Question on approach to solving Random Vibration

[heavenarmy]

I am working on a random vibration FEA analysis of an electronic housing in Mechanica. I first ran a modal analysis to extract the modes (say first 40 modes), then ran a random vibration analysis with the input profile from 20HZ to 2000Hz, levels at 0.04 g2/Hz (typical workmanship input profile). I am confused, and I hope someone can help me out here, as to how the mass participation comes in here. To determine the output response of the system, do I have to include all the modes such that the % mass participation hits 80% or a certain threshold? Or do I simply include in the analysis the modes within the input profile frequency range i.e. less than 2000Hz? The results if I include all modes such that the % participation is >80 shows peaks in the response way above 2000Hz, which I thought is counter intuitive. Do I make any sense?

 

[Co]

What do you mean by % mass participation? Shouldn't the entire mass be included?

 

[heavenarmy]

I think the approach is to include as much modes as possible to best characterise the system response, but in practice, once the sum of the participation hits above 80%, it is good enough.

http://www.eng-tips.com/viewthread.cfm?qid=126162&page=7

 

[Co]

Ah yea. Thanks for the link. I did a bit of reading, it seems that we can define this as the amount of mass that participates with one particular eigenmode. In other words whether the mode is local or global. In most cases I think we would be more interested with the global modes and frequencies, so the closer to 100% the better.

cheers,

 

[sidbrazil]

Have you tried direct analysis, if there are small amount of elements the results is better.