Pointwise non orthognal face
 

Dear all,

I followed a pointwise video tutorial online, which is generating a hybrid mesh for a horizontal wind turbine. What I want to do is similar as the tutorial. Extruding structured boundary layer mesh over the blade surface, and filled the farfield with unstructured mesh.

My problem is when I import the mesh to OpenFOAM solver, it says there are a lot of non orthogonal faces. After several test, I found that these non orthogonal faces are mainly at the boundary layer mesh. Do you have any idea how to control these faces when you extrude 3D boundary layer mesh in Pointwise? Can I check the non orthogonal in Pointwise?

Thanks a lot!

Best regards,
Ye

Ye,

Just to be clear, the poor quality cells are in the structured blocks? Or are they in the tet block adjacent to the prism block? I'll guess and suspect you mean the later.

Matching up structured meshes to tet mesh at times can be tough. The typical culprit is high-aspect ratio quads on the outer boundary of the structured blocks. Recall that when you make a tet block that has structured domains in it's boundary, Pointwise has to insert pyramids above the quads. Pointwise will automatically guess an adjacent pyramid height but if the pyramid is built on top of a high aspect ratio quad while itself has a low aspect ratio (height/base width) this will often be the cause of the high non-orthogonality you see in OpenFOAM.

Note: you can check the shape of the pyramids but selecting the tet block and entering Grid>Solve. Pointwise will draw an outline of their shape.

You can attempt to fix this in two ways. Manually adjust the pyramid heights in Grid>Solve, Attributes or modify your structured mesh so that it has more uniform quads. The later will require that you adjust the structured domains on the blade and re-run the extrusion.

Try option 1 first because it's easy but more than likely option 2 will work the best.

-Chris

I'll throw in one more helper. There's a script from Pointwise's last webinar to compute the non-orthogonality metric before exporting to OpenFOAM. You can grab it here:

http://www.pointwise.com/webinar/mixing/

Dear Chris,

Thank you so much for your reply. Actually I think the poor meshes which has large non orthogonality are in the boundary layer. Not on the interface between structured and unstructured mesh. I used hexa mesh in the boundary layer region, which means the surface mesh is quads, not triangle. i saw from the tutorial you shown me, they use triangle surface mesh to avoid high aspect ratio, right? Indeed, when I use quads mesh and extrude the boundary layer mesh, the unstructured adjacent mesh also have bad quality. I will try it like the tutorial did.

Thanks again!

Ye

The approach in the video was structured hex BL on the blade, structured hex BL at the blade-hub junction, prism BL on hub. For the later the boundary of the prism was matched to the structured blade-hub junction block.

If you can, post a picture of the problem area and I might be able to provide better guidance.

-Chris