POU_D_T Ball Joint

max123

Hello,
I'm trying to couple two POU_D_T Elements with LIAISON_DDL as a balljoint .
The structure is a simple frame:

But the I get "la matrice est singulière ou presque singulière " as long as I don't add DRX and DRZ to the connection.
I can also omit DY and DRX,DRZ to get it work.
But just the 3 translational DOF fails.

Any idea what I'm missing?
Regards
Max

` LIAISON_DDL=(

        _F(                                                                                                         
            GROUP_NO=('BJointLeft', 'BJointRight'),                                                                 
              DDL=('DX','DX'),                                                                                      
              COEF_MULT=(1.0,-1.0),                                                                                 
              COEF_IMPO=0.0                                                                                         
    ),                                                                                                              
        _F(                                                                                                         
            GROUP_NO=('BJointLeft', 'BJointRight'),                                                                 
              DDL=('DY','DY'),                                                                                      
              COEF_MULT=(1.0,-1.0),                                                                                 
              COEF_IMPO=0.0                                                                                         
    ),                                                                                                              
        _F(                                                                                                         
            GROUP_NO=('BJointLeft', 'BJointRight'),                                                                 
              DDL=('DZ','DZ'),                                                                                      
              COEF_MULT=(1.0,-1.0),                                                                                 
              COEF_IMPO=0.0                                                                                         
    ),         `

jeanpierreaubry

max123
my own guess is that the problem maybe lying in the ground fixation of the frame

none of the left or right part must be allowed to free float in 3D space

max123

thanks Mr Aubry!
the left and right ground is fixed in all 6 DOF (see below).
It is strange that if I omit to couple DY in LIAISON_DDL the solver calculates and the result seems ok as long I have no force in y-direction.
The frame can also be calculated if the coupling of the nodes is omitted in total. The frame is split in two independant systems and can be calculated.
Only the coupling of DX,DY and DZ is not working.
Regards
Max

LIAISON_OBLIQUE=( _F(GROUP_NO='PLeftG', DX=0, DY=0, DZ=0, DRX=0, DRY=0, DRZ=0, ANGL_NAUT=(0.0,0.0,0), ), _F(GROUP_NO='PRightBottom', DX=0, DY=0, DZ=0, DRX=0, DRY=0, DRZ=0, ANGL_NAUT=(0.0,0.0,0), ), ),

jeanpierreaubry

max123
in which plane is lying the frame?
and the use of LIAISON_OBLIQUE seems to me very strange
it would help if you shared the files
however to mimic this ball joint i usually use a small K_TR_D_L element as explained in my book

max123

jeanpierreaubry

the frame is in XZ-plane.
What is strange with LIAISON_OBLIQUE ?
The file is generated by TUBA.py from [https://tuba-v2.readthedocs.io/en/sm2017/#](https://)
Just found "Adding end release to the top bar" . Will try to understand
thanks
max

frame-aster.txt

15kB

(can't upload the med-file)

jeanpierreaubry

max123

What is strange with LIAISON_OBLIQUE ?

is not DDL_IMP much better in your case?

without the mesh we do not know where are the groups so it is difficult to know what is going on, it is so easy to make a mistake!

another thing, your .comm file contains a lot or irrelevant lines

max123

jeanpierreaubry
according to the docs LIAISON_OBLIQUE without ANGL_NAUT or with 0,0,0 should do the same.
I tried it also with DDL_IMP but still LIAISON_DDL with coupling all 3 trans dof fails, while only using dx and dz it is working.
I will try to simplify geometry and comm file ...

jeanpierreaubry

max123
ok for LIAISON_OBLIQUE but why make so complicated things

again without the mesh it is difficult to tell

i wonder how yo would cope with your method of modeling beam joints in a structure like the attached pics
there are around 350000 beam elements and a few thousands beam joints oriented in different angles
with small discrete elements and a clever grouping almost everything is set at mesh time

max123

I'm searching for a solution to couple a POU_D_T with another just by selected DOF.

frame-all-post.bin

4kB

frame-all.bin

9kB

If in frame_all.py the lines at LIAISON_DDL for DY are uncommented the simulation fails.

salome_meca -t frame_all.py

thanks for your support
max

jeanpierreaubry

max123
try doing what i wrote previously

max123

jeanpierreaubry

ok, but do I have missunderstood something with LIAISON_DDL ?
Why does aster can't find a solution for all translational DOF coupled ?

With this file I tried using a K_TR_D_L between the Left and Right Ponit as a hinge.

frame-all-k-tr-d-l.bin

9kB

It is working with VALE = ( 1E6 , 1E6 , 1E6 , 1 , 1 , 1 ) but setting rotational stiffness to zero fails.
Regards
Max

jeanpierreaubry

max123
setting a value to 0 means completely free l stiffness is some case that may lead to null pivot
so set a very small value
do you read the .mess file when it fails?
that gives some hints why it fails

max123

jeanpierreaubry

don't know ho to write into mess file via python

max123

jeanpierreaubry

I just can find:
Degré de liberté physique associé au noeud N12 et à la composante DY.

But don't know what that could mean.

La matrice est de taille 280 équations.
Elle contient 2556 termes non nuls si elle est symétrique et 4832 termes non nuls si elle n'est
pas symétrique.
Soit un taux de remplissage de 6.163 %.
Degré de liberté physique associé au noeud N12 et à la composante DY.

╔════════════════════════════════════════════════════════════════════════════════════════════════╗
║ <EXCEPTION> <FACTOR_11> ║
║ ║
║ Problème : la matrice est singulière ou presque singulière : ║
║ Lors de la factorisation de la matrice, on a rencontré un problème ║
║ (pivot nul ou presque nul) à la ligne 82 qui correspond au degré de liberté donné ci-dessus. ║
║ ║
║ Risques et conseils : ║
║ * Si la ligne correspond a un degré de liberté physique, il s'agit probablement d'un ║
║ mouvement ║
║ de corps rigide mal bloqué. ║
║ Vérifiez les conditions aux limites. ║
║ Si vous faites du contact, il ne faut pas que la structure ne "tienne" que par le ║
║ contact. ║
║ Vérifiez également les caractéristiques matériaux (module d'Young, ...). ║

jeanpierreaubry

max123
i will stop helping if you do share the mesh!!

max123

jeanpierreaubry
ok, this is a misunderstanding. Unfortunately I can't upload a mesh-file.
with the python-file I have sent you can generate the mesh and run code_aster.
Singularity> salome_meca -t frame_all.py
frame.mmed will be generated.
Now I also uploaded it here:
www.frenell.de/download/frame.mmed

or do you know how to upload a mesh file here in the forum?
regards
max

jeanpierreaubry

max123
with the attached mesh and the previous frame-aster.comm
and despite the numerous irrelevant lines in it
this runs without any trouble on my linux komputer (attached view)
can you share the .mess file you get

max123

jeanpierreaubry
yes, but if You omit in LIAISON_DDL DRX and DRZ the simulation fails.
If you omit the entire LIAISON_DDL the simulation is ok.
But I just don't understand why the system is not sufficiently determinded when using DX,DY and DZ.
In my mess file I can see:
" Degré de liberté physique associé au noeud N12 et à la composante DY."

jeanpierreaubry

StanislavPavlica

max123
this was the problem
the mesh created by Tuba seems completely wrong in the group assignment although it seem right at first look

so the best way is to forget about it and start with a clean geometry either wit Gmsh or Salome and the mesh it

and try to remove all the useless lines generated by Tuba in the .comm file

and finally use discrete elements for the joints it is so much more powerfull

StanislavPavlica

Hello max123, jeanpierreaubry,

there is some problem with mesh file, probably no 0D Elements?

Try this med file.
I created it in gmsh, it has only groups of 0D Elements so I included line:

_F(TOUT_GROUP_MA='OUI',),

in line 24 frame-aster.comm:

MAIL1=DEFI_GROUP(
        reuse =MAIL1,
        MAILLAGE=MAIL1,

        CREA_GROUP_NO=(
_F(TOUT_GROUP_MA='OUI',),
            _F(  NOM='GPOINTS',
...

to create groups of nodes from groups of 0D Elements.

I commented out line 442 too, in salome meca 2024 which I'm using is VISU_SP unsupported.

P.S.:
It looks like you are using mm/sec/tonne/°C UNITS system.
Shouldn't there by E=2.1E5 for steel in frame-all.py ?

Trying to put rar file on forum:

1748013667-855684-frame.rar

6609B

Stano

max123

StanislavPavlica
Dear Stano, thanks for the hint. E should be 2.1E5. You are right.
So You could run the simulation with DX,DY,DZ fixed?

With the frame.mmed you sent, I get the error:
"le GROUP_MA V0 ne fait pas partie du maillage : 00000001 "

The groups are named "BOD1","CIA2" etc ?
There is a section with POI1 that is missing in my mesh. Does this makes a difference?
POI1
M1 N1
M2 N2
M3 N3
M4 N4
M5 N6

regards
max

max123

jeanpierreaubry

ok, but would be good to know, what is wrong to be able to fix it within Tuba.
As TUBA is based on salome , how would you correctly build in salome?
Isn't it a downside to implement a 1D element just to be able to link two nodes?
thanks max