Hi!
There should be two main factors here:
- For a Re = 100, external flow is laminar, however the turbulence model is currently set to k-epsilon:
- Especially for validation cases, it’s always recommended running mesh sensitivity studies. You mentioned you have already tested different meshes, however 900k cells is definitely not the limit. Mesh convergence varies quite a bit from problem to problem.
I’d recommend running some tests, applying additional refinement around the sphere (including here the region downstream from the sphere).
Cheers