Hello all,

I'm new in the field of condition monitoring and vibration analysis. My main goal is to develop a system which will detect bearing faults in very early stages. I read a lot of articles about bearing condition monitoring and vibrodiagnostic, but I didn’t find a solution jet. The reason why I didn’t find it jet is in the specific characteristics of bearing witch I have to diagnose.

Namely, I would like to diagnose an axial slewing ball bearing with extremely low rotational speed (1RPM to 10RPM). The bearing is not standard, it has external gear and is quite big (pitch diameter = 650mm, number of balls = 80).

So my question is: what type of transducer (displacement, velocity, accelerometer, ultrasound,…) should I use for my application? ...I think that the main problem for this decision is hiding in the low rotational speed of bearing (because of the very low frequencies, bearing tones, which are present in the signal).

I would be most grateful if whoever could point me in the right direction.

Thank you in advance for any advice.

Mateo

Mateo,

We have had some success using the raw time waveform from an accelerometer and looking at the periods between peaks in the waveform to identify faults. You want to use a quality accelerometer with a good signal to noise ratio which your vendor should be able to help with. Another method is Envelope Signal Processing, but would want to take a lot of averages which is time consuming. The speeds I have been dealing with are not as low as you are talking about but the methods may still work.

If the bearings run in an oil lubrication system you should use oil analysis to monitor the condition of the bearing using the return oil line from the bearing as I think that will give you an earlier warning of a problem.

Hope this helps,

Nick

CSI's PeakVue technology works very well on low speed applications.

Mateo
this is not your typical vibration analysis application. SKF's Acc. Env. will work on some of these applications, but even with Acc. Enveloping, the key is in the time waveform as Nick has stated. I always go to the time waveform on slow speed applications. I think this is true no matter who's instrument you are using

If say your low freq of interest is 1 Hz and you're using a 1 V/g (maybe capacitive acel) and interested in mags of 0.01 IPS then you're looking at 0.0001667 volts. So, your system will have to be below this in order to keep a valid signal above the noise floor and keep it real. Your acel may need be to DC w/hi-resolution. Processing those data will mean all techniques available - FFT, TWF and if positioning is of benefit then use an eddy current probe to monitor DC component to measure clearance reduction.

Mateo--

I'm having very good luck with a combination of HFE techniques (CSI's Peakvue) and velocity measurements integrated to displacement using a 500mV/G velocity (not piezo-velocity) transducer. The latter had very clear BPFO and BPFI on two different applications running around 10 RPM. The HFE data showed it, but long after the velocity transducer did and never as clearly--especially in the spectrum.

Tony

Hello!

Thank You: Nick, Danny, Stephen, Sam and Tony. I really appreciate your help and quick replies to my question.

I have a few more questions:
1. Where could I find some basic theory of the envelope method and a PeakVue technique?
2. You mention that the key for the low speed application is in the analysis of time waveforms. What kind of analysis should be done on these waveforms?
3. I choice to use accelerometers for my analysis. Could you tell me what kind of characteristic should this accelerometers have (g-range, sensitivity,…)

Once again, thank you for your answers.

Mateo

PS: I really appreciate your help and look forward to receiving more information
from you!

Mateo--

Emerson/CSI has an excellent paper on their site by Dr. Robinson and Dr. Berry that I think you'll find pretty helpful--not only explaining principles, but also suggesting setups, alarm levels, etc. I'm not savvy enough to easily provide you a link...

In analyzing waveforms (especially for very slow speed data) I'm looking mainly for relative amplitudes, change in amplitudes (trend Pk-Pk Waveform) and repetitive impacting. The autocorrelation (sp?) function in RBMware is helpful here at times.

Finally, I'm using a PCB/IMI 326A01 500mV/g "low frequency" accelerometer. I don't remember the g-range, but it seems to do the trick.

Hope this helps! Tony

http://www.compsys.com/drknow/aplpapr.nsf/06b6f5a4de2ea...&Highlight=0,PeakVue

This is the link to get the CSI paper by Jim Robinson and Jim Berry on PeakVue. I usually use a general purpose 100 Mv/g accel even for very slow speed stuff. More important is sensor mounting, Make sure you have a good, clean, smooth, flat surface with no paint or dirt to take the data. For very slow speed (under 10 rpm or so) use as low f-max as possible.

Jim--

Thanks for posting the link; I still have VCR's blinking "12:00" at home as well...

Have you compared the results from a 100mV/g with a 500mV/g? When I did the latter seemed a lot more sensitive (i.e. higher amplitudes overall, especially on "known" problems). I'm just wandering if that was a fluke and I was just getting the "results" that I was "expecting" from some not-so-scientific tests!

Thank you! Tony

Mateo also posted this question on VibroTek's bulletin board at:

http://www.vibrotek.com/wwwboard/wwwboard.html

Mateo,

Looks like you are getting some good feedback already, but for the signal processing contact Rockwell Automation (Entek) and they should be able to email you a paper on Envelope Signal Processing. On the Time Waveform, theres a company called Universal Technologies that run good training on time waveform. Who ever supplies you with the vibration equipment should be able to provide help in terms of your analyser settings. As a general rule for time waveform you want to get 6-10 multiples of the frequency of interest. e.g If you are interested in a defect at 10 rpm (0.17 Hz) you will need to set a time period capture of 36-60 seconds.

Hello all

Thank you all for your advice, direction and ideas. I really appreciate you taking the time to help me out.

Mateo