[chandra shekhar pant]

Dear All,


I am trying to run the case similar to propeller, instead in my case the geometry is very huge. I had tried with the same set of stls file that are there in the tutorial (propellerTip, innerCylinderSmall, innerCylinder, outerCylinder etc). To see the wakes, tip vortices etc, I need the refinement in the AMI region, close to the the propellerTip (innerCylinderSmall) and in the innerCylinder region, given the refinementRegions of level (1E15 4) and (1E15 3), instead, the SHM is refining the interface.



The screenshot of the mesh and the logfiles of SHM are attached herewith. Also, if I am trying to decrease the level as (1E15 3) and (1E15 2), and correspondingly all the refinement regions by one level, its giving the refinement in the desired location, but I could not resolve the vortices, wakes etc.



Is this because the mesh is heavy my pc is not able to handle or something else is not correct?


Thanks a lot for your time and help.

[simrego]

Hi!


Line 491 in your log:
"No cells marked for refinement since reached limit 10000000."


You already reached the maximum global cell limit here so the refinement is stopped.
Increase the global cell limit (also the local cell limit if needed).
This cell limit can be huge, but after the refinement the cell count will be much smaller since you drop the cells which are out of the domain.


Also a tip:
If you have to decrease every refinement level, it is easier to make the background mesh coarser.

[chandra shekhar pant]

Thanks for the help, but if I increase the value then it complains that it's out of virtual memory, and thereby aborts the SHM. Is there any middle way around? Even though my workstation is about 64 GB RAM, 10 dual processors and 1 TB hard-drive. Do you think that it is sufficient for the meshing or bigger workstation is required?

[simrego]

Maybe you can try to skip that wake refinement, and then use snappyRefineMesh (I have never tried, but you should check it), or simply use the refineMesh utility. With this you can avoid the really big cell count during the refinement (since you refine all cells, even which is not needed and you will drop, but sHM keep them in memory at the refinement stage, and only drop after the refinement.) But if you don't drop that much cells, it won't solve your problem. You have to use a coarser mesh, or another computer...
But if you push the memory to the limit, you could be in trouble during the solution (you have to test it).
I suggest you to decrease or eliminate the wake refinement, do a calculation, and refine only where it is needed.



BTW: Based on my experience, with 64 GB RAM you should be able to handle a mesh up to 30-40 million cells easily.