Author |
Message |
Marc Zwaanswijk (marcz)
Member Username: marcz
Post Number: 4 Registered: 11-2004
| Posted on Wednesday, December 15, 2004 - 09:52 am: | |
I uses a 2D rectangular model with three layers to simulate solute groundwaterflow. My width is 2250 meters and my height is only 32 m. When FlexPDE tries to build this mesh I get the error message: BAD NEIGHBOR PATCH ---Called from coremesh::fix_nabor ---Called from coremesh2::rehook_tri3 ---Called from nodefront2::scan_legs2 ---Called from nodefront2::relax_hooks ---Called from compmesh2::newmesh When I take the height 10 times bigger for each layer there are no problems but that will not simulate my exact problem. How to solve this?? Thanksss |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 287 Registered: 06-2003
| Posted on Wednesday, December 15, 2004 - 06:26 pm: | |
We made some changes in the mesh generator in version 4.2.5 to handle floating features, and apparently have broken the mesh generation for very thin layers. If you can email me the offending script, I will try to use it to hunt down the problem.
|
Marc Zwaanswijk (marcz)
Member Username: marcz
Post Number: 5 Registered: 11-2004
| Posted on Thursday, December 16, 2004 - 03:17 am: | |
With this message I attach the offending script. Thanks for help. greetz Marc
|
Marc Zwaanswijk (marcz)
Member Username: marcz
Post Number: 6 Registered: 11-2004
| Posted on Thursday, December 16, 2004 - 07:30 am: | |
Sorry, I attached the wrong script in my message. The script i attach in this message is the offending one. I hope you can help me with this neighbor because when I increase the height 10 times the CPU time to get results will be 10 times higer to I think;) Thanks, greetz Marc Zwaanswijk |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 288 Registered: 06-2003
| Posted on Thursday, December 16, 2004 - 08:50 pm: | |
You didn't say what version of FlexPDE you are running, or on what operating system. I ran the problem as you posted it on version 4.2.4 and 4.2.5 on Windows and 4.2.5 on Linux and experienced no gridding failures on any of them. After running a while, your omega variable shows considerable oscillation, so I suspect you have a term wrong in your equations. You should not assume that ten times taller will run ten times longer. Long skinny things tend to run longer than squarish things, because they are less amenable to iterative solution (fewer information flow paths). Anyway, the long run time is because of the instability, not the shape. |
Marc Zwaanswijk (marcz)
Member Username: marcz
Post Number: 7 Registered: 11-2004
| Posted on Friday, December 17, 2004 - 02:50 am: | |
My problem ran on version 4.2.3 3D prof. on Windows 2000. I read that you haven't check my problem on this version so I want to ask you to do this too. Or do you advise me to download version 4.2.4 of 4.2.5? I uses a trial version of this version 4.2.3 3D professional. Does my trial license also work on versions 4.2.4 or 4.2.5? or have order a new license. Thanks for help, greetz
|
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 289 Registered: 06-2003
| Posted on Friday, December 17, 2004 - 03:44 pm: | |
I ran this problem as posted on version 4.2.3 on windows 2000 without gridding failure. Are you sure you don't have hardware trouble? |
Marc Zwaanswijk (marcz)
Member Username: marcz
Post Number: 8 Registered: 11-2004
| Posted on Saturday, December 18, 2004 - 01:00 pm: | |
Hmm strange But anyway thanks for help. I will check on other problems. greetz |