Random Vibration Analysis Relative Displacements

[JLBK56]

Hi all. I have been having a discussion with a colleague related to relative displacements in a random vibration analysis. Specifically, we use ANSYS for our analyses, and my colleague in the lab was recommending the relative displacement be calculated using the difference of the RMS displacements for two locations of interest. Interesting, this is how ANSYS calculates the reaction force in a spring element -- my case study has shown that it uses the stiffness of the spring times the difference of the RMS displacements at the two remote points. However, I don't think its correct to do this (perhaps my case study worked out to be a special case where this is in fact true, if I am correct in my suspicion that its an incorrect approach). I believe the correct approach is to take the displacement RPSD at each location, determine the relative displacement as a function of frequency, and then take the RMS of the resulting relative displacement RPSD. I haven't found any concrete way to prove this in the literature or in any textbooks -- can anyone provide guidance on which approach is correct, and preferably provide a reference or way to validate this is true? Thanks in advance, I appreciate it.

 

[hacksaw]

Your question is too general to warrant a specific answer to you question.

 

[dmapguru]

I feel your pain. What you describe is a concept some have called "rattle", and there is a lot of waffle on this subject. The RMS approach you describe above is one (simplistic) way of looking at the problem, but there are more elaborate methods. Have a look at the CAEFatigue (Hexagon) documentation, specifically the User Guide Examples document. I googled "hexagon caefatigue documentation" and then located their documentation center, then searched for "CAEFatigue user guide examples" to find the document; chapter 5 has a section entitled "ADDITIONAL RESPONSE INFORMATION USING ABSRESP AND RELRESP" where the RMS and more elaborate methods are detailed.

DG