When you are requesting Stress output from a solver, which one do you give consideration to - Element centroidal stress OR Maximum Nodal stress OR Averaged Nodal stress?
Which one in your opinion gives best correlation with strain-guage test data?
For structural solvers, e.g. Radioss bulk or Nastran i request stresses on corners. (Control Cards->GLOBAL_OUTPUT_REQUEST->Stress = YES->Location = Corner)
Then you get element centroid and corner stresses.
What you are doing at postprocessing is your choice.
We are using normally corner datas and switching on averaging (simple or advanced)
Averaging is a separat topic, especially since you can do a lot of mistakes. But usefully applied averaging improve the results a little bit.
For correlation with test datas i can't give a global answer, because it depends on the geometry at the location of the strain-guage (hot spots) or the simplification of reality in your FE-model.
Altair Employee
Hi,
For structural solvers, e.g. Radioss bulk or Nastran i request stresses on corners. (Control Cards->GLOBAL_OUTPUT_REQUEST->Stress = YES->Location = Corner)
Then you get element centroid and corner stresses.
What you are doing at postprocessing is your choice.
We are using normally corner datas and switching on averaging (simple or advanced)
Averaging is a separat topic, especially since you can do a lot of mistakes. But usefully applied averaging improve the results a little bit.
For correlation with test datas i can't give a global answer, because it depends on the geometry at the location of the strain-guage (hot spots) or the simplification of reality in your FE-model.
Regards,
Mario