With the proliferation of online monitoring systems utilizing permanently mounted sensors, users will need to beware of “direction sensitive” vibration and possible sudden unexpected failure due to insufficient data. The thought of insufficient data may seem incredible when thinking of constantly monitored equipment, but consider the all too common (IMHO) practice of uni-directional (one direction) monitoring of machine trains.
Many installations, due to initial cost, are mounting a single vibration sensor at each bearing. While this may be sufficient for most equipment trains, most of the time, it will certainly not be sufficient for all equipment trains all of the time. Although I don’t have hard data available, if I were to make a statement based on personal experience, and anecdotal evidence from other practitioners, my statement would be something like this: “80% of horizontal equipment could be pretty well monitored by sensors mounted at the horizontal radial position on each bearing.” I say pretty well monitored because I just can’t bring myself (as an analyst) to be completely satisfied without the vertical and axial data.
This setup would catch virtually all unbalance and roller bearing faults (excluding thrust bearings), some to most misalignment faults, and a sprinkling of others. I use the word “catch”, to mean it would give an indication of a developing problem. Accurate diagnosis of unbalance, misalignment, bent shaft, and even looseness in many cases (as well as a host of other possible faults) would require more data.
If the online vibration program manager takes these facts into account and governs the program accordingly, they should be pretty successful. If they add to the online program a “full battery” vibration survey, maybe semi-annually, just to catch the less common, but possibly very destructive defects that could develop undetected by the uni-directional monitoring, they would most likely be very successful.
What could be so destructive and yet be completely undetected by the uni-directional sensors? The Big R for one is Resonance. Resonance is often extremely directional. Consider a case history LUDECA co-published with one of our customers in the December 2012 Wastewater Processing magazine:
In the table below (Figure 1), the 1× amplitudes are displayed. I have hidden all but the vertical data, as though it were monitored only by vertical sensors.
Everything is wonderful right? Look at the motor outboard vertical, only 0.00384 inches per second—very impressive. Of course, at this point you are thinking “he is setting me up for something” and you are correct. Even though most anyone would love to have these amplitudes on virtually any machine, this particular machine was tearing itself apart with vibration!
We will give the reader a little more data, just to help add emphasis to the directional nature of resonance. We will add the axial data to our table in Figure 2:
Still very, very good… so far. Now, look at Figure 3, with the addition of the horizontal data.
The motor outboard horizontal amplitude is 162 times the amplitude of the motor outboard vertical! What if the user had only vertically mounted sensors? What about vertical with the added information of axial? You may be thinking “if I had only horizontal sensors, I would have been ok”, and for sure you would have been better off than having only vertical. You would at least have known you had a problem, but you would not have known what that problem actually was. You would likely have assumed the vertical and axial are probably vibrating badly too.
Hopefully, you would have verified the vibration in the other directions. As it was, the user had data from all directions and a simple glance told the analyst with a high degree of confidence what the problem was. Resonance is almost alone in creating that kind of directional disparity.
To reiterate, the online vibration program manager should be successful if they take into account the fact of limited data and supplement the online program with a “full battery” vibration survey at a cost-effective interval, just to catch the less common, but possibly very destructive defects that could be developing undetected by uni-directional monitoring.