Modal Analysis with Ansys classic

Not knowing any more than what you have described, the only suggestion I can offer is to look for finite element analysis software and use it.

Modal analysis is dead easy in ANSYS Workbench. These days I tend to do most preprocessing in Workbench and modal analysis in workbench then transfer to Classic for psd.

That doesn't help you though.

It would certianly save time/space to simplify your "plate like structure" to the barest minimum. (say flat plate & beams). And eliminate all uneccessary geometry: rounds, lips etc...

I assume that you've imported from CAD...If you can, have the the CAD jockey do the simplification, sit with him. If they bitch that eliminating rounds will lead to an inaccurate analysis, tell them to get you a suitable computer.

Without looking at the model I don't think anyone can advise you as to whether analysing a section of the disc is valid. Is it symmetrical?

Yes the component is symmetrical, It contains a hole, it is pipelike in form at hole, it is platelike at the outer part. The 2 forms join with an angled part from one end of the pipe shape to the outer plate shape.

The easiest was to describe it would be to use 6 xy plot points. If points 1-2 are the central x-axis and 3-6 is the basic section shape of the structure. It does contain 12 support ribs equally spaced around it, that stiffen it up. Points 3-6 are revolved 360 degrees around the x axis (points 1-2)

pt----x----y

1----0-----0

2----50---0

3----0----20

4----15---20

5----50---40

6----50----50

I have drawn a basic thin shaped section in ansys and attempted to apply an axi symmetric mesh to it for analysis? This doesnt seem to work, I take it that a 2-d axisymmetric type model is not possible for modal analysis.?

I have been attempting to load a simple 2d .dwg representation into Ansys 11 classic (I dont have 3-D modeling capabalities at the moment). I plan on revolving into a 3-D model in Ansys. Im not sure if this is possile, I only seem to have options to load in 3-d models, Im probably missing something obvious??

Thanks a lot for the help

I suppose you could split the model down the middle and run alternate symmetric and anti-symmetric restraints along the mirror-line but you you won't gain anything in total solution time. Best thing to do is keep the model down to the bare minimum to represent the distribution of mass and stiffness reasonably. It of course depends just to what accuracy you need to find the eigenvalues.

Alas with the info given there's no way we can proffer a length for your peice of string!

I think I'm right in saying that a problem can only be treated as axi-summetric if both the model and the loadings are axi-symmetric, as in the case of internal pressure in a cylinder or centripetal, or maybe thermal loadings. Hardly any of the likely modes of your structure could be described this way.

I'm sure your Ansys tool will have the capability to revolve or spin (and to extrude), but the easiest way is to import decent 'coherent' surfaces from a CAD system. Sorry. I''m not versed in this software. If it was Hypermesh and I was after the shape yo've described I don't think it would be a long job to construct it from scratch. If you want to create an true axi-symmetric shell model from scratch, model a semi-section in, say, the global XY plane by placing points (or nodes or grids) half way between the inner and outer surfaces of the section. Join the points with 1-d elements (e.g. beams) and then spin or revolve the 1d elements about the axis of rotation to create 2d shells.

Sorry if that sounds too basic.

PS Axi-symmetric models were just the job for training on a 32k PET (remember those) but probably of limited use in the big, wide commercial world. Most self respecting PCs these days shouldn't have trouble running models with between 1.0E+04 and 1.0E+05 nodes these days?? OK in your case it might depend on how many eigenvalues you're trying to get, but if you're looking at higher frequencies, just set the lower frequency limit in your modal analysis so that you only extract the ones of interest.

If you end up using a section of the model the frequencies you get will be the same frequencies that you would get with the full model, as long as you are careful about how you set up your boundary condition. The only issue is that when you model a section based on symmetry, the symmetric BC's will only produce symmetrically shaped modes, so some modes can be missed.

Yes, modes will be missed and in this case it could be the majority. It's not quite so bad if you're using a plane of symmetry, because the modes you're missing would likely be reflections of modes you found. My understanding is that axi-symmetric isn't so convenient.