Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
1/14

Organization (S): EDF-R & D/SINETICS
Handbook of Utilization
U7.0- booklet: Data exchanges
Document: U7.03.01
Macro-command MACR_ADAP_MAIL

1 Goal

To adapt a grid with software HOMARD.

This operation is placed after the first calculation on a formed grid of segments, mesh-points,
triangles, quadrangles, tetrahedrons. An indicator of the error will have been calculated. According to its value
net by mesh, software HOMARD will modify the grid. It is also possible to interpolate
fields with the nodes or constant by elements of the old grid towards the new one.

One can connect calculation and adaptation progressively process of improvement of calculation.
However, this process cannot be stopped then begun again by a “POURSUITE”. All must have
place on the way.

Software HOMARD is presented on the site:

http://www.code-aster.org/outils/homard

One finds a description of the technique used there to modify the grids as well as
examples.
To know some more about HOMARD, one can refer to the documents quoted in bibliography.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
2/14

2 Syntax

MACR_ADAP_MAIL (
# characteristic of the adaptation

ADAPTATION =
_F (

# choice of the type of adaptation

/FREE =/“RAFF_DERA”
/
“RAFFINEMENT”
/
“DERAFFINEMENT”
/UNIFORM =/“REFINEMENT”
/
“DERAFFINEMENT”
/
“RIEN”


# grid to be modified

MAILLAGE_N = man











[grid]


# new grid


MAILLAGE_NP1
=
Co
(manp1) [K8]


# If the adaptation is free


# choice of the structure containing the indicator



/
RESULTAT_N =
resun
[evol_elas]
[evol_thme]
[evol_noli]
[evol_ther]




INDICATEUR =
indic
[K16]
/
CHAM_GD =
cham_gd_i
[cham_gd]


NOM_CMP_INDICA = cmp
[K8]


Selection of the temporal parameter
/
NUME_ORDRE
=
command
[I]
/
INST
=
moment
[R]




I PRECISION
=
/
prec [R]
/
1.0E-3
[DEFAUT]





I CRITERION =/“RELATIVE”
[DEFAUT]
/
“ABSOLU”

#
If
refinement



/CRIT_RAFF_PE
= crp
[R]
/
CRIT_RAFF_REL
=
crr
[R]
/
CRIT_RAFF_ABS
=
CRA
[R]


# Finsi
#
If
déraffinement



/CRIT_DERA_PE
= cdp
[R]
/
CRIT_DERA_REL
=
cdr
[R]
/
CRIT_DERA_ABS
=
half-value layer
[R]


# Finsi

# Finsi

# If refinement, free or uniform


NIVE_MAX
=
nivmax
[I]

# Finsi

# So déraffinement, free or uniform


NIVE_MIN
=
nivmin
[I]

# Finsi






)
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
3/14

# Followed by a border

MAILLAGE_FRONTIERE
=
maf
[grid]


GROUP_MA
=
l_grma [l_gr_maille]

# Updated of fields on the new grid
MAJ_CHAM = _F (


# choice of the structure containing the field to be updated



/
RESULTAT
=
resu [evol_elas]
[evol_thme]
[evol_noli]
[evol_ther]




NOM_CHAM
=
nomsymb [K16]
/
CHAM_GD =
cham_gd [cham_gd]

Selection of the temporal parameter
/
NUME_ORDRE
=
command
[I]
/
INST
=
moment
[R]



I PRECISION
=
/
prec [R]
/
1.0E-3
[DEFAUT]




I CRITERION =/“RELATIVE”
[DEFAUT]
/
“ABSOLU”


CHAM_MAJ
=
Co
(chpmaj)
[K8]

TYPE_CHAM
=/“NOEU_TEMP_R”







/“NOEU_DEPL_R”
/
etc






)


NON_SIMPLEXE =/0






/1
/
2
[DEFAUT]

NOMBRE
=
/
“OUI”
[DEFAUT]
/
“NON”

QUALITY =/“YES”
/
“NON”
[DEFAUT]
CONNEXITE
=/“OUI”
/
“NON”
[DEFAUT]
TAILLE
=
/
“OUI”
/
“NON”
[DEFAUT]
INTERPENETRATION
=/“OUI”
/
“NON”
[DEFAUT]

LANGUE
=
/
“FRENCH” [DEFECT]
“FRENCH”
“ENGLISH”
“ENGLISH”
VERSION_HOMARD =/“V7_7”
[DEFAUT]
/
“V7_N'
/
“V7_N_PERSO”
INFORMATION =/1 [DEFECT]




/2





);
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
4/14

3
Description of an adaptation of grid

3.1
General outline of an adaptation

The principle of a calculation with adaptation of grid is as follows:

Phase 1:
Reading of the initial grid, m0
Definition of materials
Phase 2:
·
definition of the model, the loadings on this grid m0
·
mechanical calculation producing a resu0 result
·
calculation of an indicator of error, ERR0

This initial phase is the standard phase of any calculation

Phase 3:
Adaptation. One recovers a new grid, m1
Phase 4:
·
definition of the model, the loadings on the grid m1,
·
mechanical calculation producing a resu1 result,
·
calculation of an indicator of error, ERR1.

Phase 4 is the copy of phase 2. The only thing which changed is the grid. So all them
concepts while depending must be taken again. Today, there is no possibility nor of re-using them
old concepts, nor to destroy them automatically.

Then, one can continue, as many once as one wants, the tandem phase 3/phase 4. In the absence of
structure of loop in the process control language, the user must duplicate itself them
instructions.

See the reference [bib1] for a general presentation of the adaptation of grid and HOMARD,
accompanied by examples.

Caution:

This sequence of calculations and adaptations should not be stopped then begun again by one
“POURSUITE”.

3.2
Operation of the macro-command

Phase 3 carries out the adaptation of the grid. It is activated by the macro-command
MACR_ADAP_MAIL, described in this document. It has as an essential argument the name of the concept of
grid running and the name which one will give to the concept of the future grid. The other obligatory data
is the type of adaptation which one wishes: free, i.e. according to the values which one takes
indicator of error on the elements of the grid, or uniform, i.e. all the elements are
treaties in the same way.

The other data depend then on the options selected.

In complement with the adaptation, HOMARD can provide on request of the assessments on the quality of
elements of the grid, the connexity of the field of calculation, sizes characteristic or a control of
noninterpenetration of the elements. This information is obtained by the activation of the key words
associated.

In a general way, the essential impressions provided by HOMARD are inserted in the file
“mess” with the execution. In the event of error, or in mode of information 2, more detailed impressions have
place.

Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
5/14

4 Operands

4.1 Key word
ADAPTATION

ADAPTATION = _F (

This operand makes it possible to define the type of adaptation wished. The selection is done initially among
two types: free or uniform.

4.1.1 Operand
LIBRE

/LIBRE
=/“RAFF_DERA”
/
“RAFFINEMENT”
/
“DERAFFINEMENT”

This operand is employed to make free adaptation of a grid. In other words,
the decision of (die) refining an element is caught according to the value of an indicator
of error calculated before.

The choice can be done between three alternatives:

·
“RAFF_DERA”: the grid is refined and déraffiné according to the indicator
of error,
·
“RAFFINEMENT”: only the function of refinement is activated. Elements with
low level of error are not déraffinés,
·
“DERAFFINEMENT”: it is the reverse, only the function of déraffinement is activated.
The elements on high level of error are not refined.

4.1.2 Operand
UNIFORME

/UNIFORME
=/“RAFFINEMENT”
/
“DERAFFINEMENT”
/
“RIEN”

This operand is employed to make a uniform adaptation of a grid. In others
terms, all the elements of the grid are treated same manner, without holding account
of an indicator of error. The choice can be done between three alternatives:

·
“RAFFINEMENT”: all the elements are refined,
·
“DERAFFINEMENT”: all the elements are déraffinés,
·
“RIEN”: all the elements are preserved; the grid is the same one at the exit as with
the input.

4.1.3 Operand
MAILLAGE_N

MAILLAGE_N = man

Grid of the type [grid] to adapt. Attention, the adaptation can relate only to
simplexes: segments, mesh-points, triangles, quadrangles or tetrahedrons, in degree 1 or 2. If one
provides a comprising grid of other elements, two cases of figure are possible: that is to say a stop
in error, that is to say an adaptation on the zone in simplex and a restitution with identical of the remainder of
grid. The choice between these two operating modes is made by the key word
NON_SIMPLEXE.

4.1.4 Operand
MAILLAGE_NP1

MAILLAGE_NP1 = Co
(manp1)

The name of the concept of the type [grid] which will contain the grid resulting from the adaptation. This name
must respect the usual constraints of the names of concept (8 characters to the maximum) and
to be never used.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
6/14

4.1.5 Operand
RESULTAT_N

RESULTAT_N = resun

This operand makes it possible to indicate the concept of the type [result] which contains the indicator of error
to use for free adaptation.

4.1.6 Operand
INDICATEUR

INDICATEUR = indic

One specifies here which is the indicator of error which is used for the adaptation.

Caution:

The field must be present in the result; if it misses, it is not calculated of office.
The user has the choice of the indicator: maybe of the fields already defined out of standard in
Aster (cf [U4.81.02] and [U4.81.03], is a personalized field. With him to choose what is
relevant for its calculation.

4.1.7 Operand
CHAM_GD

CHAM_GD = cham_gd_i

This operand makes it possible to indicate the concept of the type [cham_gd] which contains the indicator of error with
to use for free adaptation.

4.1.8 Operand
NOM_CMP_INDICA

NOM_CMP_INDICA = cmp

Name of the component of the indicating field which must be used to control the adaptation of
grid.

4.1.9 Selection of the temporal parameter

If the structure of result contains the field of indicator of error only for one sequence number,
nothing is to be specified. In fact the values of the field to this sequence number will be used.
If not, it is necessary to specify about which number it is. That is done by the designation of a sequence number or
of a value of moment. To refer to the document [U4.71.00] for the details on these key words.

4.1.10 CRIT_RAFF_xxxx operand

In the case of free adaptation implying of the refinement of grid, it is necessary to define a high criterion of
the error. All the elements for which the indicator of error is higher than this criterion will be refined.
It is important to look at a posteriori the pace of the frequency of errors. That is possible thanks to
impressions carried out by HOMARD in the file mess. One will find there in particular a table
presenting this distribution in the form of histogram; to see chapter 5 for an example with accompanying notes.

For the choice of the criterion, three alternatives are possible:

4.1.10.1 Operand CRIT_RAFF_PE

/CRIT_RAFF_PE = crp

The criterion is defined by a proportion of elements to refine. It is a real number included/understood
between 0 and 1. The process is as follows:

·
calculation of the number of elements N corresponding to the proportion defined by crp is
N = crp X numbers total elements
·
refinement of N elements with the strongest error.

Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
7/14

4.1.10.2 Operand CRIT_RAFF_ABS


/CRIT_RAFF_ABS = CRA

The criterion is defined by an absolute value of the error. All elements with an error
higher than this value will be refined.

4.1.10.3 Operand CRIT_RAFF_REL

/CRIT_RAFF_REL = crr

The criterion is defined by a relative value of the error. It is a number ranging between 0 and 1.
The process is as follows:

·
calculation of the minimal and maximum values of the indicator of error,
·
calculation of the value corresponding to the proportion of error: v = vmin + crr (vmax ­
vmin),
·
refinement of all the elements whose error is higher than this value.

4.1.11 CRIT_DERA_xxxx operand

In the case of free adaptation implying of déraffinement, it is necessary to define a low criterion of error. All
the elements whose error is lower than this criterion will be déraffinés. Three alternatives are possible.

4.1.11.1 Operand CRIT_DERA_PE

/CRIT_DERA_PE = cdp

The criterion is defined by a proportion of elements to refine. It is a number ranging between 0
and 1. The process is as follows:

·
calculation of the number of elements N corresponding to the proportion defined by cdp is N =
cdp X numbers total elements
·
déraffinement N elements with the weakest error.

4.1.11.2 Operand CRIT_DERA_ABS

/CRIT_DERA_ABS = half-value layer

The criterion is defined by an absolute value of the error. All elements with an error
lower than this value will be déraffinés.

4.1.11.3 Operand CRIT_DERA_REL

/CRIT_DERA_REL = cdr

The criterion is defined by a relative value of the error. It is a number ranging between 0 and 1.
The process is as follows:

·
calculation of the minimal and maximum values of the indicator of error,
·
calculation of the error value V corresponding to the proportion of cdr error such as:
v = vmin + cdr (vmax ­ vmin),
·
déraffinement of all the elements whose error is lower than this value.

4.1.12 Operand NIVE_MAX

NIVE_MAX = nivmax

It is the maximum level of refinement of the grid. In other words an element of the initial grid
could not be divided more nivmax time as a whole of the process.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
8/14

4.1.13 Operand NIVE_MIN

NIVE_MIN = nivmin

It is the minimal level of déraffinement grid. I.e. only elements resulting
of at least nivmin cuttings of grid can be déraffinés.

4.2 Word
key
MAILLAGE_FRONTIERE

MAILLAGE_FRONTIERE = maf

In dimension 2, the choice of this option makes it possible the process of adjustment to follow the curvature
edges of the grid. One provides here a Code_Aster concept of the grid type which contains one
fine grid of the edges of the geometry. This grid is thus made a priori only of segments.
Their lengths are much lower than those of the segments of edge of the grid to adopt. If it
process of adjustment is brought to cut a segment of edge, the new node will be placed on
grid of the border. Thus the angles will progressively be softened adaptations.
The location of the various edges is done by the groups according to the following rule: the segments which
form an edge are gathered in a group which bears the same name in the grid of
calculation and in the grid of the border.
One will look at the case-tests ZZZZ121d and ZZZZ175a for examples of follow-up of border.

4.2.1 Operand
GROUP_MA

GROUP_MA = l_grma

If this option misses, the follow-up of the border is done for all the groups defined in
grid of the border. If one wishes to restrict this follow-up with part of the border, one gives
here the list of the groups of segments which define this part of border.

4.3 Word
key
MAJ_CHAM

MAJ_CHAM = _F (

This key word is to be employed once as many as one has fields to update old grid
towards the adapted grid.

4.3.1 Operand
RESULTAT

RESULTAT
= resu

Name of the concept [result] containing the field to be updated.

4.3.2 Operand
NOM_CHAMP

NOM_CHAMP
=
nomsymb [K16]

Reference symbol of the field which one wishes to express on the new grid.

4.3.3 Operand
CHAM_GD

CHAM_GD
=
cham_gd

Name of the concept [cham_gd] containing the field to be updated.

4.3.4 Selection of the temporal parameter

The selection of the sequence number associated with the field to be interpolated is done by the designation of a number
of command or a value of moment. To refer to the document [U4.71.00] for the details on these key words.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
9/14

4.3.5 Operand
CHAM_MAJ


CHAM_MAJ
= Co
(chpmaj)
[K8]

Name of the concept which will contain the field expressed on the new grid. This concept does not have
not to exist. It will be automatically created.

4.3.6 Operand
TYPE_CHAM

TYPE_CHAM
=/“NOEU_DEPL_R”
/
“NOEU_TEMP_R”
/
etc


One indicates the type of the concept here to be updated on the new grid. The name of this type is
built with the usual logic of Code_Aster. The first 4 characters are “NOEU”,
“ELEM”, “ELNO” or “ELGA”. One finds “_ then”. The following sequence defines the type of
field: “TEMP”, “DEPL”, etc the name ends in “_R” for a real field.

Example: “NOEU_TEMP_R”, “NOEU_DEPL_R”, etc

Caution:

There is no consistency check between the type requested and the true type of
field to be interpolated.

4.4 Operand
NOMBRE

Note:

One will consult the document [U7.03.02] describing command MACR_INFO_MAIL for
comments on the restitutions of operands QUALITE, INTERPENETRATION,
NOMBRE, CONNEXITE and TAILLE.

NOMBRE
=
/
“OUI”
[DEFAUT]
/
“NON”

If the choice is “NON”, nothing occurs.
If the choice is “OUI”, an assessment of the numbers of nodes and elements are printed on the file of
messages.

4.5 Operand
QUALITE

QUALITE
=
/“OUI”
/
“NON”
[DEFAUT]

If the choice is “NON”, nothing occurs.
If the choice is “OUI”, an assessment of the quality of the elements is printed on the file of message.
The quality of a triangle is defined as being the relationship between the length on the largest side and
the radius of the inscribed circle. The quality of a quadrangle is defined like the quotient of the product of
the biggest length and of the averages on the sides and the diagonals by smallest of
surfaces of the triangles intern with the quadrangles. In the same way, the quality of a tetrahedron is defined
as being the relationship between the length on the largest side and the radius of the registered sphere. These
reports/ratios are standardized to be worth 1 in the case of an equilateral triangle, of a square, or one
equilateral tetrahedron. For any nonequilateral element, quality is higher than 1. See
reference [bib1] for detailed explanations.
The result is presented in the form of tables, with the extreme values.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
10/14

4.6 Operand
INTERPENETRATION

INTERPENETRATION
=/“OUI”
/
“NON”
[DEFAUT]

If the choice is “NON”, nothing occurs.
If the choice is “OUI”, it is checked that the grid is correct from the point of view of the recovery:
no triangle is with horse on another triangle, no quadrangle is not with horse on another
quadrangle, no tetrahedron enters another tetrahedron.

4.7 Operand
TAILLE

TAILLE
=
/
“OUI”
/
“NON”
[DEFAUT]

If the choice is “NON”, nothing occurs.
If the choice is “OUI”, an assessment of the sizes of the under-fields is printed on the file of
messages. A under-field is defined like a whole of of the same meshs dimension and
belonging to the same groups.

4.8 Operand
CONNEXITE

CONNEXITE
=
/
“OUI”
/
“NON”
[DEFAUT]

If the choice is “NON”, nothing occurs.
If the choice is “OUI”, an assessment of the connexities is printed on the file of messages. One specifies
then if the segments, the elements 2D (joined together triangles and quadrangles) or the tetrahedrons are
of only one holding or divided into several blocks.

4.9 Operand
LANGUE

LANGUE
=
/
“FRENCH” [DEFECT]
“FRENCH”
“ENGLISH”
“ENGLISH”

This operand specifies the language in which the messages resulting from HOMARD are printed.

4.10 Operand
VERSION_HOMARD

VERSION_HOMARD
=
“V7_7”
[DEFAUT]
“V7_N'
“V7_N_PERSO”

This operand makes it possible to select the version of HOMARD which is used for the adaptation.
By defect, HOMARD 7.7 is launched. It is the version of reference. Choice “V7_N' activates
version 7.n of HOMARD which is the version of development. Active choice “V7_N_PERSO”
a version of development specific to the user. This option in fact is reserved for the team of
development of HOMARD to develop new functionalities.

4.11 Operand
INFO

INFO
=
/1




/2

If INFO is worth 2, the entirety of the output of HOMARD is included in the file of message.
If not, nothing in particular takes place.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
11/14

4.12 Operand
NON_SIMPLEXE

NON_SIMPLEXE
=/0







/1
/
2
[DEFAUT]

In its current version, HOMARD can read all the types of elements but does not make carry
the adaptation that on some: mesh-points, segments, triangles, quadrangles and tetrahedrons.
By retaining option 0, the transmission of a grid containing another thing that simplexes
a stop in error will involve.
By choosing option 1, one will be able to transmit a grid comprising any type
of element. The adaptation will relate only to the zone in simplexes or quadrangles. So consequently
from propagation of refinement, a zone in hexahedron or pentahedron came to be touched, there is
a stop in error. If not, when refinement is limited to the zone in simplexes or in
quadrangles, the other elements are restored without change.
With option 2, one authorizes refinement only on all the elements which HOMARD can treat:
mesh-points, segments, triangles, quadrangles and tetrahedrons. The presence of hexahedrons or of
pentahedrons will cause a stop in error. It is the default option.

5 Example

One will look with profit the command files associated with the ZZZZ121a case-tests, B, C, D. Ils
the processes of adjustment of grid in the form of a loop in Python language express.

Here an example of parameter setting of the macro-command.

MACR_ADAP_MAIL (
ADAPTATION =
_F
(
LIBRE
=
“RAFF_DERA”,
MAILLAGE_N
=
mun,
MAILLAGE_NP1
=
CO
(“mdeux”),
RESULTAT_N
=
remeun,
INDICATEUR
=
“ERRE_ELGA_NORE”,
NOM_CMP_INDICA
=
“ERREST”
NUME_ORDRE
=
3,
CRIT_RAFF_PE
=
0.01,
CRIT_DERA_PE
=
0.25,
NIVE_MAX
=
5










),
MAJ_CHAM
=
_F
(
RESULTAT

=
rethun,
NOM_CHAM
=
“TEMP”,
TYPE_CHAM
=
“NOEU_TEMP_R”,
INST
=
12.5,
CHAM_MAJ

=
CO
(“tempdeux”)










),
QUALITE
=
“OUI”,
INTERPENETRATION
=
“NON”





)

This sequence will adapt the grid contained in the concept mun and will restore a concept grid
of mdeux name. The adaptation is done by refinement and déraffinement free, according to the indicator of error
contents in field ERRE_ELGA_NORE of the remeun result, at the 3rd moment; the component used
is ERREST. The elements will be classified according to their level of error decreasing. The first
% will be refined; the 25% the last will be candidates with déraffinement. No element of the final grid
will not have to be resulting of more than 5 refinements.

Field TEMP of the rethun result at the moment 12,5 is expressed on the grid mun. It will be expressed
on the mdeux grid in the shape of the field of temperature to the tempdeux nodes.

A summary of the quality of the elements of the new grid is produced. One does not control
interpenetration of the elements.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
12/14

Here an example of the table presenting the distribution of the indicator of error on the grid.


**********************************************************
* Indicators of error on the grid of calculation *
* Error on the 956 triangles *
**********************************************************
* Minimum: 40.577 Maximum: 71888. *
**********************************************************

**********************************************************
* Function of distribution *
**
* Values * Nombre of elements *
* Minicomputer < < Maxi * by class * office plurality *
** 10 ** 4 * in %. numbers * in %. numbers *
**********************************************************
* 0.00 < 0.36 * 89.33. 854 * 89.33. 854 *
* 0.36 < 0.72 * 9.62. 92 * 98.95. 946 *
* 0.72 < 1.08 * 0.21. 2 * 99.16. 948 *
* 1.08 < 1.44 * 0.10. 1 * 99.27. 949 *
* 1.44 < 1.80 * 0.31. 3 * 99.58. 952 *
* 1.80 < 2.16 * 0.10. 1 * 99.69. 953 *
* 2.16 < 2.52 * 0.00. 0 * 99.69. 953 *
* 2.52 < 2.88 * 0.00. 0 * 99.69. 953 *
* 2.88 < 3.24 * 0.00. 0 * 99.69. 953 *
* 3.24 < 3.60 * 0.00. 0 * 99.69. 953 *
* 3.60 < 3.96 * 0.00. 0 * 99.69. 953 *
* 3.96 < 4.32 * 0.00. 0 * 99.69. 953 *
* 4.32 < 4.68 * 0.10. 1 * 99.79. 954 *
* 4.68 < 5.04 * 0.00. 0 * 99.79. 954 *
* 5.04 < 5.40 * 0.00. 0 * 99.79. 954 *
* 5.40 < 5.76 * 0.00. 0 * 99.79. 954 *
* 5.76 < 6.12 * 0.00. 0 * 99.79. 954 *
* 6.12 < 6.48 * 0.00. 0 * 99.79. 954 *
* 6.48 < 6.84 * 0.10. 1 * 99.90. 955 *
* 6.84 < 7.20 * 0.10. 1 * 100.00. 956 *
* 7.20 < inf. * 0.00. 0 * 100.00. 956 *
**********************************************************

100
90
80
ents
70
60
50
40
30
20
Percentage of élém
10
0
, 72
, 44
, 16
, 88
, 60, 96
, 68
, 40
, 12
, 84
< 0
< 1
< 2
< 2
< 3 < 3
< 4
< 5
< 6
< 6
00 < 0,36
36 72 < 1,08
08 44 < 1,80
80 16 < 2,52
52 88 < 3,24
24 60 96 < 4,32
32 68 < 5,04
04 40 < 5,76
76 12 < 6,48
48 84 < 7,20
0,
0,
0,
1,
1,
1,
2,
2,
2,
3,
3,
3,
4,
4,
5,
5,
5,
6,
6,
6,
Range of error

Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
13/14

The diagnosis on the distribution of the indicator of error on the grid points out initially the values
extremes met in calculation in progress. Here the minimum is 40,577 and the maximum of 71888.
Then one presents the distribution by equidistant section starting from the optimum value, 0. It is seen that
854 triangles have an error lower than 0,36 * 104, i.e. 89,33% of the total number of triangles.
92 triangles have an error ranging between 0,36 * 104 and 0,72 * 104, is 9,62% of the total number of
triangles. In cumulated, one thus notes that 946 (=854+92) triangles have an error lower than
0,72 * 104, i.e. 98,95% of the total. And so on. For example, 99,58% of the elements have an error
lower than 1,80 * 104.

On the preceding figure, one can see the representation in the form of histogram of the percentages
elements in each range of error concerned. As it could also
to note in the preceding table, one notes that very few elements concentrate a strong error.
By visualizing a representation of the cumulated percentage of elements in a range of error given,
one with the following figure.

102
100
ents
98
96
94
breadth of élém
U
92
90
88
86
84
Cum percentage
82
, 72
, 44
, 16
, 88
, 60, 96
, 68
, 40
, 12
, 84
< 0
< 1
< 2
< 2
< 3 < 3
< 4
< 5
< 6
< 6
00 < 0,36
36 72 < 1,08
08 44 < 1,80
80 16 < 2,52
52 88 < 3,24
24 60 96 < 4,32
32 68 < 5,04
04 40 < 5,76
76 12 < 6,48
48 84 < 7,20
0,
0,
0,
1,
1,
1,
2,
2,
2,
3,
3,
3,
4,
4,
5,
5,
5,
6,
6,
6,
Range of error


From this frequency of errors, one can deduce two consequences on the strategies from refinement.
If one asks for a refinement on a relative criterion of the error, key word CRIT_RAFF_REL, that returns to
to select the elements the elements which are on the right vertical line passing by it
criterion. For example if one asks CRIT_RAFF_REL = 0.85, one will select all the elements
whose L `error is higher than 0,85 * 71888, that is to say 61105. It is noted that that corresponds to very little
elements: 2 only exceed this value, i.e. 0,2% of the total.
If one asks for a refinement on a percentage of elements, key word CRIT_RAFF_PE, that returns to
to select the elements the elements which are above the horizontal line passing by
this criterion. For example if one asks CRIT_RAFF_PE = 0.85, one will select the 15%
the worst elements, is 143 elements. Among these, “the least worse” has an error lower than
3600, is 20 times smaller than the maximum.
The consequence of these remarks is that it is advisable to make a first analysis of the distribution of
the error before choosing the type and the values of the criteria of refinement. It is indeed useless, even
expensive in term of increase in the size of grid, to refine in zones where the error is not
not very strong. The adaptation will be all the more powerful as one will have known to reduce the elements to strong
error until obtaining a balance of the frequency of errors in the grid.
Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Code_Aster ®
Version
8.2
Titrate:
Macro-command MACR_ADAP_MAIL


Date:
31/01/06
Author (S):
G. Key NICOLAS
:
U7.03.01-D1 Page:
14/14

6 Bibliography

[1]
G. NICOLAS: “Software HOMARD - Volume 1 - Présentation general”, report/ratio EDF
HI-23/04/005, February 2005.
[2]
G. NICOLAS: “Software HOMARD - Volume 2 ­ Algorithmes of refinement and déraffinement
grids ", report/ratio EDF HI-23/04/006, February 2005.
[3]
G. NICOLAS: “Software HOMARD - Volume 3 ­ Interfaces with the computer codes”, report/ratio
EDF HI-23/04/007, February 2005.
[4]
G. NICOLAS: “Software HOMARD - Volume 4 ­ Structures of data”, report/ratio EDF
HI-23/04/008, February 2005.

Handbook of Utilization
U7.0- booklet: Data exchanges
HT-62/06/004/A

Outline document