FFT plot from accelerometer of beam under impact load

[StathPol]

Hi guys, 

A bullet (5kg) drops in the middle of concrete beam and the graph below shows the acceleration of the dropped bullet and mid-span acceleration of the beam. I have plotted the FFT diagram as well. At first, the bullet hits the concrete but then the beam reacts back so we have double contact between the bullet and the beam ( the beam is in its linear elastic range). However, the FFT of the beam looks a bit strange. Is it possible to let me have your thoughts?

[snordinh]

Hi StathPol,

Thank you for posting this question. I would like to check up on this post since no one in the community has gotten back to you yet. 

Do you suspect that your LabVIEW code is not functioning correctly? If so, please explain what kind of FFT you were expecting and I can try to advise you on how to make it perform in the way you would like. 

If you are having issues with the mechanics part of this question, I have found some resources that could be useful for you:

• Understanding FFTs and Windowing
• Interpreting the FFT results of a structural vibration
• Beam vibration by a shaker
• Vibrations of Cantilever Beams: Deflection, Frequency, and Research Uses

Best regards,

Sara Nordin Hällgren

Applications Engineer

National Instruments 

[Henrik_Volkers]

I see nothing unusual..

You skipped the elastic wave high frequency contend... maybe not OK for the beam....

The beam has a lot of modes you excite with the ball drop. (depending on the hit point) ..

Depending on the beam shape/contruction and by the beam FFT I would say ~600Hz and consequently and multiples of this freq. . The douple impact (not exacly in phase) will add some more contend... 

If you have a (more or less simple) FEM model of your beam, try a simulation with a force (F=m_ball*a_ball) constrain...

[Kevin_Price]

Another *very* general observation to keep in mind:

An essential theoretical aspect of the FFT algorithm is that it's based on the idea that the chunk of time data would repeat infinitely.  The FFT isn't ideal for characterizing transients -- it'll find some of the real frequencies, but it will also add artifacts as it "imagines" the need to account for that chunk of time data to go on repeating forever.

-Kevin P