Random Vibration - Relative Displacements

[rwoelfel]

I need to get random psd and rms displacements relative to the base motion of my model. The MSC/Nastran Dynamics Users Manual suggests running the random analysis at some frequency above the zero hertz mode associated with the enforced motion of the large mass. Others have used mpc equations to get relative displacements. Does anyone have experience with either technique. I would prefer the first method, if it works, since all my displacements need to be relative rather than absolute.
Thanks

 

[tomirvine]

I have made relative displacement calculations for a base mass excitation.

Here is an approach:

1. Perform a "frequency response fuction" analysis over the frequency domain from 0 to 2000 Hz, or even higher as needed. Use a constant frequency increment for the response.

2. Output the displacement FRFs at the base and at a few carefully select response nodes.

3. Manipulate the FRFs to form a transmissibility function relating the relative displacement PSD to the base input PSD. Typically, the base input is an acceleration PSD.

4. Multiply the base input PSD by the appropriate transmissibility function to obtain the relative displacement PSD at the selected node.

5. Integrate the relative displacement PSD and then take the square root to get the relative displacement RMS.

I have performed this analysis using NE/Nastran.

This method requires some postprocessing steps. There are several advantages, however. One is that the transmissibility functions yield additional insight into the behavior of the system. Another is that the transmissibilty functions can be used very efficiently if there are several different PSD inputs.

Tom Irvine

http://www.vibrationdata.com

 

[rwoelfel]

Tom,
Thanks for your reply. I'm also using NE/Nastran and use the large mass method of random response which includes a unit frequency response calculation which later gets used to do the random. I was hoping that the "simple" method of not integrating the at or near zero hertz modes associated with the base would output a relative rather than an absolute displacement. I have a lot of output points so I don't have time to manipulate transfer functions. Any ideas?
Thanks

PS I subscribe to your vibrationdata.com web site - nice work.

 

[tomirvine]

Go to:

http://www.vibrationdata.com/StructuralFE.htm

Download and run

ne_postprocess_all.exe

The resulting file to convert an acceleration input PSD to a response relative displacement PSD will have the response node number embedded in the filename _rel_disp_HH.

Then multiply this transmissibilty function by the input PSD for each response node.

Batch routines can be setup if there are numerous response nodes.

The program ne_postprocess_all should run for up to 100 response nodes.

Tom Irvine

http://www.vibrationdata.com