Using discrete/analyically rigid elements with buckling analysis

[lec235]

So I'm trying to model a column with 4 rigid links for a buckling analysis. The overall structure is simply supported and there is also a linear spring at the middle joint. Each of the rigid links are connected to the one below via a rotational spring. I attached the model for reference.

However, when I try to run it, this error pops up: DIFFERENTIAL STIFFNESS MATRIX IS COMPLETELY NULL. THE EIGENPROBLEM CANNOT BE SOLVED. IN A *BUCKLE ANALYSIS THE MOST LIKELY CAUSE IS THAT A NONZERO LOADING PATTERN WAS NOT SPECIFIED VIA *BOUNDARY, *CLOAD, *DLOAD, ETC,.

What does this error mean/how can I fix it? Thanks!

BTW, this model is with analytical rigid parts for the links but the same error occurs with the discrete rigid parts too.

 

[corus]

First look at the most likely cause quoted, ie. have you specified a load?

Tara

http://tinyurl.com/4ydjg7m

 

[lec235]

Hi,

Thanks for the response but yes I specified a unit vertical load on the top link.

 

[corus]

Looking at the cae file it seems that all the parts are rigid links. An infinitely stiff rigid part won't buckle.

Tara

http://tinyurl.com/4ydjg7m

 

[lec235]

Right. The links aren't supposed to buckle (hence why they're rigid). The deformation in the model is supposed to come from the springs that are connecting the links. So am I getting the error because the stiffness matrix for the links is essentially infinite? But wouldn't the fact that the spring stiffness matrices are not infinite just mean that the deflection will occur in those members, or is ABAQUS just not written to recognize that?

 

[lec235]

I think I've discovered a work around: use deformable parts meshed with a single beam element but specify the material to have young's/shear modulus of diamond.

 

[corus]

I didn't see springs in the parts list so I'm not sure where these elements are. Another problem that may occur is if you have rigid parts and deformable parts where the ratio of stifness between parts is zero, or infinite (depending on how you do the ratio). If it's a column then why not model the column as-is, and use deformable elements.

Tara

http://tinyurl.com/4ydjg7m

 

[lec235]

Its one of a series of models of various complexity of the same column configuration. I'm looking at possible simplifying assumptions and how that relates back to the buckling response of the actual structure.

But I think you're on to something with the stiffness ratio idea. Thanks for the help!