weird result in eigenvalue analysis

[zishendemi]

Hi all,

I'm having some weird result in eigenvalue analysis of a 3-D 2-story shear building (with rigid diaphragm). I use the command eigen -fulGenLapack to get the eigenvalue of the structure.

I'm analyzing the same model.

For the first time, I use eigen -fullGenLapack 6, which gives me the first 6 eigenvalue as follows:
3.23e2 3.83e2 4.96e2 1.97e3 2.63e3 3.85e3

Then I use eigen -fullGenLapack 3 to get the first 3 eigenvalue as follows:
2.90e2 3.87e2 4.95e2

The last time I use eigen -fullGenLapack 1 to get only the first mode. But it returns me
2.40e2

I'm analyzing the same structure and using the same command (just to get different numbers of eigenvalues), but every time it gives me a different result!

Does anyone know why it happens like that?

Thanks in advance!

[fmk]

i have never seen this before. either there is a bug in the code or your model. does a static pushover of the model work?

you might need to post a very small example that demonstrates the problem.

[MDSmith526]

At first I thought you may have misspelled fullGenLapack but I tried it and it gives an error so thats clearly not it.

[alysa]

I have seen first time this information.I am so glad after visiting this post.This is really new for me.Thanks for sharing such a great information here .

[zishendemi]

Hi all,

I think I find part of reason to the problem. The model that is creating the problem is nonlinear and the result of eigenvalue analysis depends on where you put the command.

If you put the command of eigenvalue analysis right after you build the model (without applying any load), the program will analyze the eigenvalue of the structure using its initial stiffness. However, if the command is executed after applying load to the structure (and also yield it), the post yielding stiffness will be used to calculate the eigenvalue. So there will be a difference.

This is only my point of view on this problem. I appreciate hearing your idea on this.

I also have a question. Basically, for a structure vibrating in inelastic range range, the stiffness is changing all the time after it yields. As a result, the vibration frequency of the structure is also changing. In Opensees, we usually assign Rayleigh damping to match certain 2 modes. But in this case, as the frequency is changing, which frequency shall we use in assigning Rayleigh damping? The frequency when the structure is still elastic or something else?

Thanks!

[vesna]

I am glad you figured it out yourself.

There are different ways to model the damping. Some models do not update the Rayleigh coefficients based on change of periods but some do. It is on analyst to decide which damping model to use.

This paper may be useful for you: http://ascelibrary.org/doi/abs/10.1061/ ... 4%28581%29