Static Load for VERY organic assemblies

[antran7]

Can anyone recommend or give the pros/cons of FEA packages for structural FEA of very organic assemblies?

I've been using Pro/Mechanica to simulate the deflection for a given point load of a three part assembly. The problem is: the geometry is created in Alias and are very organic (like jewelry). I usually have to fuse all three parts together, and simplify some of the surfaces just to get it to mesh. And then some more simplification to get it to run the analysis.

Ideally, I wouldn't have to do much with the Alias file (given in IGES form), and the program could run a structural FEA with three parts with three different assemblies, and it could figure out the contact regions (that could slide relative to eachother), with very little time spent on my part (not just because I'm lazy, but also because the FEA will be used in the iterative design stage).

Is it just inherant to all FEA programs that complicated surfaces take alot of tweaking to get results (some of these surfaces have vertices that are less than 0.5 degrees).

Is it also inherent that assembly FEA's are orders of magnitude more finicky than part FEA's? (not just in how long it takes to run, but also whether it can run or not).


To help simplify: I'm running pretty small deflections and the loads don't require non-linear analysis. I'm also comfortable with ignoring the friction between the sliding parts - I just can't let two objects occupy the same place at the same time.

Any input or insights would be really appreciated.
Peace out!

 

[GregLocock]

I very much doubt that the stresses and strains in an FEA model give two hoots about whther parts overlap, or about faces with an included angle of 0.5 degrees either. OK, I suppose if you are doing optical FEA they might (does anyone do that?).

Have you demostrated to your own satisfaction that such a fine model is needed? How on earth do you get loads and boundary conditions that support such finicky details?

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[antran7]

Maybe dealing with 0.5 degree angles is a weakness of p-code in Pro/Mechanica. But everytime there's a surface that has the crazy aspect ratio that 0.5 included angle induces, it causes the mesh to choke (or it meshes, but the analysis freezes).

So I go into the model, and eliminate the acute angle by patching a rectangular surface over it. Once I do that to all the acute angles, Pro/M will run and give results (surprisingly accurate... at least by my standards within 20%)

I think the fine model is needed because these crazy surfaces contribute to the structure of the parts. And since I'm interested in the total deflection for a given force, I'm trying to limit any surface modifications that could impact the relevant moments of inertia.

I should probably clarify: the solid shape doesn't have angles of less than a degree. But the surfaces that make up the solid do (as they were built with Alias).

 

[gwolf2]

It seems like contact and motion are your major needs. For this use MARC or ABAQUS which are the best at contact. The problem with poorly constrained multi-body contact is the difficulty in finding equilibrium steps up to the point where 'proper contact' is established.

Not including friction will be hurting you in this case - the friction forces add in small constraints which often help the analysis.

If you have lots of relative movement and lots of contacting bodies consider an explicit code like ABAQUS/Explicit of some flavour of DYNA.

> Is it also inherent that assembly FEA's are orders of magnitude more finicky than part FEA's? (not just in how long it takes to run, but also whether it can run or not).

Yes. Start simple and build up to your real model. If you are new to it, expect days to start getting the hang of it on complex models.

Gwolf

 

[antran7]

thanks for the leads. I'll check out ABAQUS and MARC.

I only care about the final static condition of the static load, and the load forces will be alot higher than the friction forces. So I figure I'll simplify the problem and ignore friction.

Cheers!