Hybrid (mixed-element) meshes and orthogonality issues

equon · February 2, 2015, 13:54

Hi all,

I am attempting to calculate the unsteady incompressible flow around a rotor blade using pimpleDyMFoam. I have verified that my boundary conditions are valid by running the grid with the blade surface set to free stream velocity and checking that the domain solution converges to free stream conditions. When I re-run the solution with a normal viscous wall BCs (movingWallVelocity for this case), the pressure and velocity solutions blow up at the tip. The instability appears to originate from the interface between structured and unstructured meshes.

The blade is modeled with a body-fitted structured grid (all hexes) and an unstructured far field composed of all tets. A large number of non-orthogonal cells were reported, typically at the interface between structured and unstructured grids (from plotting nonOrthoFaces). However, it was also observed that these non-orthogonal faces were minimal in number near the blade tip likely due to the more refined geometry.

I've already tried the usual tricks: relaxation factors 0.3 and 0.7 for p and U, as well as increasing nOrthogonalCorrectors. Does anyone have any other suggestions on how to stabilize the solution? Are there any drawbacks to simulating mixed-element meshes in OpenFOAM? (I've attached snapshots of the mesh with suspect orthogonality highlighted, as well as the pressure and velocity solutions as everything is starting to blow up.)

Thanks in advance!

kiddmax · February 9, 2015, 09:13

Hi equon,

Did you mesh the blade by using Pointwise? I also have similar problem. Did you find the solution?

Best regards,
Ye

equon · February 9, 2015, 13:05

Yes I did use Pointwise. Glad I'm not alone. I haven't been able to make any progress so I've taken a step back and I'm trying to simplify my current problem: simpler grid, SRFPimpleFoam instead of pimpleDyMFoam. I've seen similar issues running much simpler problems (e.g. a cylinder) with pimpleFoam, but in those cases the issues at the mixed-element interface were tolerable--perturbations were relatively small and the solution was stable. I think this is the best I can hope for because generating a perfect grid by OpenFOAM standards will be extremely difficult.

February 9, 2015, 09:13		#2
kiddmax Member Ye Zhang Join Date: Dec 2009 Location: Delft,Netherland Posts: 92 Rep Power: 17	Hi equon, Did you mesh the blade by using Pointwise? I also have similar problem. Did you find the solution? Best regards, Ye

February 9, 2015, 13:05		#3
equon New Member anonymous Join Date: Sep 2014 Posts: 2 Rep Power: 0	Yes I did use Pointwise. Glad I'm not alone. I haven't been able to make any progress so I've taken a step back and I'm trying to simplify my current problem: simpler grid, SRFPimpleFoam instead of pimpleDyMFoam. I've seen similar issues running much simpler problems (e.g. a cylinder) with pimpleFoam, but in those cases the issues at the mixed-element interface were tolerable--perturbations were relatively small and the solution was stable. I think this is the best I can hope for because generating a perfect grid by OpenFOAM standards will be extremely difficult.