Buckling strength 1

[FEAsolver]

I do have a complex structure modeled using shell element (FEMAP with NX-NASTRAN) under a combination of loads in different directions. it is required to have the stresses be less than 80% of the buckling strength of the structure. The question is how to check the buckling strength of the structure (the whole assembled structure) using Finite element. Actually I applied all the loads and run the analysis for buckling and I have the critical buckling factor value (eigenvalue) but I am not sure if this value is valid for a complex struture under a combinations of loads in different direction. I would greatly appreciate any comments in this regards.

 

[rb1957]

i think the eigen value predicts buckling of the gross section (ie long column, Euler), and as such should be sufficient. maybe you also need to think about short column failures (crippling).

 

[SWComposites]

The eigenvalue is simply a linear scale factor on your applied loads.

 

[FEAsolver]

A common practice for calculating the buckling strength of a part is to apply a unit load and check for Eigenvalue which is the critical buckling factor (the part would buckle at a load of calculated eigenvalue times the unit applied load) but the question is when in a structure there are different loads with different magnitude and direction is still the eigenvalue valid for all loads or it has to be individual loads?

 

[crisb]

You need to do a non linear analysis with initial imperfections, to establish that there is a suitable factor of safety on the applied loads.
An eigenvalue analysis is not appropriate although it may determine a suitable shape for the initial imperfections. It may be necesary to include large displacements, and yield in the material model depending on the application.

 

[mtnengr]

Crusb has a very good point. To get an estimate of the true buckling strength of a structure, a non linear analysis with the eigenvectors of a linear buckling analysis being used as the imperfection shape is appropriate.

Which imperfections to select is dependant upon the type of structure one is analyzing. For example, space frames are often thought as being connected with rigid and fixed connectors. This is a theorectical concept. One might consider a fixity factor less than 0.5. For plate components, the displacement pattern resulting from either symmetric or non symmetric loads should be considered. You may want to also consider discontinuity offsets for welded plates. If the structure is a conic, then there are a number of harmonic displacement patterns one needs to consider.

It requires considerable thought as to the selection of imperfection geometry for nonlinear anaysis to be applied correctly. Sometimes adjacent mode patterns can couple and produce lower buckling strengths than those to be considered independently.

 

[SWComposites]

FEAsolver - the eigenvalue is a linear factor that is applied to the entire set of loads that you have applied to the structure.

Whether or not you need a non-linear analysis with imperfections depends a lot on the type of structure, the type of loading and how agressive you are with your buckling criteria. If you set a large safety factor on your predicted buckling load using an eigenvalue analysis, then that analysis may be sufficient. A 20% margin on buckling is not very much for most structures, and therefore you probably need to either run a nonlinear analysis or test the structure.

 

[corus]

Some design codes (such as the DIN structural codes) actually give a factor on the buckling load calculated from FEA to take into account initial imperfections and other factors. They also give guidance on the application of loads too. Even if you're not designing to that particular code it's better to give a useful reference for your choice of factors or loads rather than just guessing at some factor or imperfection.
If you're certain that all the loads apply at the same time then use them, otherwise look for the worst case scenario to remove any doubt.

corus

 

[271828]

crisb wrote: "You need to do a non linear analysis with initial imperfections, to establish that there is a suitable factor of safety on the applied loads.
An eigenvalue analysis is not appropriate although it may determine a suitable shape for the initial imperfections. It may be necesary to include large displacements, and yield in the material model depending on the application."

I agree with this. Garbage in = garbage out