Fourier synthesis from harmonic simulations

[emoulin]

Hi everyone,

I am computing the response of an elastic plate to a transient normal load, using the SMITC solver. Transient simulations work fine, but I would like to compare them to results obtained through Fourier synthesis of harmonic simulations (which allow more physical taking into account of losses and possibly would result in shorter computation time ?).

The problem is that harmonic simulations, using the following keywords in the solver section :

Code: Select all

Harmonic Analysis = True
Frequency(n) = f1 f2 ...

end up with zero imaginary part of the displacement field. Therefore I see no way of inverse-Fourier transform the harmonic displacement to get a transient one...

Am I wrong somewhere ?

Thank you for your help.

[raback]

Hi

You're probably right. The damping matrix should be saved for later properly. I guess that the current implementation does not assume that the harmonic system could be damped.

-Peter

[emoulin]

Hi,

Thank you for your answer. But unless I am wrong in the interpretation, the imaginary part of the displacement in a harmonic simulation is not merely related to attenuation, but also to account for phase variations versus position. The displacement being represented in the equations as a complex exponential of the form u_i*exp(jwt+phi), the results should be complex numbers indeed, even in the non-damped cases...

Emmanuel

[raback]

Hi

Well, you're right that the ansatz is complex but still the eigenmodes and eigenvalues of the system without damping should be real. You can always restore the full solution by multiplication with the harmonic term.

-Peter