Bumpy Wing with snappyHexMesh
2 Attachment(s)
I have been trying to mesh a wing in a wind tunnel using snappyHexMesh. I have an issue where I end up with bumps on the wing. Initially I had problems resolving the sharp discontinuities but overcame these using a mixture of edge snapping and mesh refinement.
The resultant leading edge was highly faceted so I added a searchableCylinder along the leading edge with an extra layer of refinement. When doing this bumps appeared on the upper surface (as shown in the figure). The settings for SHM are shown below. I would prefer not to have a fine mesh over the whole surface as the cell count when we do full aircraft analyses would be too high. Any ideas or solutions would be welcome. Ian Code:
/*--------------------------------*- C++ -*----------------------------------*\ |
HI Ian,
I suppose that the quality controls in SHM will filter a few bad cells (unfortunately you didn't post those settings). When iterating, SHM will try to fill those gaps from the filtered cells which leads to your "bumps" on the surface. Try to switch off all quality controls and see where that gets you. Be sure to run "checkMesh" after SHM which will create "sets" of the bad cells that you may view in paraview/paraFoam. Then slowly raise the quality settings until "checkMesh" tells you an "OK". Greets, Jan |
Hi Jan,
Thanks for the response. I did a few parametric runs this morning varying quality parameters and the quality controls seem to have little impact on the output. The area with the bumps seems to occur where the streaks across the surface meet the refined leading edge area (See bumpywing2.jpg not very clear). So I don't know if there is an issue due to it merging an area with a refinement of level 8 coming into an area where the level is lower. Ian |
Hey Ian,
in your SHMDict you are using feature edges: are any of those possibly running across that part of the surface? Have you had a look at the SHM output esp. at the snapping iterations? You are using 300 iterations for the moving mesh, but only 5 relaxations during one iteration: see if the mesh snaps correctly during those relaxations after a few iterations (IMHO 300 iterations are quite a lot). Jan |
Hi Jan,
The feature edges just define the trailing edges and the tip aerofoil section with nothing on the wing. I agree that the 300 iterations are a lot but I have been trying different SHMDict files to see if any of them worked. I usually run with about 20 to 30. I will check the log files on Monday to see if it snaps correctly. If I run the meshing process without the refined leading edge, I don't get the same effect but I do get a rough leading edge. I wondered if it is possible to run the snap in two sections, first to get a basic mesh then a second iteration that refines the leading edge. When I have tried it, when I run the second snapping run it doesn't seem to refine the leading edge. Another that I might try is to reduce the diameter as it is only really the very leading edge that looks rough therefore it might look better. Thanks for your input. Ian |
Hi Jan
I think that I have solved the problem and your first thoughts on the quality settings seemed to be correct. I had used the quality settings from the tutorial examples. The change I had to make was to minTetQuality, in the examples this is set to 1E-30. If I increase this number (initially set to 0.5) the surface bumps disappear. I am still having issues regarding mesh quality output in checkMesh but I think that I should be able to overcome these issues. Ian |
great to hear my note was of help! Happy snapping!
|
similar issue
Hi folks,
I'm experiencing the same kind of bumps when meshing a (non-circular) pipe. Unfortunately I couldn't fix it by changing the mesh quality settings. Also, the meshed surface is perfect, if I mesh the outer region, not the inside. Not very useful for an internal flow problem, though... Any suggestions how to tackle this problem? Or any ideas for a workaround, like splitting the mesh after meshing inside and outside? Help would be greatly appreciated! Cheers, naval edit: flipping the stl's surface normals didn't change a thing |
Hi Sebastian,
The settings that I used to achieve a good mesh and pass the checkMesh (except for 8 skewed cells) include setting minTetQuality -1.E+30 (not 0.5 as I originally set in my reply the parameter seems to be highly non-linear and avoiding it seems to work okay). I tried to keep the mesh resolution down to about 1 or 2mm then gradually coarsen the resolution until it becomes poor, then I step back a stage. In the end I had up to 12 mesh refinement levels on the surface. Another area that we found was that the fineness of the underlying stl seemed to impact the resulting mesh so I used an FE preprocessor to put a fine surface triangulation down then exported this. Another thing to check is that your locationInMesh point is inside the mesh for when you are meshing the inside of the tube. I hope this helps, Ian |
Thanks for the quick reply!
It turned out to be the blockMeshDict, though. I did set inlet and outlet to type empty (to apply wallLayers there as well), that seems to mess up everything. A little disturbing, though. And a bit of a problem, since that is the only way I know to generate proper wallLayers close to another patch. Cheers, naval |
All times are GMT -4. The time now is 11:39. |