Contrainte équivalente de Von Mises

Your problems regard A) Salome, stress calc with Code-Aster

A) Salome. I know that it is not very robust. Using Salome is very tricky, and You need good nerves.... I sometimes - if solids are difficult to mesh - slice (patition) into Subsolids which I mesh seperatly (several mesh objects in the objects browser, exporting them also for having them saved as *.MED or *.UNV) then I make a Compound Mesh to export and to calculate. Then You must take care (using suitable mesh hypothesis) that the touching faces of the (compounded basic) meshes have coincident nodes, to merge them. If merging is not possible, I use LIAISON_MAIL to glue these touching surfaces together. I could continue to tell a lot of user experience, . . . , but unfortunally I have only low understanding of how Salome works internally, . May be Joel can give some hints to these problems.

stress calc
From Your stress results:

average length
0,05: min=-5,97973e5 max=7,94671e7
0,02: -5,10091e6 1,18812e8

it is clear that the stresses are not converged. -> Use more elements or better suited elements.

For mech calc quadratic elm are better than linear elm
hexa are better than tetra.

If it is difficult to mesh quadratic elm within Salome, You can mesh lin elm and make them quadratic using CREA_MAILLAGE within EFICAS. Then calc Your newly created quadratic mesh, then project the result of the quadratic mesh back onto Your lin mesh.

Question of negative van-Mises:
The amount of Your min-values is about 1% of Your max-values (comparatively small). If Your model is loaded in such a way that there are nearly stress-free regions, I wouldn't say there is no problem..

good luck Johannes Ackva

______________________________________________________________________
Ingenieurbüro für Mechanik
Dr.-Ing. Johannes Ackva
Markgrafenstr. 21
D 91717 Wassertrüdingen

www.code-aster.de