[philus]

Hi all,

my question is quite general and about the strategy on how obtain the best mesh:

Imagine to have the geometry shown in the picture attached: a semi-spherical surface (zero-thickness), a cylinder and a rectangular prism. I am interested to the external aerodynamics of these objects and I want split the computational domain along the plane of symmetry also shown in the picture for decrease the total number of cells. I expect a prism layer mesh on each surface except the volume walls. The rectangular prism must be also meshed in its interior, I hope then a conformal mesh between the inside and the outside of this solid. Imagine also that all geometries can be quite dirty so they may requires a surface wrapping.

I tried the following strategy:
1. wrap sphere and cylinder
2. wrap rectangular prism
3. create fluid block defining the symmetry plane
4. under part operation subtract all geometries to fluid block
5. under part operation intersect rectangular prism and fluid block
6. send intersection and subtraction to region
7. create interfaces selecting prisms surfaces in both regions
8. change fluid block surfaces in subtraction region to symmetry in order to avoid prism layers
9. create automated mesh (surface remesh + repair + polyhedral + prism layer + thin mesh)

But I'm not happy with that because the mesh isn't conformal on each side of the rectangular prism, because the semi-spherical surface now has a thickness (corresponding to the 25% of the wrapping cell size I think), and because I don't have any prism layer on surfaces.

You, lovely user familiar to Star-CCM+, which strategy would you prefer to correctly mesh my problem?

Thanks in advance,
philus

[philus]

news:

I discovered I have to imprint intersection and subtraction otherwise I cannot get a conformal non intersecting mesh!

thanks to cd-adapco support!