Hello, what about to solve one STAT_NON_LINE with high soft spring stiffness and then reuse the step with different AFFE_CARA_ELEM where the stiffness will be low for the same loading.

hello,
you can just do

STA_NL = STAT_NON_LINE(....)

...

STA_NL = STAT_NON_LINE(reuse=STA_NL,
                 
                       ETAT_INIT=_F(EVOL_NOLI=STA_NL),

I think I never had a problem when doing contact with symmetry, just when you want to compute reaction, you have to use contact's RN and not FORC. And you don't have to exclude nodes lying on the symmetry.
Problem comes with different bc like LIAISON_MAIL or if slave and target have common nodes.

Solidworks and other commertial softwares do it differently. If you change something all the entities that were in some feature stay there or there will be some message that something is missing. Shaper just remembers what entities names were there.

Yeah, and the file has to be named differently otherwise it will not update.

If you want compatible mesh from two imported shells just apply partition. Or, you can project the mesh from one edge to another.

I thing there exists some mesh converters to med format.
Vega Translator from *.nas to *.bdf
Meshio
export from Ansys to cgns

For example, I prepare the geometry completely in external cad and then import it to shaper, and do partition. Then, I just create groups for meshing and bc.
Like I said before, if I change my cad and update the import, if the changes are "not significant", it will all update as well.
The only problem is with the references done using names.

Hello, what about displaying the pressure from CALC_PRESSION only on elements where is contact status 1 or 2?
Of course CALC_PRESSION will give results in all areas not only in contact area, because it calculates stress in normal direction.

Hello,
no surprise since you have in EXCIT only charge

reslin = MECA_STATIQUE(CARA_ELEM=elemprop,
                       CHAM_MATER=fieldmat,
                       EXCIT=(_F(CHARGE=charge),
                              _F(CHARGE=encastr)), # added
                       MODELE=model)

All right. Correct solution is this.

Somehow friction helps.

Thank you,
however I don't think that using CONTROLE is all right. You just say give me results after 10 iterations, but it doesn't mean it has converged.

however if I understand it correctly this is not SIMP, or some algorithm where the stiffness of the elements just changes and there is no actual removal of the elements. I think that would be better.

then probably not, but an orthotropic material might work. Could you homogenize the composite?
I don't have any experience with composites in CA.

if you use FORCE_CONTOUR, it is as if you applied pressure on the whole annulus, so total force is  FORCE/(area of the whole annulus, circle in your case).

With force_nodale it would be FORCE/nodes I guess.

Just do some simple case in 3d and compare.