[piotr.mecht]

I was struggling with turbulence parameters (like omega) exploding into infinity, quickly after simulation starts, and i have found it is caused mainly because of single bad cells on the surface layer (concave, underdetermined, slightly non-orthogonal cells), due to poor evaluation of wall functions on these cells.

I'm currently working with snappyHexMesh and I really can't force any settings that would help me avoid this thing happening. I've managed to track the evolution of these cells during meshing and i'm pretty sure they come from 'underdetermined' cells in castellated mesh stage [PICTURE], which in this case are the cubes with 4 walls on the boundary layer. These walls get flattened over the geometry surface and they create rather undesirable angles, which cause even greater issues when sMH is in the layer addition stage.

I cannot flatten the whole surface, becasue its curvature is the object of my interest in this simulation (the flat surface in the front view is the symmetry plane). I tried to recreate this shape in blockMesh, but it is near impossible, mostly due to high aspect ratio between perpendicular cyllinders and this curvature...

Is this problem common? Is there any meshing setting i should be aware of? Is there a tool for automated/manual removing these cells maybe?