Buckling analysis 2

[saeedHR]

Hey friends
I want to do a buckling analysis of a simply supported beam. I have defined the beam section as shell elements and the type of buckling analysis is Lanczos. How mush should be the number of Eigenvalues requested and after analysis how can I work out the buckling moments

 

[ESPcomposites]

It depends on how well you defined your boundary conditions. The first six modes may be rigid body modes and the 7th eigenvalue could be the real one. However, it could be the first mode or a much later mode than the 7th, again depending on your boundary conditions. You will want to look at the deformed shapes to determine what the first real mode is.

The buckling load is the applied load x eigenvector. So for example if you applied a 100lb load and the actual eigenvector is 0.8, then the buckling load would be 80lb. Same would go if a moment was applied.

Brian

http://www.espcomposites.com

 

[saeedHR]

So is the number of eigenvalues actually the number of modes?

 

[ESPcomposites]

Each eigenvalue has and associated eigenvector (mode)...so in that sense, yes. However, some of those (usually the lower ones) may not be "real" modes (ones that indicate buckling). You will need to look at the displacement of each mode to determine if they are real or not.

Brian

http://www.espcomposites.com

 

[BrianInglis]

On the subject of the displacement of each mode, would i be correct in assuming that this is a visual check?

 

[ESPcomposites]

Generally speaking, visual is the way to identify the mode.

However, it may be possible to create an algorithm to identify real modes as I have done with this panel stability program.

http://www.espcomposites.com/software/eStability.html

So if you want to automate something, you would anticipate what a real shape would do versus a "false mode". By evaluating the displacements at different nodes, you should be able to identify if the mode is due to rigid body translations/rotation or a true deformation. However, this would only work if you knew what to expect first.

Brian

http://www.espcomposites.com

 

[BrianInglis]

Thanks again for the help.

I assumed that was the case. I had a vertical load which in some modes was causing displacement in some crazy directions so i assumed these were false.

Still trying to get my head round Abaqus so i think i'll leave the other program. One step at a time i think.

Cheers Brian

 

[verians]

Hi guys,

I am new in the forum, and in FEA work! I am carrying out a buckling analysis with abaqus. When I open the .odb file I can see the buckling modes and their respective eigenvalues. I can not understand the meaning of the displacements legend. When the displacement is less than one, is that a local buckling? should you take it in account?I read that the real eigevalues (buckling factor load) are which have an associated mode with displacement 1.

What is the meaning of this please??

Thank you in advance.

 

[saeedHR]

Excuse me Brian, what do you mean by actual eigenvector? How can I work it out?

 

[lec235]

verians,

It took a bit of digging in the literature to find this when I had the same question, but the magnitude of the node displacements associated with a particular eigenvalue is meaningless.

Basically, in *only* the case of a buckling analysis, ABAQUS assigns the node(s) with the highest displacement a value of 1 and every other nodal displacement is scaled accordingly. Hence the node displacements are only really good for determining mode shapes.

Hope this helps!