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

You are not logged in.

#1 Code_Aster development » Correction of small oversite in bibfor/petsc/appcpr.F90 to allow build » 2014-12-10 22:55:12

Replies: 1

I ran across a small issue with using petsc 3.5.2 on a current aster build 12.2.20 , i believe it was introduced in "#22288 Compatibility with PETSc v3.5 ". 
When compiling with gcc 4.8  (opensuse 13.2 x86_64)  waf build fails  in appcpr.F90  when build against petsc 3.5.2 with the line

call VecSetValues( coords, nterm, to_petsc_int(igp_f - 1), val,  INSERT_VALUES, ierr )

the igp_f is referenced as a array but is in fact an int.  Changing from VecSetValues to VecSetValue and droping "nterm" seems to correct the build. ...

call VecSetValue( coords, to_petsc_int(igp_f - 1), val,  INSERT_VALUES, ierr )

Note: diff/patch is attached.

#2 Re: Code_Aster usage » ANSYS to ASTER model and boudary condition translator » 2014-08-23 12:57:03

Some time back i wrote a similar translator for Abaqus to Code  Aster which you may be able to use/tweek  directly or as a "go by" . Note Ansys will output/translate to Abaqus in their "Work Bench" although  be it abit clumsy in my opinion.  I used a Translator Construction Tool kit   called http://eli-project.sourceforge.net/ to generate the translator. My intent was to translate both the bulk data and get the loads and boundary conditions translated as well a template for the complete comm file.  Attached is the complete source spec file the format of which  being a sort of hand written "tar" file. It is a flat text file including a some what  terse instructions on how to generate the executable from this input spec. Note i made some use of in memory hashing "uthash" as a linked lists were to slow with the size of the input decks being used "acouple of million dof + ". The templating for the  Pattern-Based Text Generator PTG i mentioned is  toward the bottom of the file, you may want to adjust for your needs or taste. It is a fairly straight forward  Lexial analysis is toward the top, the Syntax and Parse tree and rules are in the middle ,  and the unparser "ptg" is toward the bottom of the file.

This was a work in progress up to about a year and half ago , I know of some problems with the generated comm file with respect to the current code aster. Originally i decided i would make a copy of the mesh  when i needed to "say add mid-face nodes" on shells etc "CREA_MAILLAGE"  but the syntax has changed since so that  part need to be fixed with eficas as well as some of the options for solver call outs i know that are dated. I also didn't format the python tabs very well in my output as i always used eficas to view/adjust the run and it reformatted it for me. One other thing of note is naming conventions aster as a short, length limited , python centric naming criteria where the Abaqus names from the ".inp" files were not  so limited. So created a hashed name  translation for longer names coming from abaqus  which doesn’t help  with readability but it was all i could come up with at the time.

#3 Salome-Meca installation » Source availibilty of EFICAS Openturns and PAL? for use salome » 2013-09-09 18:24:11

Replies: 0

I would like to request the source for the EFICAS , Openturns, and i believe PAL?  plugin source  for your  SalomeMeca distribution . I was unable to use the 2013.2 binary distribution due to obsolete FORTRAN library references . So i built the distribution from source , as i was wanting to try the hexablock mesher and tie to my MPI build version of aster and saturn and i strongly prefer to build from source. I believe they are released under GPL2 and or LGPL 2.1 licensing. Thank you for the release of the ASTER plugin in source along with the code aster source , but  it seems to want to bring up solomeEficas which is part of the EFICAS solome plugin , i think ?

Is there a Git/Svn/Mercerial ... repository that is externally visible  similar to

http://git.salome-platform.org/gitweb/  (currently broken :{ )
http://git.dev.opencascade.org/gitweb/? … ;a=summary


#4 Re: Code_Aster usage » Bolted joint w/contact MUMPS rtns matrix is not factorable » 2012-10-17 14:05:46

Thank you for your time and effort,

Yes the mail and comm files were created from a C program i wrote to convert my abaqus decks to aster.  Mainly the "bulk data" and loading so i could pull up the comm file in eficas and begin exploring asters capabilities .  I do seem to have something  behaving badly in my deck.  I was not able to get my job go run w/ the SEUIL_INIT as you mentioned but i did notice a similar responce when i tried setting the
ALGO_FROT='PENALISATION' and used COEF_PENA_CONT& COEF_PENA_FROT=1000  which is << E of 206000 MPA inoder to get it to start itterate past the the first time step . There seems to be some Large "latteral" forces being generated when the bolt is pre-tensioned via thermal cooling.  I simpified my model removing the constraint equations and putting in 6  poi1's three on the head of the bolt and 3 on the unconstrained plate, applying a K_T_D_N to it with 10000 N/mm  to stablize the model.

                  AFFE=( ...

and removed the referance to the pressure loading in STAT_NON_LINE but i am still getting <Erreur> La matrice du syst?me est singuli?re as the newton method bysects down to SUBD_PAS_MINI.

am i calling out the  time steping corectly so  the thermal preload is bisected as well ?


i was also wondering if in the CONTAIN contact method if nodes are getting projected in a manner simillar to the LAISON_MAILE ?

I am not particulary fond of my mesh on my bolt to i think i will re-mesh it to refine the contact surface and remove the transition elements PYRAM13's  and try again.

#5 Code_Aster usage » Bolted joint w/contact MUMPS rtns matrix is not factorable » 2012-10-04 20:58:41

Replies: 2

I am trying to understand how to set up a classic bolted joint problem in aster. I have 2ndorder solid model with three contact zones between the two plates under the head and nut. The bolt is loaded by ramping the temperature down to -137C  in the first half of the load step then a shear load is applied to the plates, With all 3 surface defined with

Mumps returns "Degr? libert? physique associ? au noeud : N12942 et ? la composante : LAGS_F2." when i disable time stepping. This a midside node beween the two plates and i am unsure what LAGS_F2 is, from other posts it may be related to friction. with time stepping set it bisects down to a min value and exits.  I have tried varring the value of SEUIL_INIT and added springs both between the parts and then descrite springs on each part to stabilize it , set the SOLVER NPREC. but to no avail.  I am unsure how to stabilize the simplified  model  I may not have the initial contact or friction  set up correctly . Any guildance would very appreciated.  mesh , comm and a image is in attached bolt.tar..

#6 Code_Aster usage » Convering rbe2 and rbe3 with unconnected dep node to equiv aster » 2012-07-05 15:46:09

Replies: 0

I have been working on a conversion starting with a nastran deck -> abaqus -> aster. In onder to convert the deck am  running the it through "abqus fromnastran" the taking the free formatted abaqus deck and translating it into a aster .mail and .comm file . "my way of learning aster" . My conversion/converter is attached for completeness..:)

In doing this i ran across a slight problem starting with a rbe2 in nastran it gets converted to a kinetic coupling in abaqus which i have translated to either a leison_solide or containt equations with leison_dll . The problem crops up the the original source nastran deck used a large number of rbe2's and rbe3's with floating non attached nodes as the dependant node. So aster tages them as orphans and remove there dof. I have tried to counter this by placing a "light" mass and springs on these dependant nodes so aster would leave them in.
                _F(NOEUD ='N330100',CARA='M_TR_D_N',
                VALE=(0.5236E-12,0.4,0.4,0.4,0,0,0,1,0,0), ),
                _F(NOEUD ='N330100',CARA='K_TR_D_N',
                VALE=(1.0E-3,1.0E-3,1.0E-3,1.0E-3,1.0E-3,1.0E-3), ),
        LIAISON_SOLIDE=(                _F(GROUP_NO='NG175',),
     LIAISON_RBE3=(  _F(NOEUD_MAIT="N330695",

Is the the best way to handle this, i think i am probably doing the hard way?

#7 Code_Aster installation » How may i obtain a copy of U.4.44.01 documentation non-pdf? » 2010-05-17 15:59:00

Replies: 1

I would like to obtain a copy of the "source document" which i believe to be be a openoffice document for U.4.44.01. I am most regretfully  some what linguistically challenged and have been running the pdf's through google translate but some are way to large deal with .  I have noticed there are some "docs" in the english wiki area but none in the main documentation.

Any assistance would be greatly appreciated
Thanks Mike