Sectored Axi-symmetric Model

[ginsoakedboy]

My co-worker is simulating a ring-shaped geometry for me. The ring has a number of holes along the circumference such that a sectored 3D model can be used. The attached screenshots show the geometry as modeled in UniGraphics NX 6.0.

The whole ring.

Cutting planes used for obtaining sectored model.

The final sectored model.

The intention is to apply symmetry boundary conditions on the 2 cut-faces to simulate the entire ring.

Question: Is this the proper way to sector this model for FEA purposes? If the cutting planes do not pass through the holes, can ANSYS/ABAQUS tell that the holes are supposed to be all around the circumference?

 

[ginsoakedboy]

Did I leave something out of the question? Let me know if more information is needed.

 

[corus]

I think this is a case of cyclic symmetry and not symmetry as such.

Tara

 

[ginsoakedboy]

corus, true. I apologize for not using the correct term. Not very familiar with the terminology.

Will application of cyclic symmetry boundary conditions assume that the hole features exist all along the periphery?

Thank you.

 

[flash3780]

corus,
What sort of analysis are you doing? Thermal? Structural? Modal? Is the loading symmetric? Always start by asking yourself "What do I want this analysis to tell me?"

Your part looks to be some sort of spacer. Why not model the mating parts as well?

Cyclic symmetry is the correct technique for simulating partially-symmetric sectors. It's different from standard symmetry, since the solver creates real and imaginary components of the geometry.

For such a simple part, you may actually consider modeling the entire ring, rather than complicating it with cyclic symmetry.

 

[corus]

flash, I don't have a problem, it's the other guy.

My advice is to always use symmetry, cyclic or not. You'll reduce the overall size of the model whilst allowing a more refined mesh in the areas of interest. In addition, using symmetry prevents problems with restraints and rigid body motion in a lot of cases.

For this case, look at the sector you've cut out and see if it can be repeated around the circumference (without overlap) to look like the original layout. The results you'll get will be exactly the same as if you'd modelled the whole circumference, but quicker, and more accurate.

Tara

 

[flash3780]

Tara,
I'm still not sure what sort of analysis you're doing or what you're hoping to learn from it. What is the function of the part, and do the mating parts matter to this analysis? Are you concerned about stresses? Temperatures? Modal frequencies?

I agree that symmetry is often desirable, as it reduces the problem size. However, you also have to weigh the difficulty of advanced modeling techniques against the benefit. You may save a half-hour of runtime, but spend 3-4 extra hours building the model. Cyclic symmetry works well for things like turbine blades, where there is a clear benefit to modeling a single blade sector of the disk versus the entire turbine disk. For such a simple part, it may not provide such a clear benefit. You can probably attain an adequate mesh on the full geometry and still maintain a reasonable run time.

That said, consider the interactions with any mating parts. The part looks like a spacer. Is there a clamp load on it? Does that affect the results? Again, it would be helpful to know what you would like this analysis to tell you.

 

[corus]

For the last time, it's not me that's asking the question. I'll pass on your comments to ginsoakedboy, however.

Tara

 

[flash3780]

Haha oh sorry about that, Tara. I feel a bit foolish for addressing the wrong person.

I don't disagree with you about using symmetry - usually the best course of action. Still, for such a simple part, the boundary conditions would probably be much more straightforward for a full model - less chance for a mistake.

Now, that's depending on the analysis. If he's trying to do some kind of crazy elastic-plastic harmonic response analysis, then yeah, he might want to look into chopping it up.

If it's what I'm thinking that it might be, a static structural analysis of a spacer, any decent machine can chug through that in a reasonable amount of time with an adequate mesh.

 

[ajose]

Hi ginsoakedboy,

There is a diference in ANSYS about symmetry to build the model and symmetry to analize the model. It is easy to check on the project three. One of them is in model the other on analysis.

Cyclic symmetry it is to build the model, so if you have the complete model there is no saving on time computing. You can apply any kind of analysis with no worry.

If you are using symmetry analysis concept you have to be carefull, becuase have to fullfit with symmetryc and aymmetryc loads, symmetryc and asymmetryc boundary condition.

So you have to study your problem and choose the proper condition that matched the theory.

ajose

 

[utkk]

Hi Postginsoakedboy
I think you can be modeled with cyclic symmetry but you must check loading condition . Is loading symmetry?.

Rubber engineering

 

[kellnerp]

can ANSYS/ABAQUS tell that the holes are supposed to be all around the circumference?

Yes, if there is symmetry the symmetrical part will behave exactly the same as the whole if you have symmetrical loading as well. Symmetry will not behave as if the holes are all the way around the model if, for instance, you intend to model a load on just one of the holes.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

http://www.linkedin.com/in/engineeringtransport

"Node news is good news."