Welcome to the forums. Please post in English or French.

You are not logged in.

#1 2015-12-02 14:33:21

Vincent Magnenet
Member
From: IMFS - Illkirch
Registered: 2008-05-27
Posts: 64
Website

Bug in eigenvalues printed by DEFI_MATERIAU

Hello everyone,

If an orthotropic material is created, DEFI_MATERIAU prints the eigenvalues
of the stiffness matrix if it is not positive definite. These eigenvalues seem incorrect
as we can see on the following example:

``````matcalc1 = DEFI_MATERIAU(ELAS_ORTH=_F(G_LT=1.E9,
E_T=6.058199116453629E8,
E_N=5.E8,
G_TN=1.E9,
NU_LN=0.3,
RHO=2450.0,
NU_TN=0.3,
NU_LT=1.4751131221719456,
G_LN=1.E9,
E_L=2.600000000E9,
ALPHA_N=0.0,
XC=1.0,
YC=1.0,
YT=1.0,
ALPHA_T=0.0,
S_LT=1.0,
ALPHA_L=0.0,
XT=1.0,),
INFO=1,)``````

Code_Aster told me:

``````       LA MATRICE DE HOOKE A UNE VALEUR PROPRE NEGATIVE
POUR LA RELATION DE COMPORTEMENT : ELAS_ORTH

VALEURS PROPRES DE LA MATRICE DE HOOKE :

************************
!  1  !  -0.81062E+10  !
!  2  !   0.40605E+09  !
!  3  !   0.58966E+09  !
!  4  !   0.10000E+10  !
!  5  !   0.10000E+10  !
!  6  !   0.10000E+10  !
************************``````

However  the correct values from  a maxima program (in GPa)  are
8.200296      0.652950     0.41984      1.0     1.0      1.0

In the source file aniver.F90, I changed lines 305 ---> 322 to be in
coherence with the doc R4.01.02 (there are a lot of index permutations):

``````                        nu21 = e2*nu12/e1
nu31 = e3*nu13/e1
nu32 = e3*nu23/e2

delta = un-nu23*nu32-nu31*nu13-nu21*nu12 -deux*nu23*nu31*nu12

dorth(1,1) = (un - nu23*nu32)*e1/delta
dorth(1,2) = (nu21 + nu31*nu23)*e1/delta
dorth(1,3) = (nu31 + nu21*nu32)*e1/delta
dorth(2,2) = (un - nu31*nu13)*e2/delta
dorth(2,3) = (nu32 + nu31*nu12)*e2/delta
dorth(3,3) = (un - nu21*nu12)*e3/delta

dorth(2,1) = dorth(1,2)
dorth(3,1) = dorth(1,3)
dorth(3,2) = dorth(2,3)

dorth(4,4) = g12
dorth(5,5) = g13
dorth(6,6) = g23``````

and forced the printing of the eigenvalues in dortvp.F90. The eigenvalues
are now correct:

``````       VALEURS PROPRES DE LA MATRICE DE HOOKE :

************************
!  1  !   0.41985E+09  !
!  2  !   0.65295E+09  !
!  3  !   0.10000E+10  !
!  4  !   0.10000E+10  !
!  5  !   0.10000E+10  !
!  6  !   0.82003E+10  !
************************``````

http://web-code-aster.org/forum2/viewtopic.php?id=11970

but to my mind the source file (Code_Aster 12.2) was still incorrect.

A+

"Avec suffisamment de paires d'yeux, tous les bugs feront surface."

Linus Torvalds.

Offline

#2 2015-12-05 12:09:31

mathieu.courtois
From: France
Registered: 2007-11-21
Posts: 1,170

Re: Bug in eigenvalues printed by DEFI_MATERIAU

Hello,

Please report this "bug" (to be confirmed by the person in charge of the anisotropic elasticity) on https://bitbucket.org/code_aster/codeas … tatus=open.

The results are the same in the last version.
Join your patch and the testcase for analysis.

MC

Attachments:

Code_Aster release : last unstable on Ubuntu 16.04 64 bits - GNU Compilers

Please do not forget to tag your first post as *SOLVED* when it is!

Offline

#3 2018-01-21 22:48:43

tianyikillua
Member
From: Paris
Registered: 2017-11-06
Posts: 69

Re: Bug in eigenvalues printed by DEFI_MATERIAU

It seems to me that this bug is still not corrected.

I reported in at https (://) bitbucket (.) org (/) code_aster (/) codeaster-src (/) issues (/) 118

Sorry I'm not allowed to post URL's...

Offline

#4 2021-01-14 10:51:50

Renaut
Member
Registered: 2016-01-04
Posts: 4

Re: Bug in eigenvalues printed by DEFI_MATERIAU

Hi
it seems this bug is not solved in Code_Aster 13.6.0.

It worries me because it happens only with certains angles of materials.

For angles of 55° (it is a filament wound structure)
MA_1 = DEFI_MATERIAU(ELAS_ORTH=_F(G_LT=3.2208996369589444E4,
E_T=2.858130551981914E4,
ALPHA_N=3.5000000000000004E-05,
ALPHA_L=1.0036000000000002E-05,
E_N=9521.535788937483,
ALPHA_T=-4.0612999999999996E-06,
G_TN=3642.278790279185,
NU_LN=0.22185548965929472,
NU_TN=0.013564569100791339,
NU_LT=0.44663365611352024,
G_LN=3339.7203222299786,
E_L=9806.150075423458,
XC=1.0,
YC=1.0,
S_LT=1.0,
YT=1.0,
XT=1.0,),
INFO=1,)

there is no warning and buckling results look good.

but for angle of 30°
MA_2 = DEFI_MATERIAU(ELAS_ORTH=_F(G_LT=2.7949426830121825E4,
E_T=9022.813702069063,
ALPHA_N=3.5000000000000004E-05,
ALPHA_L=1.0036000000000002E-05,
E_N=9260.2940709928,
ALPHA_T=-4.0612999999999996E-06,
G_TN=3269.844111207362,
NU_LN=-0.22139505617036906,
NU_TN=2.6069997911863334,
NU_LT=1.5142569828273407,
G_LN=3712.1550013018004,
E_L=4.316872379748279E4,
XC=1.0,
YC=1.0,
S_LT=1.0,
YT=1.0,
XT=1.0,),
INFO=1,)

LA MATRICE DE HOOKE A UNE VALEUR PROPRE NEGATIVE
POUR LA RELATION DE COMPORTEMENT : ELAS_ORTH

VALEURS PROPRES DE LA MATRICE DE HOOKE :

************************
!  1  !  -0.55030E+04  !
!  2  !   0.24969E+04  !
!  3  !   0.32698E+04  !
!  4  !   0.37122E+04  !
!  5  !   0.27949E+05  !
!  6  !   0.43103E+05  !
************************
and buckling results are not coherent.

besides bitbucket bug is no longer available.

This bug is not corrected in Code_Aster 14.

How can I fix this ?

Attachments:

Offline

#5 2021-01-14 14:48:57

Renaut
Member
Registered: 2016-01-04
Posts: 4

Re: Bug in eigenvalues printed by DEFI_MATERIAU

I checked aniver.F790 in Code Aster 14.6.0 and it is consistent with R4.01.02.
The bug has been corrected.

but I don't understand why I have negative values for Hook matrix.

Last edited by Renaut (2021-01-14 14:49:23)

Offline

#6 2021-01-14 18:46:06

nicolas_p
Member
Registered: 2020-04-14
Posts: 11

Re: Bug in eigenvalues printed by DEFI_MATERIAU

I have no warning in version 15.3.2

Offline

#7 2021-01-19 14:19:21

Renaut
Member
Registered: 2016-01-04
Posts: 4

Re: Bug in eigenvalues printed by DEFI_MATERIAU

ok I will install Code_Aster 15;
Thank you

Offline