CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Problem with mesh generation (

Nestor October 30, 2006 06:32

Problem with mesh generation
Hi there,

I've got a problem while trying to mesh a volume. I just want to know what this means:

"Initialization failed; perturb boundary nodes and try again.

Initialization failed to mesh 1 nodes. ERROR: TG_Mesh_Domain failed with error code 1. ERROR: Tetrahedral meshing has failed for volume entrada. This is usually caused by problems in the face meshes. Check the skewnesses of your face meshes and make sure the face mesh sizes are not too large in areas of small gaps."

The 2d elements' skewness are below 0.8, and the most part below 0.6. I've made sure that, more less, in the zones with small elements, there are other elements of the same size. But I donīt get any good result.

The former 2d meshes for the volume are Tri Pave based.

Any suggestion?

Jason October 30, 2006 09:22

Re: Problem with mesh generation
A 2D skewness of 0.8 is high when you're using that as your 3D base, and if it's in an area where the geometry is complex, this could be the problem. I typically shoot for a maximum of 0.7 (and I only allow a couple elements to be that skewed...). That's just based on personal experience.

If fixing that element doesn't fix the problem, then I recommend decomposing the volume into smaller volumes (if you use the split command with the connected option on, and don't define any BCs on these 'internal' faces, then Gambit will write the geometry as one continuous volume).

Decomposing the volume into smaller ones does two things... 1st, Gambit's meshing algorithms aren't the most robust (Fluent support will often recommend TGrid instead of Gambit for 3D meshes) so when you decompose the geometry, you create smaller, simpler volumes to be meshed, which takes the pressure off the algorithm. 2nd, it allows you to isolate the problem area. If you fix that cell with the .8 skewness and the problem still occurs, then you really don't know where the geometry is failing... there are ways of systematically searching the volume to try and find it, but those are tedious and time consuming... splitting the geometry is often quick and can narrow your search from hundreds or thousands of surfaces to a handfull of surfaces.

Hope this helps, and goodluck, Jason

Nestor October 30, 2006 13:21

Re: Problem with mesh generation
I really thank you Jason, I'll try to split the volume in simpler ones (anyway, one of them is gonna be a little more than simple) and search something about TGrid. My objectives are serious, but still not professional, I'm just a beginner.

I'll tell here how it evolves.

thanks again,

have a nice day :)

Nestor November 1, 2006 03:54

Re: Problem with mesh generation

At last I've got to mesh the volumen, dividing it into smaller ones but also reducing the number of elements in each edge.


All times are GMT -4. The time now is 11:42.