da ansys 11 a 13

[stef_design]

Hi, I recently installed ansys 13.
I have done an old analysis (made with v.11) and I find a difference in the result of the max soll (almost double).
Why do you think?
by whom I trust:

 

[Zaxxon]

Is mesh the same?

 

[erman70]

try sending the file to assistance?

 

[stefano.garbin]

from v11 to v13 change the default mesh criteria.
If you compare the size of the 2 mesh you will see that they are different.
also changes the default management of contacts.

 

[stef_design]

from v11 to v13 change the default mesh criteria.
If you compare the size of the 2 mesh you will see that they are different.
also changes the default management of contacts.

Okay, but the final result must not change.
if an analysis made 2 years earlier by 35 mpa as max effort and then the same model 2 years later (with a new version) by a max almost double effort, of whom do I trust? ! ?

 

[stefano.garbin]

good practice: when you do an analysis and get results then try to relaunch it by varying the mesh to see how sensitive the results are!
if you arrive at a point where changing the mesh results vary, we put, + or - 10% then your mesh is ok and also results!

what you claim is that different solutors (or different versions of a solutor) give the same results with the same mesh!

 

[stef_design]

good practice: when you do an analysis and get results then try to relaunch it by varying the mesh to see how sensitive the results are!
if you arrive at a point where changing the mesh results vary, we put, + or - 10% then your mesh is ok and also results! !

Thank you, this is a good advice! :finger:
often you tend to think that a thick mesh is always better.

what you claim is that different solutors (or different versions of a solutor) give the same results with the same mesh!

with identical mesh what do you mean? the same parameters?
Because that's what I did. in my test the mesh is by default but between ansys v11 and v13 the difference is seen.

 

[stefano.garbin]

No. not the same parameters.
identical mesh = same number of knots, same position as the nodes therefore same number and position of the elements.
in the most recent versions of ansys the meaning of the parameters (e.g. the density "coarse", "medium", "fine") is a little changed, in the sense that, in order to exploit the greater computing potential of the new computers, at equal parameters are generated more dense mesh.

 

[stef_design]

No. not the same parameters.
identical mesh = same number of knots, same position as the nodes therefore same number and position of the elements.

I think it's almost impossible to have an identical mesh between two versions of ansys.

in the most recent versions of ansys the meaning of the parameters (e.g. the density "coarse", "medium", "fine") is a little changed, in the sense that, in order to exploit the greater computing potential of the new computers, at equal parameters are generated more dense mesh.

I didn't know that. Interesting!

 

[pizzaf40]

Forgive me but I ask for confirmation.. .

If a mesh, even if not exactly with the same number and position of knots, is sufficiently thick to approach the asymptote of the solution, then the problem should not be manifested with these errors.
If 11 from a half result should mean that mesh is very rare in critical areas. Have you tried to check with a refine at applied around the maximum area?

 

[stef_design]

If 11 from a half result should mean that mesh is very rare in critical areas. Have you tried to check with a refine at applied around the maximum area?

Hi, I didn't touch the mesh.
is a default mesh without various settings. the parameters are the defaults of the program.
was a quick test:finger:

 

[pizzaf40]

what I mean is that the default of what from lesser result can be unsuitable to the situation since from images there seems to be the difference fundamentally in the peak of tension, that is where there is the bending with connection that from concentration of tension.

If you do a "refine at" in the offending area and the result comes back to approaching the other (or even a little over it) should be all right.. .

 

[stef_design]

what I mean is that the default of what from lesser result can be unsuitable to the situation since from images there seems to be the difference fundamentally in the peak of tension, that is where there is the bending with connection that from concentration of tension.

If you do a "refine at" in the offending area and the result comes back to approaching the other (or even a little over it) should be all right.. .

OK I try to make a try:finger:
Thank you for the answer.
you were very kind:smile:

 

[pizzaf40]

I hope to be useful