Frame mesh problem and AFFE_CARA_ELEM problem.

MiningMinistry2017

Hello! Tried to implement the frame example from book "beginning with Code_Aster". But I get error "fichier MED : fort.20 maillage ! soit le fichier n'existe pas, soit c'est une mauvaise version de HDF (utilise par MED). !"

I suppose that this problem is caused by this:

Only one fourth of the frame is covered by mesh! What did I do wrong in meshing?

And another problem is that in POUTRE I can only define in CARA "HZ","HY", because presence of "EP" causes error:

elemcar=AFFE_CARA_ELEM(MODELE=model,


                       POUTRE=(_F(SECTION='RECTANGLE',
                                  GROUP_MA='mast',
                                  CARA=('HY','HZ',),

VALE=(40.0,20.0,),),

There was CARA=('HY','HZ','EP',) VALE=(40.0,20.0,1.5,) in the book, but this usage causes error!

Best regards,

Danil.

MiningMinistry2017

Now I started new geometry and there is no any problem with the mesh. But I don't know how to describe what I get:

NOMBRE DE GROUPES DE NOEUDS 4


                              loadN                                1
                              groundS                              1
                              groundN                              1

loadS 1

And then

! le GROUP_MA loadN ne fait pas partie du maillage : mesh !

What?? It reports that there is loadN as group of the nodes and it exists! But later it says that there is no loadN in the mesh!

This is what I have in comm file:

model=AFFE_MODELE(MAILLAGE=mesh,


                  AFFE=(_F(GROUP_MA=('horbar','mast',),
                           PHENOMENE='MECANIQUE',
                           MODELISATION='POU_D_T',),
                        _F(GROUP_MA='loadN',
                           PHENOMENE='MECANIQUE',

MODELISATION='DIS_T',),),);

How to solve this problem? I don't know what to do. I can't imagine how the group can be present and absent in the mesh at the same time.

PierreG

Hi Danil,

"loadN" is a group of nodes. What you try to affect in AFFE_MODELE is a group of mesh which called "loadN". That's why you have the message:

! le GROUP_MA loadN ne fait pas partie du maillage : mesh !

Two solutions:

- you have to create POI1 (1 node mesh) with the group of nodes "loadN",

- you can assign the model 'DIS_T' on the group of nodes 'loadN' (but it depends of the calculation you have to do next. If you use STAT_NON_LINE or DYNA_NON_LINE, I think you have to create POI1)

Regards

Pierre

MiningMinistry2017

Hi, Pierre,

Thank you! Now there is no problem with groups, only alarm appears:

le GROUP_NO :  massN  existe dÃ©jÃ , on ne le crÃ©e donc pas.    !

I'm still trying to implement the beam example from the book "beginning with Code_Aster", I even copied text for comm file from the book but it doesn't work.

Firstly, there was an error: Le calcul de l'option RIGI_MECA n'est pas possible. Il manque le CARA_ELEM.

I've read a topic about it and found a solution, I added "CARA_ELEM=elemcar" in my comm file:

stat=MECA_STATIQUE(MODELE=model,
	           CARA_ELEM=elemcar,
                   CHAM_MATER=material,
                   EXCIT=(_F(CHARGE=ground,),
                          _F(CHARGE=selfwght,
                             FONC_MULT=selfw_m,),
                          _F(CHARGE=cc,
                             FONC_MULT=cc_m,
                             TYPE_CHARGE='FIXE',),
                          _F(CHARGE=cr,
                             FONC_MULT=cr_m,
                             TYPE_CHARGE='FIXE',),),);

But now there is another error: matrix is singular, check the boundary conditions.

ProblÃ¨me : la matrice est singuliÃ¨re ou presque singuliÃ¨re : ! ! Lors de la factorisation de la matrice, on a rencontrÃ© un problÃ¨me ! ! (pivot nul ou presque nul) Ã la ligne 37 qui correspond au degrÃ© de libertÃ© donnÃ© ci-dessus. ! ! ! ! Risques et conseils : ! ! * Si la ligne correspond a un degrÃ© de libertÃ© physique, il s'agit probablement d'un mouvement ! ! de corps rigide mal bloquÃ©. ! ! VÃ©rifiez les conditions aux limites. ! ! Si vous faites du contact, il ne faut pas que la structure ne "tienne" que par le contact. ! ! VÃ©rifiez Ã©galement les caractÃ©ristiques matÃ©riaux (module d'Young, ...). ! ! ! ! * Si la ligne correspond a un degrÃ© de libertÃ© de Lagrange, il s'agit sans doute d'une condition ! ! limite redondante. ! ! En particulier, il se peut que la relation linÃ©aire surabondante provienne des conditions de contact. ! ! Peut-Ãªtre devriez vous exclure certains noeuds des conditions de contact ! ! (mots clÃ©s SANS_NOEUD et SANS_GROUP_NO). ! ! ! ! * Si le solveur utilisÃ© est LDLT ou MULT_FRONT, vous pouvez utiliser le solveur MUMPS ! ! car celui-ci est le seul Ã pouvoir factoriser les matrices qui ne sont pas dÃ©finies positives. ! ! ! ! * Il se peut aussi que ce phÃ©nomÃ¨ne soit tout Ã fait normal avec X-FEM si la fissure passe !

! trÃ¨s prÃ¨s d'un noeud. !

Attached my comm file, maybe it helps.(there is "DX=0.0" commented(#) in comm file, but this is mistake, it is not commented)

Danil.

Attached file:
240817_frame.comm, 8172kb

MiningMinistry2017

DegrÃ© de libertÃ© physique associÃ© au noeud N7 et Ã la composante DX.

and after that goes the message I wrote before.

So there is a problem with node n7, but if so, I have no idea what to do. What can i do in this case?

Danil.

PierreG

It could be difficult to solve an instability problem (singular matrix). In your case I think it is a problem of boundary conditions (mechanical problem, not aster problem).

I read the "comm" file and I saw at line 58 that "DX=0.0" is commented. I think if you uncommented this line (deleted the # character), the calculation will work.

Regards

Pierre

MiningMinistry2017

Uncommented, but it doesn't work anyway... I've set boundary conditions according to the book, I'm trying to implement a beam example.

Thanks for viewing my comm-file!

The task is pretty simple, I just want to learn how to create beam constructions this way and to learn how to get reactions. But I face this problem with singularity. Checked my boundary conditions and they're fine. You checked it too and didn't notice anything but "#dx".

Danil.

jeanpierreaubry

hello

you should post here, at least

the .mess file

the mesh file, .med and the geometry file

to find out where the error is lying

for example without that one does not know where is "noeud N7"

jean pierre aubry

MiningMinistry2017

Good afternoon!

There are .med .comm .resu .mess .hdf files in this archive.

If you have time, please, check this out, maybe you can see here something I can't.

Regards,

Danil.

Attached file:
n240817.tar.gz, 49953kb

jeanpierreaubry

as it can be seen from the attached screen capture

applying

Modification > Transformation > Merge nodes

will show a few double, even triple nodes

you should apply the above command to remove them

Salome mesher never removes double nodes while you have to be very clever to create them with Gmsh

Attached file:
merge_nodes.png, 195670kb

PierreG

the problem is that node N7 is an orphan node (see .mess file) and you affect a DIS_T on a mesh support by this node (POI1 mesh group 'massN') .

The problem is instable because there are no connections and boundary conditions on this node.

Regards

MiningMinistry2017

Thank you! I often forget merging nodes. I've merged them, but problem remains... Mess file tells the same as it told before:

DegrÃ© de libertÃ© physique associÃ© au noeud N7 et Ã la composante DX. <EXCEPTION> <FACTOR_11> ! ! ! ! ProblÃ¨me : la matrice est singuliÃ¨re ou presque singuliÃ¨re : ! ! Lors de la factorisation de la matrice, on a rencontrÃ© un problÃ¨me !

! (pivot nul ou presque nul) Ã la ligne 37 qui correspond au degrÃ© de libertÃ© donnÃ© ci-dessus.

Pierre, I'll take a look now on node 7.

MiningMinistry2017

Here it is! Node 7, massN.

It is POI1. As you said it's an orphan node means that there is no connection between the rest of the mesh and this node. Excuse me, I'm newbie, what to do in this case? Do I need to modify my comm-file or maybe the mesh?

Best wishes, Danil.

jeanpierreaubry

if i take your original .med file, perform the Merge nodes and then export the mesh

the problem solves smoothly

try to view your enclosed view with the element number displayed, N7 is not orphan

edit

POI1 means a point ELEMENT standing on a node not a node by itself

anr the martix singularity if any appears on a node not an element

MiningMinistry2017

Sorry, yesterday I forgot to export new med, now there is no errors, thanks!

I get REACTIONS in resu file, but I can't get result in paravis(depl, etc), because somehow I get an alarm that SIEQ_ELNO can't be calculated.

! <A> <CALCULEL2_89> ! ! ! ! Alarme utilisateur : !

I've found another topic, where similar problem appeared, there were said about DEGE_ELNO

So I think I should try using DEGE_ELNO, I'm going to try now

----

upd:

no, the same alarm appears

----

upd:

I tried to rotate the mesh so it would lay along x-axis(because I thought that I get alarm because my frame lays along y-axis and dimensions of beams are 'HY', 'HZ', I thought that 'HY' conflicts with frame laying along Y. But this doesn't work. Anyway, the same alarm and unability to view the result in paraview(but it calculates reactions somehow...)

jeanpierreaubry

but I can't get result in paravis(depl, etc), because somehow I get an alarm that SIEQ_ELNO can't be calculated

no this is not related

why SIEQ_ELNO ? this field is for plate or 3D elements not for beams

why DEGE_ELNO ?

do you really understand what you are doing?

i suggest you read U2.01.05 before trying to use exotic fields

for the moment keep to the fields stated in the book

i am not a specialist of Paravis in every day jobs i am perfectly satisfied with Gmsh

but i suggest to look for Wrap by Vector

MiningMinistry2017

Oh, sorry, I'm an idiot. SIEQ_ELNO was mistake, there wasn't SIEQ_ELNO in the book. I removed it and everything works now, and I also forgot that load was different according to steps. And that's why I saw only blue frame - because time step was 0 :)

Thank you for reference on u2.01.05, need to read it for proper understanding what's happening and what field should be used.

Many thanks for your help!

Best regards,

Danil.

Juselebe

Hello everybody,
Im trying to replicate the frame exercise but got the same error. However, I performed Modification > Transformation > Merge nodes, but It didn´t work. I attached the .med, .mess, .comm and python geom script (without the extension). Could anyone help me?

message.txt

41kB

runcase-4-stage-1.txt

4kB

framegeompy.bin

10kB

jeanpierreaubry

Juselebe
it looks like there are no boundary conditions (ground fixation) nor load in the MECA_STATIQUE, which explains a null pivot (the structure is free floating ins space)
there should be a CHARGE keyword

Juselebe

jeanpierreaubry
Problem solved, thanks!!