regarding damping

[blacko]

hi guys,

I am new to the field of vibrations, basically an intern.

I have been given this project where i am supposed to find damping of a part on the entire fixture under different cofigurations and loads. I have the frequency domain data.

the problem is that since, the part is constrained, the FRF curves are not very good for certain modes. The softwares I am using, uses polyrefernce time domain method to calculate the parameters. but some modes are not stable and the damping values vary a lot with the model order. and i donot know which one to pick.

now the part is circular in shape so it has repeated roots

so i thought of filtering the time domain data into different frequecny bands and calculating the damping using log decrement method.

questions:

1. is it possible to convert frequency domain data into time domain? I used ideas but the results are strange. if yes, how?
2. how can i use the curve fit on the time domain data. i donot want to eye ball it. is there any program or software?
3. is there any other conveninet way to find damping?

Again I am a beginner and these topics have already been covered in the previuos posts but if someone could try to expalain me in simple words, it will be great.

thanks a lot in advance

 

[dolsoi]

I am not sure if I undersood your question correctly, but a simple way can be that you measure a point mobility (FRF) of the subject of interest, and then take the real part of its inverse. The inverse is the impedance and its real part is the damping(i.e., power absorbing part). Finally, you draw this real part data against frequency. Cheers, dolsoi

 

[MikeyP]

1. How many references are you using (you say there are degenerate modes)?

2. Have you tried playing with the number of samples used in the Polyreference analysis? This can have quite an influence on the clarity of your stabilisation diagram.

If you can't get good results with Polyreference then you probably won't get good results with logdec.

M

--
Dr Michael F Platten

 

[xydrex]

Is there any chance of you being able to test the part on its own & suspend it in free-free conditions? That way you can get a true FRF.

 

[blacko]

Thank you guys for your responses..

well I have been able to test the part in free free condition and have the data. well the aim of the project is to find how damping changes in a constrianed condition and under different loading conditions. with the introduction of these 2, natural frequencies change, shape of the FRFs change. and the modes donot stabalize due to which when some program uses PTD, the modes donot stabalise and there is a lot of variation in the damping values on moving from one model order to next.

I am using MTS reporter and it doesnot ask for no of samples but you can play with the model order.
I am using just one reference at the moment and hence maybe not able to get good results. due to this reason i am not focusing on the frequency having repeated roots. I will be using multiple refences in a bit but before that I am trying to use time domain data to see if i can get damping values from it using log decrement.

One interesting thing i noticed is that I have a sample in which I recorded data in time domain 5 times. and it was interesting to see that the damping values vary from one data set to other. This should not be the case since the I did nothing with the system betweeen 2 measurements. i think filtering the data is causing this error. but how and why, i have no idea. or there are not enough data points to get accurate values.

any ideas what should i do??

need those damping values. can we get them with accuracy or not???

 

[GregLocock]

If your damping rates are varying from run to run, then no, you cannot really expect much physical meaning from an 'accurate' value.

I strongly suggest that you find out why the damping is changing. Your customer will appreciate that effort.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MikeyP]

If you are using polyreference with a single reference then you are really just doing a plain old LSCE curve fit. You cannot hope to identify repeated roots with just a single reference.

Does your structure conform to the proportional damping assumption? Are there discrete damping sinks involved?

Can you post some FRF data somewhere?



--
Dr Michael F Platten