PSD response with/without 1 normal mode activated?

[JBlack68]

To all,

A question about the modal approach of solving response for a random analysis
With my test FE Model I de-selected 1 mode and compared the PSD response (acceleration) to a prediction will all the modes. The mode selected has a mass participation of 2%

The PSD response with the mode de-activated shows a higher response in the low frequency range? At the frequency of the mode de-activated the response is lower which is expected.

Any idea what I get this “behaviour”


Thanks

Regards

 

[GregLocock]

Do it analytically with a 2dof system. Is the result different? I haven't done this, but I'm suggesting obvious ways of debugging the behavior.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[hacksaw]

TX,

not suprising with complex systems and broadband excitation,

some modes act to stiffen the response, while others may not.

 

[JBlack68]

Thanks for the replies & suggestion. I have not checked with a 2dof system. I did the test as I wanted to report the effect (on the rms response) of the mode de-activated. With all the modes selected I get X, with all but 1 mode I get Y. Y/X shows that the mode de-activated represents p% of X. (if this make sense!). For my test I concluded that the mode de-activated represent ~25% of the total rms i.e. without this mode the rms response is 25% lower.

In my mind if one de-activates a mode there is less mass vibrating so at the low frequency - say between 5-50Hz (where I have no modes at all) I expected the response to be lower - not higher

 

[hacksaw]

you can only deactivate a particular mode by structual changes! but you don't really deactivate it, you only change the modal response and perhaps the coupling with the force distribution.

if you are simply ignoring a given mode by asserting that it's excitation is nul, the total response drops.

you can only convince yourself by intensive, really intensive case study analysis...that can take more than a few minutes...

good luck, it will be a learning experience

 

[JBlack68]

by de-activating I meant not including it in the random response. When you use a modal technic for a random analysis one can select which mode to include. If one calculates N modes then "by default" all N modes are included. With some software (all of them!) one can request for some modes not to be included in the modal response even tough they been calculated (using Lanczos for example)

 

[hacksaw]

you have to prove that to yourself at a minimum.

eliminating a mode is tantamount to altering your structure...

simply put, the energy distributed over your defined system is now confined to fewer modes but with greater amplitudes,