Although a buckling analysis requires no more input data than a standard static analysis, there are a number of items to be aware of when preparing a model for a buckling analysis.
Restraining the structure for buckling
It is important that you restrain the appropriate degrees of freedom to prevent buckling modes that can’t occur in the real structure. For example, if a plane frame is braced in the out-of-plane direction, you must ensure that the braced nodes are restrained in that direction, otherwise the buckling load factor may apply to an unexpected out-of-plane buckling mode. The fix/release node degree of freedom tool is usually the most convenient way to achieve this. For example, using this tool to fix nodes in translation along the Z axis, will prevent all out-of-plane translations in a plane frame in the XY plane.
Conversely, it is also important that you don’t prevent node movements that can occur in the real structure. For example, consider a plane frame rafter that is restrained in the out-of-plane direction at the two ends via an RRFRRR general restraint, but which is able to buckle in the out-of-plane direction between the ends. If you subsequently add some intermediate nodes to the rafter, they will also get the general restraint and this will prevent them from translating out-of-plane, changing the out-of-plane buckling characteristics of the rafter. To avoid this, you could apply restraints of RRRRRR to the intermediate nodes so that they don’t get the general restraint.
Note that a static analysis of a plane frame is not as sensitive to out-of-plane restraints as a buckling analysis because static analysis out-of-plane displacements generally only occur if out-of-plane loads are applied. This is not true of a buckling analysis which can cause buckling in any direction, even if there are no loads in that direction.
Buckling analysis with secondary members
Structures are often modelled with the secondary members such as ties or bracing removed. If these members are required to prevent buckling of the major members in the real structure then they should be included in the buckling analysis model, otherwise the buckling capacity of the structure will be underestimated by the analysis.
This is particularly true of tower structures that contain large numbers of slender members that prevent buckling of the major support members.
Buckling analysis with tension-only or compression-only members
Extra care must be taken with buckling analysis of structures that contain tension-only or compression-only members.
For example, consider a portal frame building modelled in 3D with tension-only wall bracing members that prevent the building from swaying longitudinally. Special treatment is required for the load cases that contain predominantly gravity loads which would cause all the wall braces to go into compression and therefore become disabled. In such load cases, the buckling analysis would yield very low buckling load factors because the wall bracing members would have been disabled and a longitudinal sway buckling mode at very low load would result. Of course, in the real structure this could not happen because the wall brace members would prevent it as soon as the sway mode was initiated.
One solution is to introduce a very small horizontal load into these load cases which is small enough to have a negligible effect on the static analysis results but large enough to cause the wall brace members to go into tension. The result is that they are not removed from the buckling analysis model and are therefore able to prevent the unrealistic longitudinal sway buckling mode.
Similar situations can occur in any structures that contain tension-only or compression-only members.
Buckling analysis with cable members
Extra care is needed for structures containing cable members because of their highly non-linear nature. Because the axial force distribution in cable structures can change dramatically as the load factor is increased beyond the working load, it is recommended that the "Signcount Eigenvalue" theory be used and that the buckling analysis be performed on combination load cases that factor the working loads close to the buckling load, resulting in buckling load factors that are close to 1.0.
For example, if the buckling analysis of a working load case for a cable structure yields a primary buckling load factor of 5.2, you could create a combination load case which factors up the working loads for the particular load case by 5.0 say, and then re-do the buckling analysis for the combination load case instead. If the subsequent buckling load factor is 0.90 say, then the final load factor (for the working load case) is 5.0 x 0.90 = 4.50.
Be careful if the model contains cables and the effective lengths for a steel design are being obtained from a buckling analysis, as the effective lengths may not be correct. You would have to check that the effective lengths for the design load cases are similar to the effective lengths from the factored version of those load cases that produce buckling load factors close to 1.0.
The "Classic Eigenvalue" theory does not give accurate results when cables are present in the model.
Buckling analysis with plate/shell elements
If your model contains plates/shells and their buckling contribution is important then you should use the "Classic Eigensolver" theory, otherwise their buckling will not be considered.
Buckling instabilities
Occasionally, you may find that a requested buckling mode can't be calculated and "Unstable" appears in the buckling output report. This occurs when a node floats free due to local buckling of all of the members to which the node is connected. Sometimes it is possible to avoid this problem and calculate higher order buckling modes by adding intermediate nodes to the members that have buckled.
Modelling multiple structures in one job
It is sometimes useful to model more than one structure in a single job, however this is not recommended if you are performing a buckling analysis to obtain compression effective lengths. The buckling analysis finds the lowest buckling load factor for the entire model and then calculates the effective lengths for all the members in the model based on that buckling load factor. For example, if you have modelled structure A and structure B in one job, and structure A has the lowest buckling load factor, the effective lengths for structure B will be incorrectly based on the buckling load factor from structure A. SPACE GASS can't detect if there are multiple structures in a single model and therefore you need to put them into separate jobs if you want to use effective lengths from a buckling analysis.
Buckling analysis of spectral load cases
Spectral load cases are not included in a buckling analysis. Furthermore, if you perform a buckling analysis on a combination load case that contains spectral load cases, only the non-spectral load cases in that combination will be considered.