Buckling with focus on part of the model

[ThomasH]

Hi all,

I have a "problem" that I am fairly certain can be solved if if I don't know the answer myself .

I am working with a rather large model (1 million+ nodes). It is a bridge and most of it is steel beams modelled with shell elements. One part of the analysis is to determine the buckling modes. Thast means the buckling modes for the primary beams and that means about half of the model.

I have already made an analysis and have the relevant buckling modes, I can pick them out manually. I can also make the solver pick modes in specified ranges to control the results.

BUT, what I would like to du is simply to define a part (group) in the model where the solver should find the buckling modes.

Is that possible? And if it is, how do I set up the analysis?

I am asking more out of curiosity then necessity.

Thank you.

Thomas

 

[SWComposites]

Should be possible using superelements.

 

[ThomasH]

That may be an approach worth investigation. Thanks for the idea.

Thomas

 

[dmapguru]

Hi Thomas,
If we are discussing linear buckling in SOL 105, starting from MSC Nastran V2018, you may use PARAM,EXCLUDE,n to ignore or retain only some of the model GRID points where differential stiffness will be ignored/retained.
Linear buckling in SOL 105 solves the eigenvalue problem (K+L.Kd)U=0, where K is the linear stiffness, Kd the differential stiffness from the displaced shape due to an applied load, L are the eigenvalues and U the buckled shapes. Compressive loads cause negative differential stiffness, and when Kd becomes larger in magnitude than the linear stiffness at some location in the model, the stiffness goes negative and the structure buckles; the reported eigenvalues may be used to factor the applied load to determine the critical buckling loads.
PARAM,EXCLUDE,n may be used to select a zone of GRID points in the model where the differential stiffness will be either jettisoned if n is positive, or the differential stiffness will only be retained in the defined region if n is negative. The value of n is set to select a SET1 entry that defines the zone to be ignored/retained.
This is an engineering approach to the problem, and will result in elements in the region adjacent to the limits of the selected region having some GRID points where differential stiffness is present and some GRID points where it is not, which is non-physical. These “transition” zones should be sufficiently far away from the regions of interest that they do not affect the buckling modes, and ideally they should be defined in relatively stiff areas not prone to buckling along natural mesh lines.
DG

 

[ThomasH]

@dmapguru,

Thank you, I have to read your post more careful but it may be something worth trying.

For the model that I asked about I found that there was a lowest buckling factor and then the remaining factors of interest was in a "cluster" of something like 30-35 modes above that value. So ultimately it was easy to "filter" the results.

Thomas