How does Acoustic Emission (AE) compare with vibration?
All engineers know that machinery vibrates when it is running and that some serious fault conditions can give a significant increase in vibration. It’s a good way of detecting the presence of out of balance for example.
Unfortunately this overall level of vibration is typically insensitive to more subtle effects such as the early signs of bearing and gear teeth wear. To overcome this, Vibration Analysis has to be carried out where the vibration signal is pre-processed using subjectively set filters and then analysed in the frequency domain (uses an FFT to provide a frequency spectrum). To interpret this vibration frequency spectrum it is necessary to calculate all the possible defect frequencies which could be present (this is not a trivial task and requires precise information on machine speed and bearing and gear geometries). Once you have done this you can observe the signal levels at each of the possible defect frequencies and see if they have increased since the last time you made a measurement. You need to do all this at several positions on each machine – even the majority which are in good condition! If this all sounds complicated and time consuming that’s because it is.
By contrast our unique implementation of the AE technique homes directly in on the high frequency (-100 kHz) component of the elastic waves being generated by operating machinery. The resulting AE signal is very strongly influenced by fault processes and has a much reduced sensitivity to the effects of normal running (ie good machines are much quieter at 100 kHz yet machine faults which result in deteriorating contacting surface give rise to very loud signals). Because of this it is possible to analyse the overall AE signal (without homing in on defect repetition frequencies) in order to provide a clear indication of the presence of faults.
Experience in the field has shown over and over again that our MHC products are much simpler and quicker to use than vibration instruments yet every bit as sensitive to the earliest signs of mechanical degradation. In fact there are numerous instances where it is far more sensitive than FFT based vibration techniques.
How can Distress know the difference between good and bad if it has no history, speed or design data to refer to?
Of course there are some commercial secrets behind the exact formula of Distress, but how it works is not a total mystery requiring unquestioning faith on your part. First of all we know that rotating machinery is designed to run smoothly (ie low level of impacts) and with low losses (ie low friction). This means that machinery in good condition tends to produce low levels of AE activity. As contacting surfaces start to deteriorate the impacts and momentary rubs give rise to isolated transients. The AE signal from a machine in good condition doesn't have these features and the AE signal from a machine in poor condition does. Its difficult to imagine anything simpler more direct or more plausible as a basis for measuring machine condition.
The beauty of the Distress parameter is that it builds upon this observation with a specific analysis algorithm which does not require you to input machine specific parameters. What's more the Distress parameter has been very extensively proven across the whole of industry to provide a common interpretation on virtually all rotating machinery (including motors, pumps, fans, gearboxes and roll support bearings).
Because Distress homes in directly onto the fundamental difference in AE signals between good and bad rotating machinery the same approach is equally applicable to rolling element bearings, plain bearings and gear teeth (Note: for the case of gearboxes activity associated with gear teeth condition is detected on the casing adjacent to the shaft-end bearings).
If the MHC instruments work at 100 kHz and reject low frequencies how can they produce an audible output?
This is because its audible output is produced by a process of amplitude
demodulation in a similar way to how an AM radio works (it detects at radiofrequencies, strips out the radio frequency carrier and outputs the envelope ofthe signal which is at audio frequency).With AE signals from machinery it’s an interesting fact that the clicks,crunches and rubs associated with damage produce bursts of detectable AE activity which, when it’s demodulated in this way, sounds just like clicks, crunches and rubs as if you were listening to it directly at audio frequencies. The difference is that the high frequency detection of the AE method is insensitive to everyday low frequency sounds. Maintenance personnel tell us that this results in a much clearer signal from the AE headphones than listening with a screwdriver or stethoscope.
If I buy an MHC instrument how will I know it is working properly?
Experience tells us that there are very, very few instances of alteredsensitivity of our AE sensors or MHC instruments in service due to our highly stable designs. To provide confidence that all is well we recommend that every time the instrument is to be taken out to do a series of measurements you should as a matter of good practice : Switch on the instrument with the sensor laying on your desk and confirm that Distress and dB readings each drop to a value below
1.Rub the sensor front face and confirm that Distress and dB Level readings each rise to a value well above 1. If you are taking measurements in a region of particularly high electromagnetic noise confirm immunity by hanging the sensor in free space and checking that Distress and dB readings each drop to a value below 1.
How important is the positioning and orientation of the sensor?
Our MHC instruments are designed to be primarily responsive to the 'diffuse field' of AE activity. This differs from many other high frequency systems on the market which are peak responding. The benefit of the diffuse field approach is that in general the measurement is relatively insensitive to the precise position and orientation of the sensor on the item being monitored. This is similar to when you listen to the radio; it doesn't matter where you stand in the room, or even which way you are facing, the amplitude and information content is very similar at all positions. The exceptions are when you put your ear very close to the speaker (don't try this it may damage your hearing) and when you leave the room and these exceptions have their parallel in MHC measurements also.
How repeatable are AE measurements?
We don't know who the culprits are but someone has done a good job of convincing many engineers that high frequency or AE measurements aren't repeatable. We can't speak on behalf of other vendors of AE or high frequency technology but we can categorically state that using our unique implementation it is a simple matter to get reproducibility of measurements to within 1 dB when a sensor is removed and re-applied to a machine. In fact in a lab trial we showed that a 96% confidence of the measurement being within + 0.3 dB is easily attained. We think this is superior reproducibility to that you can achieve with accelerometer based vibration measurements. Another aspect of repeatability in a measurement relates to the repeatability of the signal within the machine being monitored. Reproducibility to within 1 dB on real machinery in the field is not uncommon but of course there could be some variation in its operational performance and this may lead to a 'month to month' variation which is typically within a 3 dB band. Of course deterioration in the machines condition will also cause an increase in the AE signal magnitude and for normal speed machinery an increase of 30 or 40 dB prior to final failure is not uncommon. For completeness its worth mentioning that the repeatability and measurement integrity referred to above is no accident but results from our paying great attention to transducer, signal conditioning and signal processing designs in all our products.
What speed of rotation can it be used down to/up to?
Standard MHC instruments characterising the AE signal in terms of Distress and dB Level are in general OK down to shaft speeds of 45 rpm or even lower. We don't know a top speed for it to be effective since most applications are at 1500 rpm or less. In reality we have detected misaligned cutting tools on 8,000 rpm routers without any difficulty. Similarly we had no qualms about on-line monitoring of the Thrust SSC wheel bearings which ran at speeds up to 8000 rpm. At high speed we would definitely recommend monitoring dB level as well as Distress. For very slowly rotating machines we have developed a new algorithm which characterises the AE signal in terms of dB Level and three new parameters rather than Distress. In its standard implementation this method is OK down to 0.25 rpm (ie 4 minutes per revolution). It’s available in the MHC SloPoint machine surveillance DIN rail module. Did you know we spent more than 2 years beta trialling this new slow-mode at several industrial sites before launching our first commercial product incorporating it; the MHC-SloPoint? This demonstrates our long term commitment to AE technology and the respect with which we treat our existing and future customers.
How do I know I haven't got cross-talk from adjacent machines?
If you are familiar with Vibration monitoring then you'll be wary of a CM instrument that gives an overall reading since at low vibration frequencies there is very strong cross talk (or sympathetic resonances ) between adjacent machines. At high AE frequencies the changes in shape and material in which the waves propagate rapidly attenuate such distant activity. Experience shows that there is less than a 1% chance of cross-talk problems from adjacent machines on the rotating machinery we are asked to monitor across the whole of industry.
How reliable is a measurement made with the MHC instrument?
No single CM technology can claim to pick up all failure modes,on all machine types all of the time. Consequently we don't make any such wild claims for the MHC instrument. The purpose of all CM instruments is to increase the chances of detecting faults in operating machinery at a timely stage in their development. Our own experience and that we know of our customers in using the MHC is that it has an excellent track record for bearing and gear fault detection. It’s difficult to put a percentage figure on it but we believe it to be in the high 90's. We have no evidence of any other CM technology or CM instrument having a higher success rate or a broader spread of applicability in general application to rotating machinery. Indeed we are aware of CM specialists who use an MHC to extend their monitoring capability to include those awkward to monitor applications such as slow speed machinery, worm-wheel drives, plain bearings and gearboxes with unknown internals. We certainly have no apprehension when prospective customers do a back to back test with an MHC instrument against any other CM instrument on the market. To date the most common outcome of such a trial has been that the MHC has been found to be at least as sensitive to faults as others under test and far simpler and quicker to apply.
Is it the same as other manufacturer’s high frequency or AE type measurements?
This is a common initial question when we go into a company who have had either demonstrations or previous experience of such equipment which is different from the more familiar vibration techniques. We can categorically state that our technology is both unique and distinct from all such other instruments on the market and because of this we have never sought to associate the capabilities of our approach with that of any other such instrument. When you see the MHC in use the first thing that will strike you as being different is the way that you just switch the MHC on and the Distress measurement gives an instant assessment of condition without the need to input machine specific information such as shaft speed and bearing details in order for it to do its analysis properly. This means that the MHC is more direct, far less prone to being misapplied (eg where you're not sure of the speed or the internal construction) and more appropriate for applying to complex machinery (such as gearboxes) where there are many bearings and gears in close proximity. Another very important aspect from a practical viewpoint is that measurements with some technologies are well documented to be critically dependent upon the positioning and orientation of the sensor with respect to the plane of the bearing and the loaded zone. For such methods sensitivity to faults and consistency of measurements can only be achieved if great care is taken. By contrast our unique implementation of the AE method makes measurements very insensitive to the precise position and orientation of the AE sensor. But you don't have to take our word for it why not compare one of our MHC products with any other CM instrument on the market and see for yourself which is the easiest, quickest and most effective for you.
Can the MHC distinguish between activity from two bearings sharing the same housing ?
Perhaps confusingly the answer is Yes and No! Take for example the case of letter sorting machinery where a single MHC measurement on a bedplate is used to simultaneously confirm that the bearings on tens of spindles bolted to the bed-plate are all in good condition. Even when only one spindle bearing has a fault it can be detected by this single measurement. However If you want you can position the sensor very close to an individual bearing (in it’s near field) and this boosts the fraction of the overall signal coming from the bearing of interest. On machinery in general where two bearings are close together you can often localise which bearing has the problem by comparing readings when the sensor is directly positioned on one then the other (however this may not be possible when two bearings are back to back). The situation is analogous to when you are at a party; you hear the general indecipherable noise of multiple simultaneous conversations in a room but it is very easy to detect if anyone coughs. Similarly if you stand sufficiently close to someone its possible to concentrate on what they are saying with less of an effect from the others.
Can AE be used on plain bearings/journal
Yes. Why shouldn't it, degradation leads to increasing impacts and friction in advance of final failure so the arguments are much the same.
If AE is that good why did people ever develop vibration?
This is a good question, it beats us as well !! Being serious though it’s a question of evolution. Whereas AE has always been beyond the range of the human senses the vibration technique grew as an instrumented extension of them, so it’s natural that it was developed first. It’s analogous to the way that optical techniques were developed a long time before X-Rays. But don't forget that if you are actually interested in how your machines are 'leaping about' then vibration is the undoubted choice since this aspect of vibration monitoring is both well understood and calibrated.
Information Source: Holroyd Instruments Ltd website