I'm sure I'm misunderstanding, but are you referring to a 2-D axi-symmetric model? Or are you looking for "off-axis" (off the global axis) symmetry? If so, is setting up a local co-ordinate system and tying your boundary conditions to the local system not an option?
Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.
For cyclic symmetry for a slice model in a cylindrical coordinate system you have 2 ways you can go in FEMAP depending on the type of loading. For a sector that is loaded symmetrically you can simply use a slice of a few degrees and update the displacement coordinate system to a cylindrical system. The boundary conditions on the symmetry plane are component 246 set to zero. At R=T=0 set these to 12456 = 0.
For true cyclic symmetry using NEiNastran V8.4 you skip the boundary conditions and just use new CYSYMGEN param which automatically sets up MPC equations on the boundaries which force cyclic symmetric behavior.
I just found out about the CYSYMGEN param from Google. Prior to 8.4 I was wondering how this was done via FEMAP as setting up in the preprocessor would allow running the model in other codes.
Garland, the model has torsion applied to it so ordinary symmetry conditions would not represent the behavior of the problem correctly. Cyclic symmetry makes the nodes on the symmetry planes move together in the appropriate degrees of freedom.
Since a couple of you helped with this I thought I would share the solution step by step:
1. In FEMAP create a cylindrical coordinate system on the axis about which symmetry exists
2. Mesh the model in this coordinate system
3. In FEMAP, MODIFY/UPDATE OTHER/OUTPUT CSYS set all nodes on the symmetry boundaries (or just all nodes) to the coordinate system defined in 1. If not CYSYMGEN will not work.
4. In FEMAP add these two lines assuming the coord. sys. in 1 is numbered 3.
PARAM, CYSYMGEN, 3
PARAM, CYSYMTOL, 1E-4 or whatever is appropriate.
I'll add one more thing to this tip.
The definition coordinate system also has to be the same cylindrical system as the output coordinate system or CYSYMGEN doesn't seem to be able to pick the right nodes.
If you set this parameter:
TRSLMODLDATA=ON
you can check that the MPCs where created correctly.
