NX3 / NX Nastran 1

[PG541]

Never used a FEA solver except for a PE solver just to check a design. Anyway, wanted to build a mountain bike frame and started playing around. I'm using NX3 structures to set up the constraints, materials and mesh. When I attempt to mesh the frame using a tetra 4 or 10 I get to 69% and NX3 goes into a 'not responding' mode. I've tried making all the frame members solid and also a constant wall thickness but get the same results. It will however let me mesh each individual frame member separate, I would think you should be able to mate the meshes together but I can't find anything in the help files.
I can provide further information if needed as to what I'm doing if anyone would like to help. Just thought it would be a fun activity for a hobby.

 

[FEMdude]

There may by too small surfaces or edges hidden in the model; try feature suppression, or select every meshable surfaces one by one, not using "pick box" or "select all" button.

 

[PG541]

Thanks for the reply but forgive me for my ignorance. Essentially what i would like to do is fix the head tube (tube that has the handle bars attached) and the two drops (what the rear axle attaches to) and applying a load downward on the bottom bracket (tube that has the crank arms) to see the flex of the bike. I am able to select each component of the frame individually and mesh it but shouldn't the meshes need mating? I was able to solve this time, however, I would expect to see stresses in the down tube and chain stay tubes but the only thing shown in the results is on the bottom bracket (only surface the load is applied). By pushing downward on the bottom bracket and only one face I'm creating a moment on the bottom bracket and would think the seat tube would bow.
Can you or do you have to mate the meshes if you do each frame member separately or does the program recognize it as a whole?
I'd appreciate all the help possible. A work colleague and I are taking our hobby a bit far but it's become fun now and this is a bit interesting.
Maybe I should just stick with design and release and leave the validation to someone else.

 

[GregLocock]

If you mesh them as separate parts then they won't be joined together. A crude way to do it would be to merge coincident nodes, or use rigids to link the meshes at each interface.

Either way you will get a horrible result, but it will at least point you in the right direction.

The semi-proper way of doing it is to either merge the solids before you mesh them (so that they are one volume), or to mesh each part individually and consecutively, but forcing the program to pick up the previous meshes at the contacting surfaces as necessary. That at least has some merit.

This assumes that it is correct to treat the entire structure as a homogenous lump of aluminium or whatever. That would be an error.

Really you need to model the welds or brazes. That gives you the same problem, repeated many times over.

I'm glad you realised that that plot of stress was indicative of major errors. You may find that a deflection plot gives you a better feel for what is going wrong.


Cheers

Greg Locock

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

 

[PG541]

Thanks for the tips, I have all the filet welds in the model but they're blanked because I'm just trying to produced some decent results first and then I'll bring in the welds. I took out the drops (components that holds the rear axle)and am working strictly with the frame members.
If anyone can give some pointers on set up using NX3 or NX4, maybe a typical set up procedure I'd appreciate it.