Author |
Message |
Svitlana Subota (photinka)
Member Username: photinka
Post Number: 5 Registered: 02-2007
| Posted on Wednesday, February 25, 2009 - 05:32 am: | |
When I use "transfermesh('transfer_U0_3D.dat',U1)" I get an error "cell 2829 is inverted". What is the problem if the geometry is the same in both (output and input) programs? |
Marek Nelson (mgnelson)
Moderator Username: mgnelson
Post Number: 113 Registered: 07-2007
| Posted on Wednesday, February 25, 2009 - 01:52 pm: | |
If you post the output and input scripts, we can take a look at this. |
Svitlana Subota (photinka)
Member Username: photinka
Post Number: 6 Registered: 02-2007
| Posted on Thursday, February 26, 2009 - 09:20 am: | |
|
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 1218 Registered: 06-2003
| Posted on Friday, February 27, 2009 - 12:13 am: | |
I believe the trouble stems from the fact that you have a plane surface and a spherical surface becoming tangent at the top center. This kind of gradual approach to tangency is difficult for the mesh generator to deal with. In this particular case, I think that on read-in, roundoff error is making the zero-thickness points appear to have crossed. There are several ways you can deal with this: 1) increase the thickness of the top layer so the top surface is not tangent to the sphere. 2) place a circular void patch at the top center, so that all mesh cells have non-zero thickness. (radius=4e-4 seems to do the job). 3) truncate the spherical surface with a small flat spot at the center, so that the tangency is avoided. I have done the first two of these in the attached files, and both transfer successfully.
|