Decouple the strain gauge measurement into individual modal response

[Domainpower]

Hello Experts ,

I have a structure with multiple strain gauges attached to it. It is under fluctuating load.
We know the response at any point is basically linear combination of scaled mode shapes.
Is it possible to accurately decouple the total strain response from each gauge into modal coordinates ? My objective is to identify the contributions of the dominating modes , so even if i can extract modal response of those modes it will be interesting for me.

Which technique can help me to perform this decomposition of strain gauge measurement?

Can you recommend any textbook reference or article which help me understand the steps involved to work .....

Regards
RK

 

[scheah]

Hi RK,

even if i can extract modal response of those modes it will be interesting for me.

If you're after the mode shapes from test data, the best way is to perform a modal test to extract the modes. OMA works but has some limitations.

Which technique can help me to perform this decomposition of strain gauge measurement?

I've done it in the past by associating test data with strain mode shapes from FEA. Here's a paper that might be useful.

Note that the measurement locations should be able to distinguish between the modes. A MAC should help verify this. Modal coordinates can be computed at every instance in time of the test data.


Best regards,
Jason

 

[Domainpower]

Thanks Jason for paper and tips , I will have a closer look at them.

Just few more points for clarity
I actually don't want to obtain mode shapes what i want to do is to decompose strain measurements into modal response.

I was able to find a Paper which exactly is dealing with the kind of problem i have (attached).He uses Empirical mode decomposition (EMD) to decouple the strain measurement.
Is EMD a robust approach? Please let me know if any body has any alternate / robust way to do the same.

 

[scheah]

Is EMD a robust approach?

From what I gather, the EMD method is very similar to what was previously discussed. You need mode shapes on hand to decompose the strain measurements.


Best regards,
Jason

 

[GregLocock]

There is no difference in information between 'knowing' the mode shapes and understanding their contributions. That's the same thing to all intents and purposes. Actually you need slightly more information to work out the contributions, as you need to assume a damping and frequency and an amplitude scaling (in whatever form) for the mode shape, which the mode shape doesn't actually care about.

Given that you have the structure strain gaged the quickest fastest simplest approach would be to hit it with a hammer and load cell of appropriate size and do an experimental modal analysis. This will give the mode shapes and the global parameters for each mode in a far more robust fashion than operating deflection shape data.

Cheers

Greg Locock


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

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

 

[Domainpower]

Hello Greg,
First thanks for stepping in.

From paper attached in my previous reply i understood, limited strain gauge measurements under actual operating load is all that is required to decompose and evaluate modal responses as function of time (refer page 8 in attachment).This is exactly the information i want to work further.
I have the FE model of structure and did the modal analysis to get the mode shapes and natural frequencies from it.
Exactly like in the paper i too want to capture peak to peak at location which is away from strain gauge.

Now as you propose to do Experimental modal analysis(EMA).I am not clear with :
- why in the paper EMA is not required.
- for my understanding with EMA (experimentally derived mode shapes, natural frequencies and natural frequencies) how can i decouple my strain guage measurements.