CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   3d geometry mesh generation (

mach13 September 6, 2009 04:04

3d geometry mesh generation

I have a 3D diffuser - collector geometry. Using Gambit - Fluent. Having difficulty meshing - should I go by edge- face - vol sequence,or direct volume; scheme:cooper or hex core ----I'm confused.How do I validate my mesh - Is monitoring of fluxes after each refining,enough for this? Please help.

And no books - or any user guide give any input on practical geometry meshing. Request advice on this too.... does it come by practice?

-mAx- September 7, 2009 01:17

if you attempt to mesh your volume on the fly, I assume Gambit won't mesh it properly (especially if your geometry is complex)
What kind of issues do you have with Gambit?

mach13 September 7, 2009 09:49

3d complex geometry meshing
Thank You Max,
Thanx for the reply. I tried Cooper Scheme and Hex Core on my diffuser - collector combo.Both are giving me errors - mainly taking up too much memory for just about 3,00,000 nodes - I assume my meshing schemes are wrong. How do I trouble shoot it, and how can I be sure of my mesh before Fluent analysis.:confused::confused::confused::confused:

Thanx again and really appreciate your helping all us 'idiots' on this forum- seen most of your replies to other queries.:):):):)

-mAx- September 7, 2009 10:19

memory problem has nothing to do with cooper.
So first, if you get this memory issue, then you may have to divide your domain into subvolumes and treat each subvolume one by one. (it will ask less memory)
Second Cooper can only be used if your volume is Cooper-able... I mean with projection source, surround surfaces which are mappable, etc...
I don't know how looks your geometry but if you want to mesh it "quick" I would use the tet-hexcore scheme.

mach13 September 8, 2009 05:19

3d mesh gen
Hi Max,
Thanks again. That memory problem has been sorted out - was b,coz I've been foolishly using lower interval size. I had tried Tet-hex Core also after that, but got errors regarding geometry - but the scheme is meshing part of my geometry. My geometry is a diffuser inside a collector, viz like a diverging passage inside a semicircular prism - will try sending a graphic tiff to u - if u could give ur mail ID - only if its OK with you.I will work by trial n error till I succeed, and will let you know what happened.

Thanks n regards....

-mAx- September 8, 2009 07:01

what does gambit prompt after meshing?

mach13 September 8, 2009 12:49

3d complex geometry meshing
Hi Max,
Thanks.I think the problem was with default settings. I had to adjust those - tetmesh_pyramid_angle...etc. Part of my geometry I've meshed with Cooper and part with hex core. grid check looks OK.But how do I know for sure my grid is OK. My plan of action is to use the same schemes with lesser interval sizes and then compare the fluxes for grid correctness - am I right in my approach?????


-mAx- September 9, 2009 01:38

check the skewness of the mesh.
I take care having max below 0.9 (but some people, may have mesh with skewness below 0.8)

mach13 September 9, 2009 11:04

Thanks Max,
U are right. Skewness is a problem, but right now I made a very course mesh of 1.5 lakh elements - out of which 100 or so are above 0.97. I think it will reduce once I 'fine' it to 10 lakh or so...Am i right?

Also this mesh is giving me negative values at my pressure outlet in FLUENT. Will fine meshing sort out this too ?????

-mAx- September 10, 2009 01:43

what does it mean?
Skewness 0.97 is too high
What kind of BC did you set at your outlet?

mach13 September 10, 2009 11:03

Hi Max,
Thanks again. Lakh means 100000 cells.I've given pressure inlet at inlet and pressure outlet at the outlet. I know following values:- mass flow(kg/s),pressures - both in & out,mach nos - in & out, temp - in & out.


All times are GMT -4. The time now is 12:30.