[Red Ember]

High to all
I'm trying to mesh blade inner cooling for CFX, geometry is not simple and there are relatively sharp edges where I can not add fillets. I managed to get mesh for medium quality options (relevance center - coarse, smoothing - medium). I got impressing 9 mln nodes and 25 mln elements. When I tried to use fine and high options, process AnsMeshingServer.exe get stuck at such point:
"meshing completed on 4/5 parts
meshing faces 1174/6843"
I left it on for 2 days and there was no progress. The process was still occupying the same RAM quantityas 2 days earlier.
Is it because of hardware is weak for such task?

Thanks in advance!

[divergence]

Sounds pretty odd to me. My guess would be that the program is in a limbo of some sorts without a way out. Judging from your message, I would say that your hardware is not the problem here. In my experience Meshing has simply ended its process if the memory amount has not been sufficient.

Have you tried meshing the problematic part separately? Are there some mesh definitions that might distract other definitions? If you used DesignModeler to create the volume, was the ShareTopology process succesfull - are there any rogue volumes or surfaces in the five parts?

[Red Ember]

Well, actually there are 2 big parts - inner cooling and external flow, I just divided original body using NX, not design modeler. I do it sometimes to get parts with different sizing options and it often takes more time to mesh separated bodies, but now it was just stuck.
May be I should create 1 part of 2 bodies in DM.
I think there are no rogue volumes or surfaces, cos I succeeded to mesh it using more coarse options and solver started although it failed after few dozens of iterations because of overflow. The problem was about boundary layer that time - first layer should be about 0.001 mm and it caused bad orthogonal quality elements (but no negative volumes yet). So I decided fine sizing options could help.
May be I should use relevance or refinement options?

[divergence]

It could be so that the program is trying to create a mesh consisting literally of hundreds of million cells due to the finer sizing options. Correct me if I'm wrong, but it sounds that you have defined the sizing in the global settings. This could lead in very big meshes. You could combine the two parts and define reasonable global sizing and then fine tune the mesh by setting local sizing into chosen boundaries and volumes. Additionally by combining the parts, you would get a conformal mesh already in the mesher.

The relevance option is a solid choice in my opinion for most of cases. However, I suppose it is not mandatory - each should find their own way to create suitable meshes.

[OI_M5]

I observe a similar behaviour while meshing a relatively simple geometry of some 250 thousand elements in both 18.0 and 18.2.

1. If the generated mesh data are first cleared, the meshing completes. After changes in the mesh details are made, however, the mesher hangs and the process AnsMeshingServer.exe keeps going on without generating any progress. The "Stop" in the progress monitor does not do what it should but AnsMeshingServer.exe keeps consuming CPU as before Stop was hit.

If this process is killed in the TaskManager, after a short while the meshing completes to the end(?) and the message "The mesh translation to Fluent was successful" is shown. Very odd, as one does not know which mesh is now shown, the old one before the parameters were changed or the one after this change?

The whole meshing process is a total frustration under these circumstances and takes much more time than it should.

2. Incidentally, with 18.2 in meshing the progress monitor "ANSYS Workbench Mesh Status" does now show any more. Neither are the parts being meshed highlighted any more. I can not find anything in the documentation nor in the Properties of ANSYS WB on how to switch this progress monitor on or of.

It is also useful to note here, that meshing of the same geometry worked as it should even with many more elements with 18.0. Apparently through a combination of parameters now meshing can not come to completion.

Any suggestions on how to further debug the situation would be welcome.

PS. To point 2 above. The problem is caused by incompatibilities between a remote multimonitor SW and ANSYS. Whereas the main GUI windows of WB are controllable by the user, the progress monitor pane (or other information popup windows) may land in a physically non existent virtual monitor, which is then no more visible.

As many people run WB in various multi-monitor environments it would be beneficial if the user were given direct control of what monitor is used for what purpose.