Mesh sensitivity study VS sub-modelling

[yudhichen]

Dear FEA user,

I'd like to have your opinion regarding mesh sensitivity study vs sub-modelling.
I've just started doing FEA analysis and AFAIK, it is normal that we run analysis 2 or 3 times with refined mesh
to determine what is the best approximate solution.

on the other hand, I also heard there's a technique called sub-modelling in which
same principle applies but the model being simplified (cut from original model), refine the mesh
to get solution particularly at our point of interest with still carrying the initial fea result.
(i use ABAQUS software for doing this, but i believe other software, i.e. ANSYS also has similar feature)

The way i see these sub-modelling is way more efficient while serving the same purpose.
In case i do sub-modelling,do I still have to run mesh sensitivity?

Can anyone share experience which method better?

Thank you

Yudhi

 

[gravityandinertia]

They aren't separate concepts. You always need to determine mesh sensitivity. Sub-modeling is based on the fact that displacements converge faster than stress values. Therefore for a large assembly you can use less mesh, then break up the submodels and start refining mesh on the submodels to speed up solution times.

 

[yudhichen]

Thanks