As Arne suggested, a discussion regarding HFD may be good to help those not so familiar to learn something. Here is current example of an inner race defect, possibly a split inner ring on a 34000lb. press roll. The bearing is an SKF 23260CAK/C3. This data was collected using an Entek dataPAC 1500 with a stud mounted 100mv/g accel. The sensor is connected to an online system through an XM-122gSE vibration module for 24/7 monitoring. I collect data once a day right now because of the defect and am able to get higher resolution with the portable collector. The XM module is limited to 800lor. The filter being used is the 5KHz gSE.

Here is some vendor information....

"Spike Energy measurement was originally developed to detect the signals emitted from defective rolling-element bearings. The term "Spike Energy" describes the very short pulses, i.e., spikes, of vibratory energy generated by the impact of rolling-elements against microscopic cracks and spalls. Spike Energy is a measure of the intensity of energy generated by such repetitive transient mechanical impacts. These impacts or pulses typically occur as a result of surface flaws in rolling-element bearings, gear teeth or other metal-to-metal contacts, such as rotor rub, insufficient bearing lubrication, etc. The measurements showed that Spike Energy is also sensitive to other ultrasonic signals, such as pump cavitation, high pressure steam or air flow, turbulence in liquids, control valve noise, etc.

Spike Energy measurement utilizes an accelerometer to detect the vibration energy over a pre-determined high frequency range. The mechanical impacts tend to excite the mounted natural frequencies of the accelerometers as well as the natural frequencies of machine components and structures in this high frequency range. These resonant frequencies act as carrier frequencies and the bearing defect frequency modulates with the carriers. The intensity of impact energy is a function of pulse amplitude and repetition rate. The signal induced by such impacts can be measured by accelerometers and processed by a unique filtering and detection circuitry. The measured magnitude of the signal is expressed in "gSE" units (acceleration units of Spike Energy). Since its introduction, Spike Energy has been used successfully in many industrial applications and gained acceptance in various industries. Spike Energy measurement can provide early indications of machinery faults and is a very useful tool in vibration analysis. In addition to the traditional Spike Energy overall measurement, Spike Energy spectrum and Spike Energy time waveform were also developed and used in diagnostic analysis in recent years."

This message has been edited. Last edited by: vibbase,

hfdexample.doc (73 Kb, 54 downloads) HFD example

Here is an example of the acceleration data without the gSE filter from the same point.

accelerationexample.doc (72 Kb, 48 downloads) Acceleration example

Ron,

This is definitely a very informative educational material. Demodulation produces much nicer pattern, although TWF in normal data acquisition also speaks volumes.

I think if you did not take normal data with such a high Fmax=300,000cpm, (and probably with high LOR) detection of BPIR would've been impaired in the spectrum.

I wonder how the TWF would've looked like if data was taken with more common Fmax=120,000cpm and 800 LOR? Will you still see clearly these impact spikes?

Thanks,
David

In reality, much of the explanation given for Spike Energy is not valid - it was just hype. When used to get a numerical value, it is/was just a high pass filtered acceleration measurement, mostly the same as HFD. The implimentations changed over time as hardware changed; original spike energy used a 5.5 kHz high pass filter and later this was reduced to 5 kHz with neither version having a sharp cut-off filter. For most machines wtih shaft rotating frequencies above 600 RPM, either 5 kHz or 5.5 kHz was usually a good choice. For most rolling element bearing cases, the statement about accelerometer resonances were irrelevant as were other statements IRD made about ultrasonics, etc.

BTW - I implimented the same thing except for the filter frequency in my IRD 350 plus an envelope detector and showed it to IRD well before spike energy was "invented" by IRD. I didn't invent it either, by 1971 it was obvious, unpatented, prior art.

For an example of the frequency response of the signals that bearings produce, see:

http://vibrotek.com/article.php?article=articles/factors/index.htm

This message has been edited. Last edited by: Duncan Carter,

Duncan, thanks for the update. The information is supplied in the help topics for my software. There may be political issues or proprietary issues or hyped statements involved but the fact is that it works. Using gSE has saved alot of unplanned downtime in this facility over the years. One thing I never do is jump to conclusions. If I see a possible defect in gSE, I will take many higher resolution readings in acceleration and spend time looking at the waveforms as well to verify the defect. The reason for this is because on our paper machine clothing rolls, we have some skidding happening occasionally which will show up as an outer race defect in gSE alone usually low amplitude but nothing shows up in the acceleration spectrum at the higher frequencies.

What's the speed of the roll? Well above the 3 Hz HP filter?

Has spike energy been used successfully anywhere? I've seen it used unsuccessfully a lot and it does early detection but beyond that to point to a time of failure??? Are there case histories showing success?

I'm curious as I have not seen them, which doesn't mean they don't exist or success hasn't been realized - just never heard of it. I've only had people call me asking why they can't see the faults in paper machines (dryer cans) using spike energy.

Don't mean to be controversal but; would like to know more.

spike energy, like all the other methods, can be useful, but you need to understand its limitations which include response to non-bearing signals and internal bearing path limitations. I added it to my 350 for good reason.

All test methods have limitations. For successful use, you need to understand the limitations of the methods.

HFD
Excellent thread very informative.
This filter that Filter who decides where and what filter to use?

I know a vib guy that when asked to give his opinion and help the first thing he would say was did you take spike energy regardless of how it was taken When informed it was taken with a nine inch probe the response was well it isn't ideal but!
We soon reckoned that this fellow had PhD in "did you take spike energy"

If I have got this right enveloped acceleration can be very useful in identifying potential early stage faults, but in the real world where the majority of" run of the mill" machinery is being monitored who is going to call a bearing fault this EARLY? Or even CONVINCE a client that there is some bearing wear or damage.
I have found that when enveloped acceleration levels have increased and a recommendation is made to lubricate the bearings, if the peaks go away it will be ok until the next survey.
However isn't there a time limit for the lubricant to take effect maybe 24 hours? Will a client on a tight budget pay for you to re-monitor this machine? I have yet to find one the usual response is that's what you always say.
High profile plants with big budgets and expensive equipment may take heed but this is dark magic and can be unpredictable.
I do not mean to offend anyone or any particular brand but I am fed up with clients getting some info on enveloped acceleration then quoting how "I'm not that worried that this fan is vibrating at 30 mm/sec because the??/?? Levels are low

Now I know that there are many types of enveloped acceleration products available but honestly how many of us are comfortable and use this can you share your filter settings for Run of the mill 2 /4 6 pole machines
A very cynical vib guy

mcdm, I hear ya man. I would never call a bearing on gSE alone. But look at the attached velocity data on this bearing. This would be the "normal" reading. There is no indication of a defect. And I should give some more details here. The roll in question is large and turns at around 500rpm. The distance from accelerometer to the bearing surface is about 4" of cast iron. My experience with this roll tells me that I better heed the early warning signs. The attached picture is from a previous failure where the data was similar.

velandphoto.doc (86 Kb, 32 downloads)

Thanks vibbase
Now this is what we need actual instances and reasons
"taking into account the distance from accelerometer to the bearing" is a positive engineering tip and one I WILL remember
a simple thing but I never thought of it sometimes you need a
a very good example and reasons to further our Knowledge
anyone else got any?
A lesson lernt "If experience talks listen carefully"
Mike