FEM Mesh Dependency Test

[mechanicaljw]

Hello All,

I have done an FEM simulation and have this question. Is there a way you can know that your solution (Stress analysis or modal analysis) is mesh independent? I have varied the mesh resolution and each time i am getting a variation mostly in the first and second mode frequencies. The third, fourth, fifth and sixth seem to be changing less. So I am wondering whether the various values i am getting are physical (real) or non-physical (due to numerical errors, etc).

Any hint on how to get a mesh independent solution in FEM would be appreciated.

Thanks in advance!
Jimmy

 

[rstupplebeen]

Typically you continue to refine the mesh (mesh sensitivity study) until the pertinent outputs no longer change within your desired accuracy. I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[missil3]

How much are your results varying ? 5% or 10% or 15% ? For our analysis, we stop refining the mesh when the difference between two consecutive refinements is less than 5%.

So as rstupplebeen also says, the mesh you choose should depend on your desired accuracy and available computational power.

 

[IceBreakerSours]

Nonlinearities, if not modeled properly, can result in noisy output like contact pressure. Mesh refinement does not help in such a situation.

http://www.eng-tips.com/faqs.cfm?fid=376