Coastdown measurement to determine natural frequency of structure

[bootlegend]

Hello all. I am looking for some tips on determining the natural frequency of a structure supporting a vibrating screen during coastdown. My understanding is that I can measure the machine speed over the duration while simultaneously measuring the displacements of the structure. I currently have a Dewesoft 4 channel data collector and a tri-axis accelerometer. I am planning on using the tach on the 4th channel of the data collector. I'm new to this type of measurement so I may not be providing enough info or even asking the right questions. Is there a better way to do what I'm trying to accomplish? what tools would be easiest to do a coastdown analysis in your experience?

 

[271828]

I am not sure if that will work. It might, but I haven't tried it. It seems like all of the spectra of accelerations on the structure will have huge peaks at the machine frequency and maybe nowhere else.

Typically, we would need to input a force with frequencies across the band of interest. I'm not sure how to do this with the machine running.

The Cadillac method is a shaker with a sine wave that changes frequency as the test goes on. These go by the names "chirp," "swept sinusoid," etc. You can also run the shaker with white noise with frequencies spanning the bandwidth of interest.

A common method is to hit the structure with an impulse hammer or unmeasured impulsive force. That one strike will include frequencies across a band, which depends on the hardness of the contact interface.

 

[GregLocock]

That sounds fine. We call them operating deflecting shapes. @271828, it's a very common technique in automotive, the excitation is equivalent to a swept sine or chirp of unknown amplitude. Typically I'd analyse it using some sort of peak hold, defined by the logic that when the excitation signal is at frequency F, grab the phase and magnitude at F for each response channel.

Cheers

Greg Locock


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

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

 

[bootlegend]

What if the structure is too large to be excited by the hammer? I'm interested in the overall structure natural frequency in addition to any individual member vibrations.

I think the Cadillac method might be out of reach. I'm shooting for the KIA method.

 

[271828]

They make sledge hammer sized impulse hammers. I've always wanted one of those. Ha. They can be used to excite floors in buildings.

 

[GregLocock]

They also make hydraulic shakers, which can bend cars (how do I know that?). One technique for applying large forces is to tension a cable and then snap it. A rule of thumb in general is that if you can't feel the vibration then you are going to have trouble measuring the response.

Cheers

Greg Locock


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

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

 

[bootlegend]

I do have access to one of the large sledge hammer sized impulse hammers, but like Greg mentioned, I doubt it would be enough to excite the structure.

Regardless, I am interested in learning how to perform/interpret impulse hammer testing so if you have any go to references or resources I'll definitely track them down.

 

[271828]

Look for articles on Experimental Modal Analysis.

Some places to start:

https://www.uml.edu/docs/s-v-jan2001_modal_analysis_tcm18-189939.pdf

https://www.youtube.com/watch?v=SHOYDzgMVAQ

I bet the sledge hammer would work. These are used for full-scale building floors. You don't have to knock the crap out of it. LOL

 

[BrianE22]

If your machine is any thing like our deburring tank then the natural frequency is quite noticeable when coasting down. The time history will show a large amplitude at some point when coasting down. I think your coast method will work fine. To back it up maybe use a stop watch to monitor how long it takes to get to that bad point. Use that time in looking at your time histories.

 

[GregLocock]

Bruel and Kjaer have a nice series of downloadable booklets and books on noise and vibration. The analysis of an impact signal is straightforward, you gate the signal to eliminate noise in the tail, and just do FFTs as usual. Do not use Hanning (seen it done!).

Cheers

Greg Locock


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

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

 

[bootlegend]

Thank you for the references. Those all look very helpful. Going to spend some time with them.