CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Fluent Non-Positive Volume (http://www.cfd-online.com/Forums/fluent/99574-fluent-non-positive-volume.html)

eishinsnsayshin April 6, 2012 13:18

Fluent Non-Positive Volume
 
I'm trying to run a Fluent simulation and am having some trouble getting my solution to stop diverging. I created my 3D mesh entirely using ANSA13.2.1 (both surface and volume). When I do a mesh check in ANSA, there are no intersections, the skewness is below 0.8, there are no duplicate nodes, and it finds no negative volumes. I import my .msh file into Fluent 13.0 and there are no errors. When I run my model I get an epsilon divergence about 200 iterations into solving. I know the parameters/boundary conditions are set correctly because I have a very similar model meshed with tetras (instead of hexas as in this model) that I use a journal file to read in the bc's.

The only thing in fluent that hints at the source of the problem is if I do a mesh check. It says "warning: non-positive volumes exist", and lists the minimum volume at about -2.03e-03 (m^3). It also finds about 15 left handed faces. Perhaps I have a problem with my mesh. In ANSA, I can't find any issues though (as I described above). For the left handed faces, I use the orient tool in ANSA to ensure all my normals are pointed in the right direction, so I'm not sure where it's getting left handed faces...Does anyone know if there is a way for fluent to spit out coordinates of problematic areas so I can go look up that region in ANSA? My model is pretty large (60 million cells) so if there was some way to pin point a problematic region, it would be great.

I'd really appreciate any advise!

vangelis April 9, 2012 07:21

Hi there.
In ANSA v13.2.x you can check for left-handed elements if you activate this criterion in F11 Solids tab.

Does your model consist only of trias and tetras, or do you have other elements as well?

Also I would like to ask if it is a matter of precision.
ANSA v13.2. is double precision. If you have a great spread of length scales in your mesh, you could try to staet Fluent in double precision mode and read and check the mesh again.


All times are GMT -4. The time now is 20:24.