Orthogonality and grid expansion control
 

Hi all,

I am relatively new to meshing in ICEM CFD. I have followed Simon's airfoil tutorial for a problem I am working on. I have run into an issue while trying to enforce cell orthogonality in conjunction with a set grid expansion ratio. I am hoping to have my first layer (ideally the first 2-3 layers, but I will be happy with just the first cell) orthogonal while maintaining a maximum grid expansion ratio of 1.2.

I will be using an in-house FEM solver for this mesh, so with ICEM I am only concerned about the near-wall element geometry, and not so much whether or not the mesh is structured or unstructured (I will convert as necessary). The unstructured function "Smooth Hexahedral Mesh Orthogonal" preserves my first-element thickness very well, but abandons the 1.2 growth rate I established in the pre-mesh. If I use the structured function "Smooth Multiblock Domains Globally", I can retain the growth rate and first-element spacing, but I cannot restrict the smoothing to the near-wall area. Instead it wants to make all of the cells at the perimeter of each subface orthogonal, meaning that the elements at the edge of my boundary layer also try to establish orthogonality (leading to unnecessarily skewed elements in the mesh--see picture).

Basically, if I am using a blocking approach, is there a way to achieve near-wall orthogonality WITHOUT using a smoothing function? I feel like I have exhausted all of the options associated with each smoother. Am I missing something? Any advice is appreciated. Thanks!

I don't have time to look into your smoothing question right now, but as for how to get it without smoothing...

You could add an edge split just out from your airfoil and then match edges between the surface of the airfoil and the split so that the mesh is orthogonal between them.

Thanks for the tip, Simon! My initial attempts to match edges to a nearby split are having trouble due to the curvature of the airfoil, but I will keep playing with it and update here if I figure it out.

Yea, the outer split is a little longer, so you need to adjust the initial spacing slightly to get it all proportional and orthogonal...

Just look at the edge length and use that over the original edge length to calculate a factor to multiply the initial spacing by... The rest of the distribution will already adjust proportionally...