[davidwilcox]

Is it possible in ICEM to create a surface mesh and then fill it up using the Volume mesh method? I read on one of the Forums here that you would need to do a volume mesh first, delete the volume, smooth like crazy and then do a delauney. My question is : Can u do a surface mesh and then do a delauney without having to go through the Octree method first?

[cfd seeker]

Yes you can do it. First create the Shell Mesh and then directly fill the volume by Delauney Method. There is also a tutorial that came with the Ansys Help Package Wing Body tutorial about it.

[siw]

I've always used the octree volume mesh then deleted the volume cells method because I find it makes the better quality surface meshes (after the checking and smoothing). All that's needed is a watertight domain from the build topology. I've tried the surface mesh and then Delaunay but I found the octree volume generated surface mesh always was of a higher quality and nicer element distributions. Also you are forced on the mesh sizings by the octree rule, which is a good thing for me as I'm someone who tries to justify their mesh sizings and if I'm given too much freedom I'll spend ages trying too many variations and never get around to the CFD.

The main down-side is that it can take up a lot of PC memory with all those volume elements, which your PC might not have. But just last week I made a bunch of 10-12 million element meshes on my 4 core 20GB RAM PC this way and each octree meshing time was a good excuse to take a tea break.

[Far]

Is it possible in ICEM to create a surface mesh and then fill it up using the Volume mesh method? I read on one of the Forums here that you would need to do a volume mesh first, delete the volume, smooth like crazy and then do a delauney. My question is : Can u do a surface mesh and then do a delauney without having to go through the Octree method first?

Yes. I have gone through this method due to lots of problems with octree. One of them is memory issue as suggest by Stuart (siw). My first problem was setting the approapriate sizing which I didnt know due to lots of different parts of different sizes. If I set too fine mesh, computer goes for meshing forever and if I set coarse size it did not captuture the many features. So after tyring many times, I switched to shell meshing and then delaunay. See thread here http://www.cfd-online.com/Forums/ans...ula-one-5.html

But the real advatange of ICEM is in using hte OCTREE method if you can think in power of 2 and you can probabaly mesh the very complex model in less than hour and submit to solver.