- All Topics
- GearTalk Podcast
- Training & Events
- Buyer's Guide
The condition-monitoring universe consists of a myriad of tools and technologies that are available to help detect and remediate machine reliability problems. No single tool comes close to doing it all, circumventing the need for all others. Everything depends on using these tools at the right time, the right place and in the right way.
These tools include our senses (used in inspection), oil analysis, vibration, thermography, motor current, and ultrasound. Because of the broad swath of science and function related to their use, the application and benefit of these methods are frequently misunderstood. For instance, years ago the following statement was published in a major trade publication in an article written by a vibration expert:
"In a purely technical sense, lubricating oil analysis is not a predictive maintenance tool. Rather it is a positive means of selecting and using lubricants in various plant applications. This technique evaluates the condition of the lubrications, not the condition of a machine or mechanical system."
Needless to say, I was stunned to read these words and immediately fired off an email to the author followed by a telephone conversation. I strongly encouraged him to attend one of our oil analysis courses, even going so far as to giving him a complimentary pass. He never did.
That experience punctuates the importance of good training across the many condition-monitoring disciplines. At minimum, there is a need for awareness. In such a dynamic era of “digital-this” and “wireless-that”, continuing education is an endless quest.
This article primarily addresses two specific technologies that collectively represent a large portion of the annual spend on condition monitoring: oil analysis and vibration. They are not competitive or alternative methods. Instead, they should be viewed as collaborative.
They are powerful when fully and skillfully deployed. Used alone, they leave gaping holes on the state of health of our machines.
I’ve written extensively on the power of inspection to exploit our keen senses and the supercomputer in our head. Our sense of sight (or “eyeometer”) is not rivaled by any optical camera technologies available today. In perhaps the most powerful “camera” of all, we use our senses to continuously examine our surroundings as we drive through the streets of our cities, subconsciously taking mental “snapshots” of what we see.
Photons instantly alert us to dashboard data, traffic conditions, stop lights, pedestrians, and road hazards. Vehicle sounds and horns flow more data to our brain to process. Our tactile senses are applied to the pedals and movement of our vehicle as we turn, change acceleration, and course along undulating pavement conditions.
We process this data to make real-time changes in the control of our vehicle (steering, braking, acceleration). Why do I bring this up? When you put technology and instruments in the hands of people, they begin to trust their own senses less.
Despite all that we can and should extract by the power of inspection, our machines are largely exoskeletons made of cast iron and plate steel. Photons will not penetrate these barriers, but we can instead extract the data we need from other methods including oil analysis and vibration analysis.
But, aided by the use of sight glasses, along with audible sounds and the sense of touch, we can also use inspection to pick up critical oil-analysis and vibration data. See this article on Sight Glass Oil Analysis.
Oil touches the most critical and concerning surfaces in our machines. These include the frictional zones in our bearings, gears, pistons, and cams. As a common medium, it’s darn hard for a machine to fail or have a failure-inducing condition without the oil knowing about it first.
Just name your poison: corrosive agents, misalignment, overloading, material defects, lubricant starvation, wrong viscosity, dirty/wet oil, etc. — something will show up in the oil. It’s like a flight data recorder.
Let’s take particles for example: They are known to be both a major cause of failure (abrasion, contact fatigue, stiction, erosion) and a major consequence of failure (wear debris), regardless of the cause. So, if particles are the cause and/or effect of failure, wouldn’t we want to be rigorously monitoring them in our lubricants?
Consider this: If our machine’s lubricant is tested and found to be extremely clean, what can we conclude about the health of our gears and bearings?
There are exceptions, of course, and oil analysis doesn’t alert you to certain well-known failure modes, such as machine imbalance, shaft cracks, misalignment, looseness, soft foot, and resonance. However, these conditions soon transition through failure stages to produce wear debris along the way.
Wear debris are excavated particles from bearing and gear surfaces due to mechanical impact and sliding conditions. These particles will trigger analysis alarms.
Catching abnormal wear and critical machine faults long before catastrophic failure is one of the main advantages of oil analysis. If accelerated abrasive wear suddenly occurs in a bearing with 90% remaining useful life (RUL), this can be detected and the condition corrected. In contrast, abrasive wear may actually attenuate bearing and gear vibration until so much material is lost and precipitous failure commences.
Palo Verde Nuclear Generating Station found that “for oil lubricated bearings the oil analysis results usually show up earlier than the vibration indications. If the bearings removed for only oil analysis alarms had been allowed to run, the vibration symptoms would have eventually shown up.”
The pie chart in their report (presented to the Vibration Institute), shows that oil analysis detected bearing faults 40% of the time not reported by vibration, while vibration detected bearing issues 33% of the time not reported by oil analysis.
Only 27% of the time did both technologies see bearing faults in unison.
For rolling element bearings, Palo Verde claimed to pick up incipient failure using wear-debris analysis 18 months out, in contrast to vibration, which typically detected failures only one month out using Spike Energy. Similarly, a Monash University study induced various failure modes in gear boxes and reported that oil analysis caught the threatening conditions on average 15 times sooner than vibration analysis. They also pointed out that certain types of failures were undetected by oil analysis.They also pointed out that certain types of failures were undetected by oil analysis.
A well-known disadvantage of oil analysis is the requirement to pull a sample that needs to be packaged and sent off to a commercial lab. At times, several days can pass before data arrives on the client’s dashboard.
However, many modern oil analysis programs include the use of portable instruments and, for certain machines, online oil sensors. Onsite oil analysis labs are also common that allow tests to be run at the plant the same day.
It is widely known that vibration analysis has limited use in particular applications involving slow-moving, reciprocating, hydraulic, and articulating machines. Even journal bearings and multistage gearboxes can pose real challenges for vibration analysis.
That said, vibration analysis is highly important — almost foundational — to machine condition monitoring.
The list of faults that can show up at one-times (1X) running speed is long. These can include misalignment, unbalance, looseness, partial rub, broken gear tooth, rotor bow, coupling problems, cracked shaft, and broken motor rotor bar. Most of these issues relate to an increase in dynamic forces or a reduction in effective stiffness.
In contrast, oil analysis would struggle to detect any of these faults at the time of initiation. Think of the P in the P-F interval. Like oil analysis, vibration has blind spots or at least areas of blurred vision. Let’s take journal bearings as a case in point:
The following is a commonly seen timeline to bearing wipe. It is apparent that early detection depends on alarms from oil analysis, inspection and temperature.
No doubt, when in comes to vibration analysis a great deal of the focus should be on mitigating root causes, also referred to as “proactive maintenance”. As aforementioned, root causes for vibration include balance, alignment, looseness, soft foot, resonance, to name a few.
While using vibration analysis to catch impending bearing failure (predictive maintenance) is also valuable, often the P in the P-F interval is way too close to the F. In other words, most of the remaining useful life can be gone.
Vibration analysts have reported that 90% of the problems they detect are balance and alignment related, or in other words, destructive precursors to bearing failure. If such conditions go uncorrected, the vibration will use up all the available bearing clearance, and in doing so, will “nibble away” at the bearing surfaces, effectively making its own clearance (wear) while vibration continues to increase.
This condition robs thousands of hours of bearing service life. Take a look at the chart in Figure X. When vibration overalls are high (say, due to alignment and/or balance), the penalty on bearing life is severe — cause and effect.
Ref. Heinz Bloch
Particle and water contamination of lube oils can have a similar effect of bearing life as shown in Figures XY and XZ.
Vibration analysis of low amplitude signals in the high frequency domain (ultrasound) has been used for years to detect abnormal wear in rolling-element bearings. These methods go by different names including:
Values are usually trended with no absolute limits. Success depends on proper sensor mounting as well as the filter settings that are chosen.
One such method, referred to as stress wave analysis, claims to overcome many of these challenges and limitations. Application can be either route-based or via online accelerometers through an IoT network. More importantly, it has promising potential to tease out time-domain events like impact, bearing and gear defects, rubbing, severe sliding, mechanical friction, jerk/slip-friction and electric arching.
Why is this important? Primarily because mechanical asperities can collide or rub due to various reasons. These metal-to-metal contacts result from mechanical causes such as overloading, misalignment, unbalance and shock. Impaired lubrication conditions are frequently responsible too, caused by restricted lubricant supply, viscosity starvation, contamination and AW/EP additive depletion (loss of film strength).
Regardless of the cause, when rubbing, sliding, galling, and abrasion occur, they transmit an ultrasonic signal that often gets “lost in the sauce” with traditional FFT spectrum analysis. Conversely, stress wave analysis appears to be able to quantify these time-domain events and send alerts as conditions and severity warrant.
Follow this link for more information on stress wave analysis.
In vibration, analysts look at such things as gear mesh frequency, blade pass frequency, roller/ball pass frequency and cage frequency. Vibration frequency identifies the source, amplitude tells us how bad it is.
In oil analysis, we look at changes in an oil’s physical and chemical properties, particle size distribution, machine metallurgy, particle morphology and composition, corrosive agents and varnish potential. It’s a whole different ball game.
The initial onset of wear from dirty oil, or “impaired lubrication”, does not cause the excitation forces that lead to frequency-domain vibration alarms. However, given enough time for wear to advance, vibration signals become clearly detectable. Sadly, this often occurs at less than ten percent of the bearings remaining life.
Ranking failure modes by probability and consequences (also called “criticality”) is a good strategy for laying out your condition-monitoring game plan, machine by machine. Each ranked failure mode needs one or more matching condition-monitoring strategies for early detection. Using this approach, vibration and oil analysis can be seen to work closely together. But don’t forget about other condition-monitoring methods, including inspection, ultrasound, motor current, and thermography.
Combining condition-monitoring technologies gives us a much more complete picture of the state and health of our machines. One helps fill in the gaps of the other. The need for combined monitoring is especially true for multi-stage gearboxes, plain bearings, rotary screw compressors, roots blowers, and rolling element bearings. When both technologies pinpoint the same problem, the diagnosis and follow-up recommendations are rarely inaccurate.