I'm trying to make use of symmetry and anti-symmetry when modelling quite a large model. Let's say I have a cylinder, which is geometrically symmetric. Then I have two loads, with a tangential load, a radial load and an axial loads on the same position away from the symmetry line. The problem is that the tangential load is asymmetric on one side. I have been trying to split the analysis into two parts by using an anti-symmetric boundary condition in one case, and a symmetric boundary condition in another case and then sum the results. I can't think of how to do this, however.
Any suggestions would be appreciated. I am attaching a very simplified image to show the force vectors.
Thanks
Björn
Where did the reply go? Weird.
Anyway, in this particular case it's actually an MES, but fully constrained and without any non-linear effects and rather small deformation so it might as well be a linear stress analysis. But usually these analyses that we are conducting involve lots of surface contact and large deformations, so what if I would like to do this in a MES instead?
Thanks
Björn
I assumed you are doing LSS and I replied in a fast way without carefully considering the load situation. And I deleted it.
If it is MES model, anyway we do not support cyclic symmetry yet. So never mind the disappeared post :-). Sorry.
-xli
And for your more detailed description of what your intend to obtain from analysis, in MES particularly involving to contacts, you probably need full model. No idea to simplified the model.
To deal with none symmetry loads on symmetry structures, there is a method called semi-analytic FEA. It models the structure in r-z dimension to reduced size by 1 order. And then sum solutions by Fourier series (loads are expensed in series too). But that whole theory is based on linear principal too. No similar method exists for nonlinear analysis.
-xli
Bjorn,
I think the real problem is that your tangential load is not symmetric and it is not antisymmetric. If it were antisymmetric, the tangential load at the "8 o'clock" position would need to be pointing in the "radial direction" -- the exact opposite vector of the tangential load at the "4 o'clock" position.
You should do a simple test to confirm if this is correct. (I did not try it.)
But IF you have a linear example where some loads are symmetric and other loads are antisymmetric (or cyclic symmetric), then you can analyze each load separately in different design scenarios. In the Results environment, use the "Utilities > Load Combination". This lets the user combine multiple linear stress results together to create a combined result. This utility is based on the method of superposition, so it is only valid in those situations. Likewise, symmetry, antisymmetry, and cyclic symmetry are only valid if the geometry, loads, AND RESULTS fit those criteria. I do not know how your system is responding, but I wonder if the surface contact would violate one of these conditions. (Contact is nonlinear, so superposition would not apply in some cases.) Again, a small test model would be a good idea.
The load combination utility is not available for MES since superpositioning of the results is rarely valid.
Thanks, both of you. I suspected a full model was needed, but I wasn't sure. Nevertheless, I will try the load combination method through a linear analysis and compare it with the results from a full-model with contact. With a bit of luck, maybe the results might agree.
Regards
Björn