rolling elements bearings defect frequency and harmonics 1

[kancil]

i have conducected a few experiments to detect rolling element bearings fault using envelope analysis. The envelope sepctrums showed that the fundamental bearing defect frequency and its harmonics are in exponentially decaying in their amplitude. Is there anyone know why this phenomenan happen. If possible pls email me leo9421@hotmail.com. Thanks.

 

[dgallagher]

As an example, say a bearing is developing a crack in the outer race. Each time a ball passes over the crack it creates a high burst of vibration energy that last for a short period of time. Enveloping traces an envelope around these repiticious busts of energy and passes this envelope through the FFT process, providing you with the distinct bearing frequencies and harmonics. As the defect progresses, becomes larger and smoothes out, the impacts become softer, lower busts of energy. As a general rule of thumb - if peaks are only found in Enveloped Acceleration, this is an indication that a fault is developing or that the bearing requires lubrication. If the fundamental peaks are present in both Enveloped Acceleration and normal FFT, then it is time to schedule a replacement. If fundamental peaks are present in normal FFT, but now hidden in the Enveloped Acceleration noise floor - replace the bearing now.

Dave G.

http://www.reliabilitydirect.com

 

[kancil]

Thanks for your reply.
Actually in my experiments, i purposely created an artificial fault on the bearing raceways and rolling element for both ball and roller bearings. The results are presented using an envelope spectrums. The spectrums showed that fundamental spike and its harmonics are detected at their defect frequency location. The similar results are obtained when i repeat the experiments for a few times.
My quention is why the fundamental spike and its harmonics magnitude decayed over frequency?Is this due to the total energy spread or the calculation involved in the FFT algorithm? Thanks.

 

[electricpete]

I have tried to explain the shape of the harmonics in the NORMAL spectrum (not demodulated) at the following link:

http://www.geocities.com/pschimpf/impacts/electricpete.htm

For that case the spectrum can be identified as the product of three components:
#1 - The natural frequency of the bearing components
#2 - The fourier transform of a single impact
#3 - A frequency pulse train which "samples" the frequency waveform at points f=k*f0, leaving only harmonics k*f0.

How would this be modified in looking at the demod spectrum? I believe that the demod spectrum would ignore/discard effect #1 (natural frequency of bearing), since the ringing at bearing frequency gets smoothed out by the demod/envelope. That leaves the primary factor that explains the relative shape of the demod harmonics as #2 - the fourier transform of a single impact.

If you have perhaps something between outer ring and the housing, then you will get a very smooth low frequency impact as a ball rolls over this location... much lower harmonics.

If you have a sharp defect on the inner race which causes a high-rise time high frequency impact, then your higher harmonics will show up higher.

There was a related article by Jan Krepela in one of the predictive maintenance magazines last June which gave several examples of bearing defects and the associated shape of the impact time waveform, and the associated spectra. I think his article answers your question exactly with good examples. I personally think he neglected to consider the effect of bearing ringing on the shape of the harmonics in the normal spectrum, but that's irrelevant to your question on demond spectrum.

Let me know if you want an exact article citation... I have it lieing around somewhere.

 

[kancil]

Thanks for your informations electricpete.
I will very glad if you can send me the article that u mentioned. Thanks again.

 

[kancil]

Can anyone explain for me why the harmonics amplitude decayed when the frequency increased. pls