Hi Thomas,
Ok, so a litle bit background about me. I am a strength engineer and work with FEM evryday. Usualy it's NASTRAN, but I also have quite experience with ANSYS, ABAQUS and Optistruct. And this is also my problem, I am to much influenced by those FEM.
I have started to learn code_aster to use it for my hoby projects. And to keep my brain busy with new stuf.
The preprocessing in code_aster is not very different. I think it is very comfortable to work with the element and node groups, it is similiar with ABAQUS and CalculiX. Only limitation is 8 chaaraters for naming.
The basic concept of CA, that it is nearly independent from any pre processor software is a BIG plus. Only thing what is needed is the mesh, which, for simple cases can me defined by hand.
For meshing of the complicated geometries I am usualy using more than one software: cgx, Salome and gmsh, all of them are quite capable and getting better with evry new release.
The pre processing cabapilities of the CA are quite impresive, creating new groups, transforming mesh, etc, no other FE solver, to my knowleg, has such capabilities.
Computation makes also no big problems. Only thing one has to be caruful is nonlinear and large displacements, especialy if shells, plates, beams and rods are mixed in the model.
The bigest diference with other FEM, even with open source is, how the results are obtained, and that's where I have diffculties at the moment. As I mentioned before, I am to much influenced by other FE software. I will try to list how are the results basicly obtained in FE software I have experience with:
ABAQUS, Calculix: results are devided in dependece if they are node or element based
*node
displacement, reaction forces, etc
*element
stress, strain, sectin forces, etc
NASTRAN, Optistruct: There are no separate request filds, for evry result type there is one control comand
STRESS(), STRAIN(), DISPL(), ELFORCE(), etc. for each comand with additional options
For code_aster, If I have understood it correctly :) the results depend not only if they are node or element based but also on element type, like beam shell and solid. SIEF_ is stress for shells and force for beams. And it is very often necessery to have intermidiate commands, like calc_champ and post_champ, before one gets the results in impr_resu.
Regards,
Sergej
p.s. If I may add few wishes I would like to see in CA, better composite capabilities, I know it's not a priority for EDF but still the best from all of the opensource FE I know, inertia relief capabilities and forces due to rotational accelaration not only centrifugal forces.
