background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
1/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
Organization (S):
EDF-R & D/AMA















Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
Document: V6.04.173



SSNV173 ­ Barreau fissured with X-FEM




Summary

The purpose of this test is to validate two aspects of elementary calculation within the framework of X-FEM [R7.02.12]:
· the integration of a discontinuous size thanks to a under-cutting of the element,
· the enrichment of the functions of form by the Heaviside function.

This test brings into play a parallelepipedic bar fissured on all its section, subjected to a displacement
imposed, which has as a consequence the total opening of the fissure and the separation of the two parts of
structure.

The influence of the mesh and the boundary conditions is also studied.
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
2/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
1
Problem of reference
1.1 Geometry
The structure is a right with low square and healthy parallelepiped. Dimensions of the bar
(see [Figure 1.1-a]) are: LX = 5, LY = 5 and LZ = 25. It does not comprise any fissure.
The fissure will be introduced by functions of levels (level sets) directly into the file
order using operator DEFI_FISS_XFEM [U4.82.08]. The fissure is present in the middle of
the structure by the means of its representation by 2 level sets LSN and LST (see [Figure 1.1-b]) of which them
equations are as follows:
2
/
:
fissure)
of
plan
(for
LSN
LZ
Z
-
éq
1.1-1
10
:
fissure)
of
melts
(for
LST
-
X
éq
1.1-2
Thus, the bottom of fissure is located to Z = LZ/2 and at X = 10. It is thus located apart from the structure, it
who allows to have the structure completely cut into two.
Geometry of the bar and positioning of the fissure
1.2
Properties of material
Young modulus: E= 205000 MPa
Poisson's ratio:
= 0.3
1.3
Boundary conditions and loadings
The nodes of the lower face of the bar are embedded and a displacement is imposed on those of
the higher face. One wishes to show here the possibility of separating a finite element into two with
X-FEM.
Appear 1.1-a
Appear 1.1-b
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
3/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
2
Modeling a: only one finite element
2.1
Characteristics of the mesh
The structure is modelized by only one finite element of type HEXA8. The fissure is thus present at
center of this element by the means of the level sets.
2.2
Boundary conditions
Considering that the nodes close to the fissure, i.e. here the 8 nodes of the mesh are enriched by
ddls additional, the boundary conditions are written a little differently. That relates to
the enrichment of the conventional functions of forms [R7.02.12] by the function Heaviside H (X).
To impose a null displacement on the nodes of the lower face amounts writing a linear relation
between the ddls. For each node, one imposes DCX ­ H1X = 0 (idem according to y and Z) where DCX is the ddl
conventional and H1X the enriched ddl.
For the nodes of the higher face, one imposes a displacement according to Z being worth 10
- 6
and no one following
two other directions, i.e. DCX + H1X = 0, DCY + H1Y = 0 and DCZ + H1Z = 10
- 6
.
2.3 Resolution
analytical
The solution of such a problem is of course obvious. It is seen well that mechanically speaking, both
left the structure will be detached: the lower part will have a null displacement and the part
higher an overall movement equal to imposed displacement will have (see [Figure 2.3-b]).
In our case, it does not occur anything according to X and y, and according to symmetries of the problem, one can
to study only on one vertical edge (for example that joining the noted nodes N1 and N2).
There are four conditions to satisfy:
· displacement of null the N1 node
· displacement of the point medium lower A
-
no one
· displacement of the node N2 equal to imposed displacement, noted uz
· displacement of the point medium higher A
+
equal to uz
According to the formulation, these four conditions are written respectively:
(
)
(
)




=
+
+
+
=
+
=
-
+
-
=
-
uz
Z
H
dcz
Z
H
dcz
uz
Z
H
dcz
Z
H
dcz
Z
H
dcz
Z
H
dcz
2
2
1
1
2
2
2
2
1
1
1
1
1
1
2
1
1
0
1
1
2
1
0
1








=














-
-
-
=








-
2
2
2
2
0
0
0
0
1
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
1
1
0
0
1
1
1
2
2
1
1
uz
uz
uz
uz
uz
uz
Z
H
dcz
Z
H
dcz
The analytical solution is then the following one: all the ddls according to X and are null there and all the ddls
according to Z uz/2 is worth.
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
4/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
States initial and final of the structure

2.4 Functionalities
tested
Controls
DEFI_FISS_XFEM
MODI_MODELE_XFEM

Appear 2.3-a
Appear 2.3-b
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
5/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
3
Results of modeling A
3.1 Values
tested
One tests the values of displacement after convergence of the iterations of the operator
STAT_NON_LINE. It is checked that one finds the values well determined with [§2.3].
Identification Reference
Aster %
difference
DCX for all the nodes
0.00
0.00
0.00
DCY for all the nodes
0.00
0.00
0.00
DCZ for all the nodes
0.5E-6
0.5E-6
0.00
H1X for all the nodes
0.00
0.00
0.00
H1Y for all the nodes
0.00
0.00
0.00
H1Z for all the nodes
0.5E-6
0.5E-6
0.00
To test all the nodes in only once, one test the MIN and the MAX of the column.

3.2 Comments
It is noticed that the values of the ddls Heaviside in Z are not null because there is discontinuity of
field of displacement following this direction to the interface.
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
6/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
4
Modeling b: Several finite elements
4.1
Characteristics of the mesh
One discretizes the structure in 5 finite elements HEXA8
The nodes on both sides of the fissure are nodes nouveau riches, therefore them
three central meshs having such nodes are they also enriched.
Only the two extreme meshs are conventional meshs having only
conventional nodes.
One will be able to thus impose boundary conditions on the extreme meshs
in the usual manner.

4.2
Boundary conditions
The boundary conditions applied represents the same phenomenon
physics that for modeling A. One embeds the nodes of the face
lower and one imposes a displacement of the nodes of the face
higher:
Lower face (Nodes N1, N6, N11, N16): DX = 0, DY = 0 and DZ = 0
Higher face (Nodes N21, N22, N23, N24):DX = 0, DY = 0 and DZ = uz.
This constitutes the 1st case of loading.
In fact, one takes freedom to move the higher part of the structure according to the three directions,
one will thus choose like 2
ème
case of loading:
Lower face (Nodes N1, N6, N11, N16): DX = 0, DY = 0 and DZ = 0
Higher face: DX = ux, DY = uy and DZ = uz
ux = 10
- 6
uy = 2. 10
- 6
uz = 3. 10
- 6
4.3 Resolution
analytical
The solution of such a problem is of course still obvious. In a way similar to [§2.3], one can
to show that the solution is as follows:
Nodes of the lower face (stage n°0):
DX = 0, DY = 0 and DZ = 0.
Nodes of the stage n°1:
DX = 0, DY = 0 and DZ = 0.
Nodes of the stages n°2 and 3:
DCX = ux/2, DCY = uy/2, DCZ = uz/2,
H1X = ux/2, H1Y = uy/2 and H1Z = uz/2.
Nodes of the stage n°4:
DX = ux, DY = uy and DZ = uz.
Nodes of the higher face (stage n°5):
DX = ux, DY = uy and DZ = uz.

4.4 Functionalities
tested
Controls
DEFI_FISS_XFEM
MODI_MODELE_XFEM
AFFE_CHAR_MECA LIAISON_XFEM
Appear 4.1-a: Mesh
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
7/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
5
Results of modeling B
5.1 Values
tested
One tests the values of displacement after convergence of the iterations of the operator
STAT_NON_LINE. It is checked that one finds the values well determined with [§4.3] for the 2 cases of
loadings. One obtains the following table for the 2
ème
case of loading.
Identification Reference
Aster %
difference
DX in N1 (stage n°0)
0.00
0.00
0.00
DY in N1 (stage n°0)
0.00
0.00
0.00
DZ in N1 (stage n°0)
0.00
0.00
0.00
N2 DX (stage n°1)
0.00
0.00
0.00
N2 DY (stage n°1)
0.00
0.00
0.00
N2 DZ (stage n°1)
0.00
0.00
0.00
DCX in N3 (stage n°2)
0.5E-6
0.5E-6
0.00
DCY in N3 (stage n°2)
1.0E-6
1.0E-6
0.00
DCZ in N3 (stage n°2)
1.5E-6
1.5E-6
0.00
H1X in N3 (stage n°2)
0.5E-6
0.5E-6
0.00
H1Y in N3 (stage n°2)
1.0E-6
1.0E-6
0.00
H1Z in N3 (stage n°2)
1.5E-6
1.5E-6
0.00
DCX in N4 (stage n°3)
0.5E-6
0.5E-6
0.00
DCY in N4 (stage n°3)
1.0E-6
1.0E-6
0.00
DCZ in N4 (stage n°3)
1.5E-6
1.5E-6
0.00
H1X in N4 (stage n°3)
0.5E-6
0.5E-6
0.00
H1Y in N4 (stage n°3)
1.0E-6
1.0E-6
0.00
H1Z in N4 (stage n°3)
1.5E-6
1.5E-6
0.00
DX in N5 (stage n°4)
1.0E-6
1.0E-6
0.00
DY in N5 (stage n°4)
2.0E-6
2.0E-6
0.00
DZ in N5 (stage n°4)
3.0E-6
3.0E-6
0.00
DX in N21 (stage n°5)
1.0E-6
1.0E-6
0.00
DY in N21 (stage n°5)
2.0E-6
2.0E-6
0.00
DZ in N21 (stage n°5)
3.0E-6
3.0E-6
0.00
One does not test with each time only one node of each stage.
5.2 Comments
It is noticed that the values of the ddls Heaviside in X, y and Z are not null because there is discontinuity
field of displacement following these three directions to the interface.

background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
8/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
6
Modeling C: Several elements and interfaces leaning
6.1
Characteristics of the mesh and the interface
One considers a structure of dimensions LX = 5 m, LY = 5 m and LZ = 25 Mr. This structure are
discretized with 5 finite elements HEXA8. One is interested in a plane interface of normal
1
N
1
1
-
=
passing by point A of co-ordinates (5, 5
, 5). [Figure 6.1-a] a zoom of the 2 shows
ème
element
where the trace of the interface is represented in red.
Appear 6.1-a: Mesh C and zoom
The interface is characterized by the level set normal having for Cartesian equation:
5
-
+
+
-
=
Z
y
X
lsn
Note:
The parameter
a strong influence on the problem has. If 0 or 1 are worth, then point A coincides
with a node, and the interface passes by this node. If
is nonnull, but small in front of 1, the interface
will separate the element into 2 part, very different volumes. In this situation, enrichment
N9 node by the Heaviside function becomes almost useless, and led to very small pivots
during the factorization of the matrix of rigidity. That results in a significant loss of
a number of decimals and with a false result. For
= a 0.1 (either point A accounting for 10% of
the edge), one loses already 8 decimals (default value causing a fatal error) and for
=
0.05, one lose 10 decimals. The idea thus consists in not enriching the N9 node by the function
Heaviside when such cases arise. An algorithm of detection was set up, based
on the volumetric ratio for an element crossed into two. This problem makes it possible to test the good
operation of the algorithm, when the parameter
becomes small.
In the continuation, one will take
= 0.02. This value leads to the loss of 13 decimals at the time of
factorization (before the installation of a special processing).

The algorithm will be validated if calculation proceeds normally, without loss of decimals in
factorization. One will check also solution displacement in a selected node.
N9
background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
9/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
6.2
Boundary conditions
The boundary conditions are the same one as those of modeling B. One embeds the nodes of
the lower face and one impose a displacement of traction on the nodes of the higher face:
Lower face (Nodes N21, N22, N23, N24): DX = 0, DY = 0 and DZ = 0
Higher face (Nodes N13, N14, N15, N16): DX = 0, DY = 0 and DZ = 10
- 6
.

6.3 Functionalities
tested
The determination of the nodes where Heaviside enrichment is superabundant is carried out in
order MODI_MODELE_XFEM [U4.44.11] and the cancellation of these ddls is carried out by the control
AFFE_CHAR_MECA [U4.44.01] with key word LIAISON_XFEM=' OUI'.

Controls
MODI_MODELE_XFEM
AFFE_CHAR_MECA LIAISON_XFEM

background image
Code_Aster
®
Version
8.2
Titrate:
SSNV173 ­ Barreau fissured with X-FEM
Date:
25/11/05
Author (S):
S. GENIAUT, P. MASSIN
Key
:
V6.04.173-B
Page:
10/10
Manual of Validation
V6.04 booklet: Linear statics of the voluminal structures
HT-66/05/005/A
7
Results of modeling C
7.1 Values
tested
The good course of calculation makes it possible a priori to validate the case. Moreover, one tests the values of
displacement with the N1 node of co-ordinates (0, 5, 0) as well as contact pressures (LAGS_C). One
make sure that the displacement of the N1 node is correct and that contact pressures all are
null (loading of traction).
Identification Reference
Aster %
difference
MIN (LAGS_C) 0.00
0.00
0.00
MAX (LAGS_C) 0.00
0.00
0.00
DCZ in N1
5.E-7
5.E-7
0.00
H1Z in N1
5.E-7
5.E-7
0.00

Recall:
DCZ is the component according to Z of the conventional ddl
H1Z is the component according to Z of the enriched ddl


8
Summaries of the results
The objectives of this test are achieved:
·
It is a question of validating the taking into account of enrichment by the Heaviside function of
conventional functions of form.
·
Moreover, modeling B allowed the validation of the operand LIAISON_XFEM which allows
to remove the ddls nouveau riches in excess.