Hi!
Visually the mesh used in the first study looks much finer (i.e. much higher node count) than the mesh used in the second study. I’d suggest using comparable (and optimally the same mesh) for the test.
