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

You are not logged in.

#1 Code_Aster usage » PILOTAGE - understand keyword COEF_MULT » 2022-08-03 14:03:00

Duc Vu
Replies: 0

HI all,

I'm doing some simulations with a softening material that I developed. I've got some trouble with the convergence, so I'm trying option PILOTAGE in STAT_NON_LINE.

From the test case SSNV147, the document R5.03.80 and two post below
ht  tps://code-aster.org/forum2/viewtopic.php?id=20005
ht  tps://code-aster.org/forum2/viewtopic.php?id=12930
I see that I should choose to pilot 'DEFORMATION' but I do not clearly understand how to choose COEF_MULT.
It seems like this coefficient defines how fast we move on the list of time steps and thus, the big value will give better convergence ?

Thank you

#2 Code_Aster usage » Option 'TANGENTE_SECANTE' » 2022-07-13 02:02:22

Duc Vu
Replies: 0

Hi everyone,

I'm trying to model a softening material and I've found in doc R5.03.01 (Algorithme non linéaire quasi-statique) an algorithm to across the problem of convergence, that is activate the option TANGENTE_SECANTE='OUI' in the keyword 'COMPORTEMENT'. However, I couldn't find these keywords in STAT_NON_LINE. So, how can I use the option TANGENTE_SECANTE? If this option is eliminated, what is the alternate solution?

Thank you.

#3 Code_Aster usage » Problem of visualization in Paraview » 2022-06-21 14:44:55

Duc Vu
Replies: 1

Hi all,

I've got some problems with visualizing my result in Paraview (sorry if it is duplicate).
I have a mesh with 2 groups: A and B. With Code_aster, I have a rmed file with DEPL and SIEF_ELGA.
I only want to visualize the result on group A so in .comm file I did the following:

IMPR_RESU(FORMAT='MED',
          RESU=_F(GROUP_MA=('grA', ),
                  NOM_CHAM=('DEPL', 'SIEF_ELGA'),
                  RESULTAT=RESNON,
                  TOUT_CMP='OUI'),
          UNITE=80)

But in Paraview, when I visualize the result of DEPL field, I still see group B. Meanwhile, the result of SIEF_ELGA only shows group A as expected.
How can I solve this?
Moreover, I want to check the VARI_ELGA field (internal variable at Gauss point) but Code_Aster said that it doesn't exist. Any idea on that?

Thanks

#4 Re: Code_Aster usage » Working with Orphan Nodes » 2022-05-19 10:27:32

In my experience, orphan nodes are usually occurred by the problems of mesh or boundary condition.
It will be better if you specify your stud (mesh file, command file,...)

#5 Code_Aster usage » testcase for shear behavior » 2022-05-16 18:58:39

Duc Vu
Replies: 0

Hi all,

I'm performing a simple simulation using CA15.5: a cubic with the edge of 300mm, one surface is entirely blocked (DX=DY=DZ=0) while the opposite surface is imposed an increasing displacement in X-direction, from 0 to 3mm (DY=DZ=0).
In the first step, the simulation was executed with the behavior MAZARS and I got the convergence problem when the X-displacement was higher than 0.5mm.
I've tried several solutions: refine the computation step, use RECH_LINEAIRE or tried different SOLVEUR but nothing changed.

Do you have any ideas to solve this problem? I need the simulation at the high displacement with the material that uses the small deformation hypothesis.
I attached below the .comm file and the mesh file.

Thank you in advance

#6 Re: Code_Aster usage » [SOLVED] Current stable version? » 2022-04-26 14:48:46

At the moment, the 'stable' version of CA is 15.4, version 14.6 is identified as 'old-stable' that can be found in SM2021, the Download section has not been updated for a long time.
On sourceforge, version 15.5 is 'stable-update' which means version 15 for the developer if I understand well.
You also have in sourceforge the version 16.3 which is identified as 'unstable'.

#7 Re: Code_Aster development » [Solved] Solver MUMPS and PETSc in CA15.4 » 2022-04-25 15:59:34

A little update, I forgot that I have to compile the parallel version
so I followed your tutorial here (from bind /home step) inside the containers you provided and I had a mpi_version with PETSc.
https:__//github.com/emefff/Code-Aster-MPI-in-Singularity-of-SM2021
Everything regarding the solver is good to me now. Thank you again.

#8 Re: Code_Aster development » [Solved] Solver MUMPS and PETSc in CA15.4 » 2022-04-25 13:37:21

Hello Mario,
These are what was written in my .mess file.

<INFO> Démarrage de l'exécution.

           -- CODE_ASTER -- VERSION : CORRECTIVE AVANT STABILISATION (stable-updates) --           
                               Version 15.5.3 modifiée le 04/03/2022                               
                                       révision bfb7596fc1be - branche 'v15'                       
                                   Copyright EDF R&D 1991 - 2022                                   
                                                                                                   
                                      Exécution du : Mon Apr 25 03:56:15 2022                       
                                             Nom de la machine : ductam                             
                                                Architecture : 64bit                               
                                            Type de processeur : x86_64                             
                      Système d'exploitation : Linux-5.4.0-107-generic-x86_64-with-debian-9.13     
                                          Langue des messages : fr (UTF-8)                         
                                     Version de Python : 3.6.5                                     
                                                  Version de NumPy : 1.15.1                         
                                     Parallélisme MPI : inactif                                     
                                    Parallélisme OpenMP : actif                                     
                                          Nombre de processus utilisés : 1                         
                               Version de la librairie HDF5 : 1.10.3                               
                                Version de la librairie MED : 4.1.0                                 
                               Version de la librairie MFront : 3.4.0                               
                          Version de la librairie MUMPS : 5.2.1consortium                           
                                  Librairie PETSc : non disponible                                 
                               Version de la librairie SCOTCH : 6.0.4                               

                            Mémoire limite pour l'exécution : 7648.00 Mo                           
                                          consommée par l'initialisation : 886.16 Mo               
                                       reste pour l'allocation dynamique : 6761.84 Mo               
                          Taille limite des fichiers d'échange : 46.88 Go                           

.. _stg1_txt127

#9 Code_Aster development » how to stop interpolate VARI_INTERNE » 2022-04-22 15:02:01

Duc Vu
Replies: 0

Hi all,

I'm developing a new material behavior for fiber concrete. It is a continuous model based on "smeared crack concept".
I have an internal variable that indicates the crack status of each element: opening, sliding, closing, ... and tells what kind of equation should be solved in the next step.
This variable should be constant if the crack status is unchanged. As you will see in the file logcrackH200 attached below, the variable V54 should be 1.0 while V27 remains 0.0 and change to 2.0 when V27 is not equal to 0.0.
The test with SIMU_POINT_MAT gives a good convergence, however, when I test with a mesh of 10 elements HEXA8, it seems like the variable V54 is interpolated from Gauss point, I've doubted that might be the reason for the divergence when the displacement becomes larger
I've tried to recompute this variable inside the law but it does not give better convergence.

Does anyone know how to tell CA to not interpolated an internal variable? If not, does anybody has a better solution for my problem?

Thanks.

#10 Re: Code_Aster development » [Solved] Solver MUMPS and PETSc in CA15.4 » 2022-04-22 14:40:40

Thank you mf,

I tried the container that you gave and it works well.
It contains only MUMPS and the PETSc is not included.
There are several alarms but that is okay with me.

#11 Code_Aster development » [Solved] Solver MUMPS and PETSc in CA15.4 » 2022-04-19 12:51:49

Duc Vu
Replies: 5

Hi everyone,

I'm compiling a new material in CA15.4 within the singularity shell. I've found that the MUMPS and PETSc were not configured in the version for development.
The material's law works well with solver MULT_FRONT but I'd like to test my material with these solvers (at least with MUMPS).
Does anybody have a solution to my problem?

Thanks

#12 Re: Code_Aster development » [solved] Correct folder for development in SRC » 2022-04-01 17:04:24

Merci AsterO'dactyle,

Je confirme que les bons chemins pour les routines sont:

- Catalogue des comportements et leurs variables:
./src/code_aster/Behaviours/ -> ma_loidecomportement.py & cata_vari.py
./src/code_aster/Cata/Commands -> modification du fichier defi_materiau.py
./src/code_aster/Cata/Commons -> modification des fichiers : c_relation.py

- F90 de la loi de comportement:
./src/bibfor/lc/lc.00xx.F90

- Fichier headers pour appeler fortran
./src/bibfor/include

#13 Re: Code_Aster development » waf configure failed » 2022-03-01 13:35:00

Hello,

Check your branch with hg branch. If you want to configure the CA 14.6, you should use branch v14, (default = v15)
-> hg up -> hg branch v14
If it is not the case, try ./waf configure --disable-mumps to see what happens next. If the configuration step work, you might have to compile mumps first and then build CA. I got this error several times and I'm not sure about the cause, it just worked big_smile

#14 Code_Aster development » [solved] Correct folder for development in SRC » 2022-02-22 16:12:18

Duc Vu
Replies: 2

Hello,

I would like to clarify several processes for adding a new material behaviour in CA.
I've compiled the CA15.4 development from singularity shell and I have some files that describe the material law to copy to SRC folder.
The document D5.04.01 says: "ajouter dans le repertoire bibpyt/Comportement (page 8)" and "chercher dans le répertoire bibfor/algorith un routine lc00nn" but there are neither folder name bibpyt in SRC nor the file lc00nn.F90 in bibfor/algorithm.
I think it should be replaced by folder code_aster/Behaviours and bibfor/lc?
Please confirm if there are some errors in the documentation or correct me if I'm wrong.

Thank you in advance

#15 Re: Code_Aster development » Problem when compiling new material law: hid_t unknown » 2022-02-18 14:57:55

Thank you,
I've known about the container with singularity recently and it's quite new to me.
I've installed the SM2021 and compiled the development version in shell mode successfully.
However I've confused with how to use the Singularity shell mode to make some development on the new version. It will be great if someone could explain how to develop on the new version step-by-step (In another post, perhaps!)
Moreover, the new version seems work slower and I have some problem with graphical modules.  So, I raised again the question for the one who used to (or prefer to) work with the older version like me sometime.

#16 Re: Code_Aster development » Problem when compiling new material law: hid_t unknown » 2022-02-16 19:19:08

Hello,

I would like to renew this question. I've got this error when compiling CA 14.6 from source.
I've thought that the problem is hdf5 library is missing or nor working properly, but I don't know how to fix it.
Any ideas?

#17 Re: Code_Aster usage » Calculation cost - problem with joint element in 3D » 2020-11-06 10:50:50

I got the simulation work, it seems like the coefficient Hpen is too small, I wonder if the relation JOINT_BA is applicable only for the rebar and not for the wire steel. Otherwise, I've tried several load path and I found that the convergence of the model is sensitive with shear : The model converged easier if the shear component was small and was not converged if there existed a sudden change in the load path.
Do you have any idea in this problem?

#18 Re: Code_Aster usage » Find the node in singularity error » 2020-11-06 10:30:57

Hello,

This error may be caused by a lot of reason, it will be more helpful if you post at least your mess file (.comm and mesh file are more recommended).
The line you asked belong to the tangent matrix not a specific node

#19 Code_Aster usage » Calculation cost - problem with joint element in 3D » 2020-10-23 17:58:06

Duc Vu
Replies: 1

Hi all,

I'm modelling the pullout process of a steel wire form concrete under the shear load (tangential displacement).
I degenerated the joint element from the meshes and then assign the JOINT_BA as the material to those elements. The steel is in elastoplastic and the concrete is in elastic, all the mesh was linear element.
In 2D, code_aster performed very well and gave the result in less than 10 min, but when I turned to 3D, the model that I'm interested in, the computation time was too(...)oo long. Even if I changed the steel's behavior back to elastic, it still took more than 1hour to finish.
I also refined the step of computation, increase the RESI_GLOB_MAXI but it was not better.
I tried another kind of joint element as CZM_XXX_MIX or JOINT_MECA_RUPT but the calculation was even slower.

Do you have any idea to optimized my model?, I attached below my .mess file. Thanks

Je suis en train de simuler l'extraction d'un fibre métallique sous le déplacement tangentiel d'une extrémité.
J'ai dégénéré l'élément de joint lors du maillage et l'ai affecté par le comportement JOINT_BA. La fibre est en élastoplastique et le béton est en élastique, le maillage ne constitue que des éléments linéaires.

En 2D, le calcul a bien fonctionné et sorti le résultat très vite. Par contre, en 3D, le coût de calcul soit trop élevé. J'ai augmenté le résidu global, raffiné le pas de temps, tenté avec des autres types de comportement pour les éléments joints mais la situation n'ai pas été améliorée.

Pourriez-vous partager vos expériences pour traiter ce problème? Merci d'avance.

#20 Re: Code_Aster usage » [SOLVED]3D mesh of zero thickness? » 2020-10-14 08:52:57

Hi,
I got it, that's right, salome defined the element as 3D and the code_aster run. Thank you.
Now I got another question, I assigned several type of material behaviour as : CZM_XXX_REG, CZM_LAB_MIX, JOINT_BA, JOINT_MECA_RUPT but I've got the result quite strange.
What I want is to evaluate the pullout process of steel wire under combined load (tension and shear) :
- With the cohesive element, I've got the steel wire went through the matrix of concrete
- With the JOINT_BA, the stress in wire was too high (about 10GPa)
- With JOIN_MECA_RUPT, the calculation was not converged.
Can you give me some advice, what kind of element I should use in my model?

#21 Re: Code_Aster usage » [SOLVED]3D mesh of zero thickness? » 2020-10-12 21:23:15

Thanks for you reply, I've already done that but I just got a 2D element while code_aster requires a 3d element for 3D_JOINT.
So can you give me more detail to do this stuff. For now, I'm using a very thin element to introduce the joint but the result I've got was very strange.

#22 Code_Aster usage » [SOLVED]3D mesh of zero thickness? » 2020-10-12 17:54:06

Duc Vu
Replies: 4

Hi all,

I'm trying to model the adhesion between the concrete matrix and the steel wire by using the joint element (AFFE_MODEL='3D_JOINT'). So I want to degenerate a 3D mesh with zero thickness from 2D mesh (interface fiber/concrete) as which is introduced in the test case: SSNA112 but in 3D instead of 2D.
Can anyone show me how to assign or build something like that in GMSH or Salome, please. I've tried to group the two 2D face (interface of fiber and the one of concrete) and used : CREA_MAILLAGE -> CREA_FISS but it gave me just another 2D face.

Thanks

#23 Re: Code_Aster usage » Saving table on a txt? » 2020-09-08 10:46:05

I usually print the result by format .resu and change the extension to .txt without any error, but i only do that when i need a post-treatment complexe (i.e analyze the result in MATlab, python,...).
The name of the output is always assign in the step with ASTK as far as I know.

#24 Re: Code_Aster usage » [solved]linear tranlation in contact » 2020-08-25 16:26:56

Hi nirmaljoshi, mecour and JP,

I'm performing the same problem with yours except the steel rod is pulled out by a horizontal force (DY and DX instead of DZ as describe in your mesh). So If you solved this problem with contact, could you provide your .comm file, if not, did you tried to play with cohesive element as the suggest of @mecour? As I never used cohesive element, can you give me some indication?

#25 Re: Code_Aster usage » contact "rasant" » 2020-08-23 06:09:14

Hi,
I take a look to your result file and it seems like I had the same question with yours.
I got the result with the same alert but the stress value was very strange (too small = "rasant" as you said) and my object which must slide on the other one passed through instead.
Do you have the answer for this and did we misunderstood something?
Thanks.