Corner vs Centroidal data types: what is considered a large difference
When post processing my FEA results, the question comes up quite a bit on whether to look/use the corner or centroidal stress values. I understand the differences between the two types and have read over the years that large differences between the two data types indicates that further mesh refinement is required to get a more accurate view of the stress.
What I've never seen discussed anywhere is what is typically considered a "large difference" between the 2 that should make you consider refining the mesh. For example, a current linear static analysis i am running shows a roughly 45% increase changing from centroidal (31 ksi) to corner ( 45 ksi) looking at max prin stress.