HOT!!! Results of Random Vibration Analyses...

[koenigsegg]

Hello all,

Sorry for the long post but it is worth it.

I am currently analyzing the response of a beam shaped solid (clamped at one end) to a base (displacement) random excitation. I am interested in knowing the 1-sigma displacements (RMS values) of some nodes in the same direction of the applied excitation.

Now, I know that there are 2 ways of getting this values:
1 - list/plot the displacement nodal solution;
2 - plot the response PSD curve of those particular nodes, integrate over the frequency range and take the square root of the result.

Either way, the results should be same.

What I have found is, if the input PSD curve is a "white noise" (constant values over the frequency range), this holds true.

HOWEVER, if the input PSD values vary with frequency, the values do not match! And I am not talking of disparities in decimal places; depending on the shape of the input PSD curve, I am talking of differences in one order of magnitude and more...
WORSE...: if I look at the response over the clamped nodes, where it should be equal to the input, I found that the input RMS and the response RMS values do not match either!

Now the question is: what is happening? Which results can I trust? Does any one have any idea?

Thanks in advance for your feedback.

Koenigsegg

PS1: I have also verified this behavior with a simple SDOF spring-mass model.
PS2: All of Ansys Verification Manuals test cases are performed with white noise PSD's as inputs...

 

[Transient1]

A couple of comments:

Have you reviewed the Theory Guide for PSD excitations? Also, can you post the input files for both cases? You might not be plotting what you think you are plotting.

Are there very closely spaced modes in the frequency range you are working with? Sometimes things can get screwy with overlapping modes.