[CIG-SHORT] RuntimeError: Determinant of Jacobian (1.74667e-08) for cell 0 is smaller than minimum permissible value (1e-06)!

[Lucas Abraham Willemsen]

Sorry, I realize the words I used were somewhat poorly chosen.

"The resulting mesh elements have great aspect ratio" with 'great' I mean good. The elements are very close to being cubes so the aspect ratio is close to 1.


________________________________
From: cig-short-bounces at geodynamics.org [cig-short-bounces at geodynamics.org] on behalf of Lucas Abraham Willemsen [lawillem at MIT.EDU]
Sent: Saturday, March 16, 2013 23:37
To: cig-short at geodynamics.org
Subject: [CIG-SHORT] RuntimeError: Determinant of Jacobian (1.74667e-08) for cell 0 is smaller than minimum permissible value (1e-06)!

Hello,

I'm trying to use Pylith in combination with a flow code to model the Cappa and Rutqvist 2011 paper. This is a plane strain problem with no derivatives in the out of plane direction. The flow code however only works with hexes. So I extended the mesh in the out-of-plane direction by a very short distance.

I first generated the mesh with the journal file 'alternativeapproach_loft.jou'. Cubit has features for extruding a 2D quad mesh into a 3D hex mesh. There is a (confirmed) bug however which destroys any refinement done. So I try to work around that by using 'loft'. The resulting mesh is sometimes multiple elements thick in the out-of-plane direction. This could explain the error I got in the subject of this email.

I then decided to completely rewrite the meshing script from scratch. The result is in 'mesh_python.jou'. This can be run by using the python version of the CUBIT journal editor. I use a rather fine mesh in the entire domain now instead of using refinement, because of the previously mentioned bug. The resulting mesh elements have great aspect ratio. The out-of-plane thickness of the mesh is 10 meters. The thickness of the mesh in the out of plane direction is exactly 10 meters (1 element) as well. The hexes are approximately cubes with 10x10x10 dimensions. When I try to run Pylith I still get an error like:

RuntimeError: Determinant of Jacobian (1.74667e-08) for cell 0 is smaller than minimum permissible value (1e-06)!

This message doesn't really help me a lot. I have no idea why the Jacobian is bad, when the mesh elements seem to have such a nice aspect ratio.

I'd be very grateful if someone would be able to find out where things are going wrong. The journal files are very long and a similar result could probably have been achieved by much less code, but I don't see mistakes in it. Especially the 'mesh_python.jou' is long, but straight forward.

I was in the process of changing the pylithapp.cfg from 2D to 3D, but since the mesh gets rejected I have not been able to run it and verify that there are no more mistakes left in it.

cheers,
Lucas


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geodynamics.org/pipermail/cig-short/attachments/20130317/7a575f03/attachment.htm