Learning Forum

[jiayangchng]

[Rob]

Orphans tend to be formed when the solver can't pair the cells or rather can't figure out how to pass information from one cell to the next. It's typically caused by the cell size difference at the transfer point being too large.

[jiayangchng]

Hi Rob, thanks for the tip.
I have meshed both background and component to the same size in the picture above. Even reducing the background mesh slightly still left me with some orphan cells. Could it be due to the different type of mesh - structured for the background and unstructured for the component? If so, can that be overcome?

[Rob]

It's unlikely the cell type is causing a problem, but matching sizes can be more difficult. Have you inflated the overset (airfoil) mesh?
Turn on the expert options, /define/overset-interfaces/options expert yes in the TUI and see if the verbosity function will give you more information. Looking at the mesh the orphans are far too close to the airfoil - how big is the component fluid domain?

[jiayangchng]

I have tried applying an inflation layer before, but the mesh near the trailing edge is very skewed - so currently it is not inflated. Would an inflation layer help with the overset?
Yes you are right, some of the orphan cells are on the airfoil. The computational domain is rather small (approx 0.8m x 0.8m x 1m) as I am working with the student license and the number of cells are limited.

[Rob]

Inflation won't help the overset but should improve the drag predictions.
OK, the domain size may explain the orphans. There needs to be a number of cells between the overset walls and background mesh. If the domain is over constrained you may prevent this.
Is the aerofoil moving? Just wondering if you need the overset.

[jiayangchng]

Thanks! Adjusting the domain size helped with removing the orphan cells.
The airfoil is not moving, however the research involves overset - hence utilizing it.
One more question, I am now trying to create an inflation for the airfoil. I now want to generate a structured mesh using hex dominant method but an inflation layer is not valid for that method. How may I overcome that?

[Rob]

OK, I was wondering as we tend to try and keep commercial models "simple" to get the results more quickly. With academic studies they're often too simple or excessively complicated!
If you're in Workbench Meshing do NOT use the "hex dominant" method. It's for Mechanical and produces meshes of dubious (dreadful) quality. I don't know what shape the zone is so it's hard to judge but inflation may break sweep/multizone due to the way the source and target faces interact.

[jiayangchng]

The geometry is shown below.
Yes - sweep/multizone resulted in the inflation being invalid. Sizing the curves individually doesn't seem to work too

[Rob]

It won't. For swept inflation you don't inflate the surface, it's in the training examples and/or documentation somewhere. You inflate the edge onto the source face and sweep that. That'll not work here anyway, you'll need to do some decomposition to get a swept mesh and even then I'd expect a few triangles/wedges to get an acceptable cell quality.