inverted cells in Gambit
 

Hello all,

I would be very grateful for some help here if anyone knows how.

I'm building up a structured mesh consisting of a number of connected volumes. Some of these volumes however, when meshed, report inverted (negative volume) elements. Some of these I realize are because the volume geometries are not ideal for obtaining a structured mesh and I have tried my best to improve them or the way in which they are meshed. Some of the volumes, however, are relatively simple in geometry and should be allowing nicely structured meshing yet report every element of the mesh to be highly skewed and inverted.

What I am wondering is first of all why Gambit is doing this and secondly whether there is any tool in Gambit that can amend these inverted cells. I've tried the mesh smoothing tool but to no avail.

Fluent cannot read the mesh structure I have buit and I am sure it is due to this issue. The error reported is 'Cell Centroid is xc 300.876183 yc 10.127939 zc 10.439100WARNING: no face with given nodes. Error: Build_Grid: grid error. Clearing partially read grid.

Error: Null Domain Pointer'

I would be extremely grateful if anyone can help me here.

Thanks a lot.

Re: inverted cells in Gambit
 

The only thing you can do is remesh your geometry. A lot of times what happens is there are edges or vertices that are really close together. When you go to apply a mesh to these faces, then it puts a node at each vertice, and these are extremely close relative to the size of the mesh you want, which creates skewed elements. Use the examine mesh tool... button at the bottom right of the screen... to locate the problem areas and understand what's going wrong that way you can figure out what you need to do to fix it.

Goodluck, Jason

Re: inverted cells in Gambit
 

thanks a lot Jason - a gentleman

Re: inverted cells in Gambit
 

well when u r creating a sturctured mesh , you got to be shrewd for complecated geometries, the very first thing that you can do (if u r not already doing) is to spend time on thinking how you can further subdevide the geomtery into smaller volumes, with the property that each subvolume is a topological cube. (that is it should have 8 vertices or 4 in 2d, and 6 faces).

so if u have a cube and u substract a sphere from it, it is not a topological cube since it might show u 8 vertices but now it has 7 faces.

once this thing is done, and all the faces are connected, creating a structured emsh is then a work of minutes if not of seconds.