High frequency noise

[Farid]

I have a bridge model with a linear-elastic composite deck in which the girder and slab are connected by two-node links. When I apply dynamic force on some nodes (representing moving vehicle), I get some very dominant high frequencies (at Nyquist frequency) in the recorded acceleration. However, double differentiation of the displacement is good. Does anyone have same experience? I am using the Newmark integrator (0.5, 0.25).

[selimgunay]

How does the recorded displacements look? You may be exciting some high frequency higher modes depending on the frequency of your forcing function

[Farid]

Hello Selim,

Recorded displacement looks smooth (although has a small level Nyquist rate high frequency). I agree with you that the impact load could have excited all modes (while I believe they should appear as low frequencies noise due to aliasing issue, not at Nyquist rate), but I expect to see them in the double differentiated displacement too. Here is a link to see both cases:

https://www.dropbox.com/s/q5mquob5lw30s ... e.png?dl=0

[selimgunay]

From the plots that you sent, this looks like a numerical problem. I have seen this issue with another integrator before, but not with Implicit Newmark. To overcome this issue, you can try HHT or Generalized Alpha with numerical damping

https://opensees.berkeley.edu/wiki/inde ... lor_Method
https://opensees.berkeley.edu/wiki/inde ... pha_Method

[Farid]

I have already tried other integrators with numerical damping, but no success. I will try again. Thanks.

[Farid]

Could you please tell me which parameters I should play in the Generalized Alpha method?

[selimgunay]

You can play with alphaM and alphaF. You can also decrease the integration time step to see if it helps.

[Farid]

Thanks.

[Prafullamalla]

Have you solved the problem. I am also facing same problem in my model high frequency noise

[Farid]

Record displacement/velocity and numerically differentiate to get the acceleration.