Comparing Modal analysis FEA vs Experimental / High modal density problem

[Domainpower]

I have a bladed disk with 21 blades. By FEA i have obtained the First family of in plane bending modes. See the attached file(FEA_results).

http://files.engineering.com/getfil...8b-4bfa-bb74-05a88111a23e&file=FEA_result.png

After i did an experimental modal analysis using an impact hammer.

I am attaching the measured FRF for condition (impact on the blade 1 and response on the blade 1).

http://files.engineering.com/getfil...-194527517817&file=Measured_FRF_Amplitude.png

http://files.engineering.com/getfil...8769-56182ab0eb9e&file=Measured_FRF_phase.png

here i am getting a single peak may be due to high modal density Or does this response includes all nodal diameter modes of first family?How to extract from it all nodal diameter modes.
Can somebody guide me how to segregate the nodal diameters and identify the mode shape from measured FRF for such a particular case.

Thanks.

 

[GregLocock]

I'm afraid your first plot is meaningless to me and the experimental results are of insufficient resolution to tell what is going on.

Cheers

Greg Locock


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

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

 

[Domainpower]

Hi Greg first thanks for the interest.Experimental modal analysis is new area for me. So may be i was not able to provide you with adequate information.

By first plot i wanted to show that by FEA modal analysis i get a set of natural frequencies whose values are very close to each other (Case of high modal density).

i did measurement using two sensors which were attached to blade extremity , one on each extreme corner and the impact was made on all the blades one by one from 1 to 21.

This time I have attached the amplitude of FRF (acceleration / Force) graph with zoom out for drive point measurement ( sensor on blade and impact on same blade) :

http://files.engineering.com/getfil...8d-483d-9076-273666a3e9f2&file=Amplitude.png.

If something specific is required please tell me.

However what i require is :

[li]general technique on how to proceed if the theoretical analysis shows i have a number of natural frequencies very close to each other and experimental analysis FRF shows just one peak in same frequency range. How to correlate then with FEA values.[/li]


[li]I have another doubt for mode identification.Since i am giving impact with hammer as i am moving from blade 1 to 21 , each time the shape of the input force is changing , is it possible then as the frequencies are very close , one of the impacts by hammer may excite a different frequency and then when i connect the peaks of the imaginary part to identify the mode shape ,i am eventually working with different mode shapes? [/li]

Thanks again

 

[GregLocock]

That tells me very little more. I don't understand your description of the plot but I'll take it at face value and say the data quality looks pretty poor.

Could you show me the coherence, phase, and magnitude plot for (a) the driving point and (b) another point, across a frequency range of 500-700 Hz, if for some bizarre reason you don't want to show the whole thing. Also post a nyquist plot of the frequency range 600-700 hz for the same points.

As to your specific (good) questions

1) yes that is very common, the reality is that a system with many identical structures in it will have clusters of modes where say 10 blades move one way and 11 the other, but exactly which blade does what is determined by non linearities. I'd have to see the above plots before suggesting a strategy.

2)yes that is a typical occurrence. Again the physical system is 'selecting' which mode gets excited by non linearities. The trick there is to use non contacting forcing and response measurement, typically, ie shake the thing as a whole and measure the response with eg doppler lasers or cameras, depending on amplitude. If you can't do that then do your survey with all 21 blades instrumented at once and don't move the excitation point.





Cheers

Greg Locock


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

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

 

[FRANKOK]

I doubt you have actual diametral modes of concern unless all blades are exactly the same as the FEA analysis assumes.
Mistuning - one of many references is:
Castanier, M. P., Ceccio S. L., Epureanu, B, I., Pierre, C, (2007) “Next-Generation Modeling, Analysis, and Testing of the Vibration of Mistuned Bladed Disks,” AFOSR Grant FA9550-04-1-0099, The University of Michigan, Ann Arbor, MI, USA.

You could apply clay to all blades except one and get a quick rap test; then repeat to obtain variation of all 21 blades.