[MBDyn-users] eigenvalue extraction

masarati at aero.polimi.it masarati at aero.polimi.it
Wed Jun 16 16:03:30 CEST 2010


> LS,
>
> I am attempting to extract the eigenvalues and -vectors of the TM4760
> example model (in rotating conditions). The eigenvalues are partly
> reported in the .out-file and the frequencies are close to the expected
> values. However, the alpha matrix in the .m-file returns purely real
> eigenvalues of magnitude unity. Also, the file does not contain the idx
> vector mentioned in the documentation, so I cannot plot the eigenvectors
> in a meaningful way.
>
> I understand the eigenanalysis functionality is experimental, but the
> results in the .out-file are satisfactory, so it appears to be a
> post-processing issue. Could the problem be related to my build of MBDyn
> or, for instance, LAPACK?
>
> I am using MBDyn version 1.3.15.

Hi.  The results you find in the .m file are "raw".  First of all, the
"idx" vector and other post-processing data related to structural models
are returned only when you add the "output geometry" request to the
eigenanalysis card
<http://www.aero.polimi.it/~masarati/MBDyn-input/mbdyn-input-1.3.13/node26.html#SECTION008111200000000000000>.

Moreover, the eigenvalues in the "alpha" matrix need to be transformed
according to Eq. (30) of [1].  Let me know if you can't access that paper,
I can send you a draft.  I have implemented a script that is supposed to
perform all the required transformations, it's not released yet because it
is not yet general enough.  The suggested approach, right now, is that you
write your own script in order to suit your needs.  I can send you mine to
serve as a guideline.

Hope this helps.  Cheers, p.

[1] P. Masarati, "Direct Eigenanalysis of Constrained System Dynamics",
Proc. IMechE Part K: J. Multi-body Dynamics, 223(4), 335-342.
doi:10.1243/14644193JMBD211 <http://dx.doi.org/10.1243/14644193JMBD211>.



More information about the MBDyn-users mailing list