Eigenvalues analysis

[bassking]

Hi guys, i have a little problem with an eigenvalues analysis. I'm modelling a simple deck bridge, without the piers, only to obtain the periods of the superstructure. The deck is made with 4 span of continuos box-girder, 50 meters each. So i define 5 nodes, first and last are fix, over the second, third and fourth i assign lumped mass in 3 direction (X,Y,Z). I must obtain 9 periods (3 mass for 3 degrees of freedom), but seismostruct give me only 7 (but the last is "immaginary or null") eigen and "Warning: Could not re-orthogonalise all Lanczos vectors" error. Model is very simple, i don't understand where is the problem.

[huffte]

Hi bassking.

I'm not sure what is going on with your model. I threw together a quick model similar to the one you describe and I had no trouble getting the modes using either Lanczos or Ritz solution for the eigenvalues.

Check your settings - if you are using Lanczos, then ask for more modes than you really want. If you are using Ritz Vectors, then ask for 5 vectors in each of the 3 directions - this will be more than enough so some of them will, in fact, be imaginary.

Tools -> Project Settings -> Eigenvalue

I hope this helps you zero in on the problem bassking.

[Stelios_Antoniou]

With smaller models the Ritz+Jacobi method is considerably more stable.