My results are changing with change in the density of mesh in the non linear analysis. There is no option of P and H controls in the non linear analysis. What can I do to get stable results.
The model is a large sheet of 2012 mm length x 992 mm width x 5.8 mm thickness. the sheet is fixed at four supports and have a pressure of 2400Pa acting on the top surface in the downward direction.
I will appreciate any comments/suggestions or tips.
Thanks in advance.
Refine the mesh in the critical areas and re-run the study.
Thank you Janko for the suggestion. The problem is when there is a singularity or very high stresses in a region then further refining the mesh in the region only worsens the results. That's what happening when I am trying to refine the Mesh in the critical areas. And the results are not stable, as I am increasing the mesh density the stress is also increasing.
Well you can always ignore that part of the model if you know the result is nonsense. However if you do need an answer then that depends on what you're modelling, I don't think there's a simple answer.
However have a look around this guy's blog: Stress singularity - an honest discussion | Enterfea it's full of practical advice for problems such as yours.
Thank you Eric for the suggestion and the link. It is very informative link. Unfortunately I can not ignore the area where I am having this problem because those are the same areas from where the part fails. However, I tried non linear material and now I am getting close results which are also stable. So it seems like the high stress area did pushed a very small area into plastic deformation. But my main concern was stress which is now under control.
Great result Ehtesham!
I'm glad my advice was helpful.
Refining the mesh will also refine the results . Maybe use node to node contact instead of surface also if you see that the result of different meshes are close and not having high difference then assume it as right output. Also check singularity as mentioned!
Thank you Husein for the suggestions. I did tried node to node contact and it was still the same.Also the difference was quite high between the two different mesh results so cant assume any answer to be the right one.
As I mentioned in my reply to Eric, I was able to solve the problem after taking non linear material.
The good news is that is what is supposed to happen. You're not getting worse results; you're getting more accurate results.
Keep remeshing and refining until the stresses stop getting bigger. If possible, use auto-refinement. I usually go the full five rounds.
But it takes longer!!!
What good is a quick analysis if the result is wrong?
Thank you Roland for the suggestion. I actually have the empirical results already from the experimental analysis. Refining the mesh further was just making the stresses bigger, I have tried about 20 rounds with different mesh densities the result was changing in each round.
The answer is what I have replied to Eric. The areas where the stresses were infinite are the points where actually a small amount of plastic deformation is happening. So I used non linear material properties as mentioned in the article and it worked. The results are stable and are very close to the empirical results.
Well you can always ignore that part of the model if you know the result is nonsense. However if you do need an answer then that depends on what you're modelling, I don't think there's a simple answer.
However have a look around this guy's blog: Stress singularity - an honest discussion | Enterfea it's full of practical advice for problems such as yours.