Simulate a flow over hemi-sphere
 

hi everyboby, I'm simulating a flow over a 3D hemi-sphere in two velocity direction.
With the direction is perpendicular to the arc plane, I got the Cd is about 0.42 which approximately with experiment results. It is good result.
However, with the remain case, when the velocity is perpendicular to vertical plane, I got the Cd is about 1.277 ( while experiment result is about 1.12). In this case, I use k-omega SST model.
Should my mesh has something wrong or should I use another model?
I attached me mesh for this case.
Please help me solve this problem. Thanks in advance :)
my mesh : http://www.mediafire.com/?7q4x1t7iyy8s7av

Mesh resolulation in wake and boundary layer
 

Your mesh on surface of REV is good. But I see problems in wake region and boundary layer. I would suggest you to mesh in ICEM Hexa and use hemisphere shape at inlet and cylindrical shape at outlet.

You can also find the thread on meshing the same geometry in ICEM on cfd online ansys meshing forum and also I have uploaded YouTube video on the same.

http://www.cfd-online.com/Forums/ans...-tunnel-2.html

In fact, if you smart enough, you can use same files with little modification (mostly on geometry part as per your dimensions)

thanks for your advice, I'll try it.:)
And I have one more question: whether haxagon mesh is better than Tet mesh for fluid flow? :confused:

Mostly it is better than the tetra, specially for the cases similar to yours.

well, I think I have to change to Hex mesh :(

Only if you can justify the extra pain and time ;)

For drag prediction, I guess, Hexa is must. See AIAA drag prediction workshops.

Should I mesh a boundary layer on the hemi-sphere? The mesh,which I attached, was generated using Size function only.
:confused:

can you show that mesh?

Can you attach size function in the boundary layer and wake region? If so you can successfully predict drag. However hexa mesh mesh is always needed in the boundary layer region.

here is my mesh :)
http://www.mediafire.com/view/naq828...08k-mesh_2.msh
Is it good enough to calculate?

One more thing, I know we create a inflation to model the boundary layer, but in my case, the flow seperates quickly because of the geometry of the body. So, if it is neccesary to create inflation on the surface of hemi-sphere :)