multi mesh configuration
31 views
Skip to first unread message
abraetz
Apr 3, 2008, 9:45:33 AM4/3/08
to FDS and Smokeview Discussions
Dear all,
I placed a simple drawing 'mesh-alignment.jpg' in the files section
which shows a configuration of three meshes. Can FDS handle such a
configuration or should I avoid such constellations?
Regards
Andreas
I placed a simple drawing 'mesh-alignment.jpg' in the files section
which shows a configuration of three meshes. Can FDS handle such a
configuration or should I avoid such constellations?
Regards
Andreas
F-Sim.de
Apr 3, 2008, 10:25:00 AM4/3/08
to FDS and Smokeview Discussions
You are using more than 2 cells the two mehes at the right side?
Less than 4 cells caused problems at least in FDS 4.
P.S. interesting article at FZ-Juelich!
Less than 4 cells caused problems at least in FDS 4.
P.S. interesting article at FZ-Juelich!
dr_jfloyd
Apr 3, 2008, 10:34:34 AM4/3/08
to FDS and Smokeview Discussions
Cells need to line up like building blocks. So it is OK to have one
mesh large than another, but it must be by an integer multiple and the
gridlines must line up. There is no restriction in terms of that
integer multiple, but you as the user must satisfy yourself that a
particular gridding strategy is resulting in good answers.
mesh large than another, but it must be by an integer multiple and the
gridlines must line up. There is no restriction in terms of that
integer multiple, but you as the user must satisfy yourself that a
particular gridding strategy is resulting in good answers.
abraetz
Apr 4, 2008, 8:06:03 AM4/4/08
to FDS and Smokeview Discussions
The picture in mesh-alignment.jpg is just a modification of figure 6.2
in the User's Guide to FDS. Consider all of the meshes to consist of
cells of the same size. The main difference to the first grid
configuration in figure 6.2 is that there is a certain point where
three rather than only two meshes are connected. Is there a
restriction that only two meshes should connect or are there no
limitations to that particular issue. I ask this because I encountered
a numerical instability with FDS version 5.0.x some weeks ago. I found
goofy velocity vectors at that point where three meshes met.
Unfortunately the numerical instability appeared very late in the
simulation so I had no time yet to try the case with FDS 5.1.4. We
then just reorganised the meshes, to avoid points with more than two
meshes abutting. Thereafter the numerical instability did not occur.
Please let me know if it is unclear what I want to know. I will upload
a more definite sketch to the files sector if necessary.
in the User's Guide to FDS. Consider all of the meshes to consist of
cells of the same size. The main difference to the first grid
configuration in figure 6.2 is that there is a certain point where
three rather than only two meshes are connected. Is there a
restriction that only two meshes should connect or are there no
limitations to that particular issue. I ask this because I encountered
a numerical instability with FDS version 5.0.x some weeks ago. I found
goofy velocity vectors at that point where three meshes met.
Unfortunately the numerical instability appeared very late in the
simulation so I had no time yet to try the case with FDS 5.1.4. We
then just reorganised the meshes, to avoid points with more than two
meshes abutting. Thereafter the numerical instability did not occur.
Please let me know if it is unclear what I want to know. I will upload
a more definite sketch to the files sector if necessary.
Kevin
Apr 4, 2008, 8:28:24 AM4/4/08
to FDS and Smokeview Discussions
Your mesh configuration is allowed. The general idea is that each
boundary grid cell should have a clearly defined "neighbor" with which
to exchange information.
boundary grid cell should have a clearly defined "neighbor" with which
to exchange information.
Kevin
Apr 4, 2008, 8:36:43 AM4/4/08
to FDS and Smokeview Discussions
Let me clarify my last post. The idea of mesh alignment is that each
cell FACE that is on the boundary of two meshes should have a clear
opposite neighbor.
cell FACE that is on the boundary of two meshes should have a clear
opposite neighbor.
Reply all
Reply to author
Forward
0 new messages