FE Mesh convergence based on un-average min and max principal stress ratios at corner nodes?

[AeroStructAnalyst]

Has anyone come across this FE mesh convergence approach which is based on the ratio of the un-averaged min/maximum principal stress ratios at the element corner nodes or at integration points?

 

[burningDiesel]

You should do the convergence study on the measure that you are analysing.
If you are analysing strain, perform it on strain.
Mises on mises, max principal stress on ... you get the point. No idea why you would do it on the ratio of min/max principal, but if that is your application ...
Unaveraged should be always the case.
On integration points preferably (although if your mesh is converged most of the time the extrapolations to the points are 'good enough', esp. for linear calcs)

 

[inline6]

i sometimes look at unaveraged stress to see if my mesh is fine enough. i rarely do convergence studies usually just start coarse and then then once i'm happy everything works as far as BC, loads, deformed shape etc i just choose a finer mesh. with experience you dont need to do a full convergence study you can choose a suitable mesh by looking at the problem.