Inspecting Reciprocating Compressors with Ultrasound

Article extract from ReliablePlant newsletter:
http://www.reliableplant.com/Read/29745/reciprocating-compressors-ultrasound

Anyone who has benefitted from the advancements made in ultrasound testing will agree that this technology is no longer just for leak testing. It is a complimentary companion technology to vibration analysis, infrared imaging, motor-circuit testing and oil analysis. Some will assert that the foundation of any predictive maintenance program should be based on ultrasound first due partly to its versatility, low cost and easy implementation.

Nowadays, meeting reliability professionals who still believe in archaic methods for detection of machine faults is rare. That's why it was surprising to read Dan Wise's technical paper, "Knocking Compressor? All you really need is a screwdriver to find the source of the problem." His message compelled me to climb back on my soapbox and scream, "You need ultrasound!"

Safety should be taken very seriously. Inspector training is as much about safety as it is about procedure, so it's hard for me to look past the obvious hazards that could come from sticking a greasy screwdriver in your ear and then placing the other end against a knocking compressor valve. It is also difficult to ignore the added risk of working in a noisy compressor room without wearing approved safety gear.

Wise's paper suggests that ultrasound is unnecessary because the technology "takes time to learn." Instead, he urges inspectors to develop their skills with a screwdriver "by spending time, each day, paying attention to the sounds of compressors." His advice is conflicting. Are we better to spend time each day learning his "old-school" screwdriver technique, or are we better off to spend our time learning how to use an ultrasound instrument that can do so much more?

Figure 1. Inspecting reciprocating compressor
valves with an ultrasonic data collector

For more than 20 years, ultrasound has been providing early warning signals for mechanical failure. If you can hear compressor valves knocking with a screwdriver, you are too late. However, there is a fundamental difference in philosophy between what ultrasound inspection proposes to accomplish and what is hoped to be achieved with a screwdriver. Ultrasound monitors the condition of that compressor valve from the time it is new until the time it is ready to be replaced, all the while providing feedback about subtle changes in its operational condition. The screwdriver method is used "to find the source of the problem," which suggests the knocking is already perceivable to the human ear and it is only a matter of days, hours or minutes before the compressor will fail. I call this firefighting. It is certainly not condition-based monitoring, and it leaves no room for planning the repair.

The paper warns that "valves can break without warning for a variety of reasons" and goes on to list some common causes for failure. The real reason machine systems fail "without warning" is because no warning system is in place. Using a predictive technology such as ultrasound condition-based monitoring to indicate subtle changes in machines over time allows industry to avoid firefighting catastrophes.

Advanced Valve Testing

A huge benefit of ultrasound is its ability to record dynamic wave files on reciprocating compressors. Using simple time waveform analysis software, it is easy to expand 1/10th or even 1/100th of a second of data to see a valve open, exhaust and close. During that blink of an eye, you can see if there is complete sealing of the valve seat and can even assess the strength of the valve spring. All of this data can be printed in a logical report, which serves as a summary for management to decide if further action is necessary. Because the document becomes a historical archive, the next decision will have a basis of comparison.

Reciprocating compressors contain one or more cylinders that compress gases. A simplified example is shown in Figure 1. Pistons travel up and down within the cylinder. On the downstroke, new gas is drawn into the cylinder through an inlet valve. On the upstroke, gas already trapped in the cylinder is compressed and pushed through an outlet valve to a receiver or tank. In this example, a vacuum pulls the outlet valve closed and the inlet valve open, while pressure does the opposite. In more complex reciprocating compressors, more robust valves operate with springs.

Opportunities for failure are many. Seal integrity around the piston cylinder wall reduces the efficiency of compression. Dirty and worn valves do not seat properly due to corrosion and buildup around the valve head. Internal leakage around the piston and valve head can be detected with ultrasound inspection by recording a dynamic ultrasound signal and analyzing it over time. Early mechanical failures, which can eventually lead to screwdriver-detectable valve-knocking, can be seen this way, too.

In Figure 2, you can see the impact made by the valve when it is seated and the suction of turbulent air flow when the valve is opened. High amplitude signals between the valve impacts can reveal mechanical looseness (early knocking) or internal leakage from piston rings or the valve seat. A comparison of similar valves can be made easily by scaling the Y-axis of the time signal and presenting it in overlay or side-by-side mode.

A documented report that cites the historical evolution of your compressor's health can be an extremely valuable and professional way to present your data. It is a great tool for conveying to maintenance managers that a problem is escalating and that action is required. It will hold much more credibility than walking into your boss's office with grease dripping from your ear and suggesting that a compressor valve sounds like a "bongo drum" and should immediately be removed from service.

There is no substitution for walking the plant floor and visually taking account of the state of affairs. Much can be learned by talking with operators. Operators have the best pulse for machine condition because they work around them every day. For those who are ready to take the next step (implementation of a condition-based monitoring program), getting to know the expectations of operations, management and front-line staff is the best approach. From that point, defining the assets that need the most critical monitoring will be as an important of a step as figuring out which technologies should be used to collect the data.

Documentation is the final piece of the puzzle. Without it, all other initiatives are pointless. Needless to say, modern techniques for collecting data that can be trended and reported are preferred over archaic methods that rely on individual interpretation and old-school tricks.


Figure 2. Dynamic time signal captured on a recip valve head

About the Author

Allan Rienstra

Allan Rienstra is the president of SDT Ultrasound Solutions and co-author of "Hear More: A Guide to Using Ultrasound for Leak Detection and Condition Monitoring." He has spent the past 20 ... Read More

Vibration and Oil Analysis Techniques Reveal Root Cause and Severity

Article extract from ReliablePlant newsletter:
http://www.machinerylubrication.com/Read/29123/vibration-oil-analysis

When evaluating mechanical defects in roller elements and gears commonly found in rotating machinery, there are several techniques that can be applied to reveal root cause and severity. One is wear particle analysis using either a filter patch or glass slide. Another technique is a wear particle ferrous-density measurement method often referred to as ferrous index. A third technique is an impact-detecting vibration stress-wave analysis method called PeakVue®.

Each of these techniques independently provides insights as to the nature of a damage-causing mechanism, including abrasion, fatigue, boundary wear or corrosive wear. Root cause and severity information learned from these techniques can then guide operators to call for the best proactive and corrective actions. Uncorrected, these mechanisms will produce significant mechanical deterioration.

Figure 1. Vibration and oil analysis reveal proactive root causes and predictive failures in progress.

Industrial plant maintenance departments typically use walk-around and on-line vibration analysis together with periodic oil analysis to determine condition-based maintenance on rotating machines like pumps, motors, compressors, gearboxes, turbines, fans and rolls. Figure 1 shows how vibration and oil analysis collectively address proactive condition monitoring by finding root causes before damage occurs, as well as by identifying component failures in progress, which might range from incipient to catastrophic.

PeakVue Technology

The PeakVue technology is a highly sensitive and trendable impact-detection method for quantifying defects in roller elements, bearing races and gear teeth. It is a unique stress-wave analysis technique distinguished from demodulation, shock pulse, spike energy and other peak detection methods in that PeakVue detects and holds a scalar maximum peak value from each oversampled plurality of sample values collected during each sample interval.

For example, Figure 2 is a comparison of demodulation with PeakVue in six different measurements of the same vibration signal having a periodic impact. The three plots on the left show how demodulation captures and represents only a small fraction of the total signal, so the portion detected is substantially diminished by the sampling bandwidth (1,000 hertz top, 200 hertz middle and 50 hertz bottom). The three plots on the right reveal how PeakVue captures and consistently represents the same data independent from bandwidth selection.

Figure 2. Demodulation (left) vs. PeakVue (right) showing waveform for bandwidths of 1,000, 200 and 50 hertz respectively (top to bottom).

Ferrous Index for Ferrous Density

Abnormal wear mechanisms of abrasion, fatigue, boundary wear or adhesion along with corrosion are common to rotating machinery such as gearboxes, pumps, motors, compressors, rolls, presses and transmissions. A ferrous-density measurement like ferrous index is an excellent method for detecting abnormal wear in the 5- to 60-micron size range. This is usually the result of abnormal fatigue caused by cyclic loading and boundary wear where lubrication is inadequate.

Emerson, Spectro Form Alliance

Emerson and Spectro Inc. recently announced an alliance to combine technical innovation and expertise to deliver best-in-class oil analysis solutions. By combining Emerson’s oil analysis application for the process industry with Spectro’s extensive oil analysis product line and expertise, users can gain more meaningful and accurate information on the condition of their machinery.

Under the new agreement, Spectro gains rights to a suite of Emerson-developed and patent-protected intellectual property in the field of oil analysis for predictive machine maintenance. Included in this portfolio is Emerson’s AMS Suite oil analysis software module as well as the CSI 5200 Machinery Health Oil Analyzer, which is part of the Trivector minilab package.

Effective Oct. 1, Spectro will become the exclusive worldwide supplier of the Spectro 5200 Minilab product and services offering. The onsite minilab is used to quickly test incoming and in-service lubricants with immediate retest when needed, enabling improved lubricant contamination control with effective root-cause detection and elimination.

This ferrous-density measurement is highly sensitive to wear debris or dust contamination produced by abnormal abrasive wear. It is a measure of wear particles typically containing iron alloy particle debris in sizes as small as 5 microns and extending to at least 60 microns. This size range is frequently produced by abnormal severe sliding (boundary lubrication) wear or fatigue wear.

Figure 3. An ASTM D7416 minilab reports
ferrous index and wear particle analysis.

The ferrous index is defined in ASTM D7416 as a “ferrous-density-type parameter measuring the relative concentration and size of magnetically responsive iron particles greater than 5 microns collected on a dielectric-permittivity sensor.”

Wear Particle Analysis

Wear particle analysis is a direct approach to visualizing damaging causes and effects taking place in lubricated machinery by capturing and viewing particles extracted from lubricating oil. ASTM D7684 is a new “Standard Guide for Microscopic Characterization of Particles from In-Service Lubricants.” It provides excellent recommendations for terminology and techniques, such as filter patch analysis or analytical ferrography, and offers insight as to what is likely happening inside the rotating machine.

This new guide defines rolling contact fatigue wear as being “caused by loaded rolling contact typically between the roller and race in bearings or between gear teeth in the vicinity of the pitch line, typically forming spall-type pitting and releasing rolling contact fatigue particles.” Rolling contact fatigue particles are described as “flat platelets, with a length more or less equal to their width, with smooth surfaces, random, jagged and irregularly shaped circumferences, and a major dimension-to-thickness ratio in the range of approximately 5:1 to 10:1 or more.”

Figure 4. PeakVue data from a bearing pedestal.

Performing Ferrous Density and Wear Particle Analysis Onsite

The onsite minilab shown in Figure 3 can be used to measure wear, contamination and chemistry parameters of lubricant samples, including ferrous index and wear particle analysis. Industrial plant maintenance departments employ onsite analytical tools like this to get immediate feedback with retest capability if needed.

Figure 5. Evidence of boundary
lubrication (sliding wear)
due to inadequate lubrication revealed
by wear particle analysis.

On the following page is an example of PeakVue and oil analysis, including wear particle analysis, combining for effective identification of root cause and severity. PeakVue data from a bearing pedestal is shown in Figure 4, while Figure 5 reveals wear particle analysis results from the same machine. Together, these technologies indicate a problem of inadequate lubrication.

In conclusion, vibration and oil analysis each provide valuable insights into the health of machine components and lubricant systems. The combination of PeakVue, ferrous density and wear particle analysis produce three independent perspectives into the root cause and severity of an anomalous condition. Armed with this information, predictive maintenance technicians are able to accurately recommend appropriate corrective actions in order to improve maintenance and plant reliability.

Machinery Lubrication (10/2012)

A benchmark for the flexibility and adaptability of maintenance management system

Article below was extracted from:
http://www.reliableplant.com/Read/28900/condition-monitoring-saving

Although the content is suppose to focus on the advantages of good condition monitoring, but pay attention to how maintenance parameters such as criticality, priority and frequency changes dynamically. Notice also the flexibility and adaptability of their system to cope with the dynamism. That is what I would consider a World Class Maintenance practice.

Saving Time and Money with Condition Monitoring

Andy Gailey    

A recent acoustic emission (AE) study identified a potential critical bearing failure that became a planned preventive action for a leading food manufacturer. This also avoided considerable cost and unplanned downtime.
The acoustic emission equipment and the main tool used during the planned inspection routes were manufactured by Holroyd Instruments. This example will show the value of this type of equipment in avoiding a major unplanned event that could have had massive cost consequences for the business. Collateral damage to the associated equipment would have proved very costly, and the lead time to rebuild could have caused extensive downtime that would have meant disgruntled customers not being able to rely on stock availability.
The story began in April 2010 when some initial elevated readings were noted at two node points on a large step-down transfer gearbox that were sampled on a seven-day routine. The distress readings were elevated and triggered the alarm level. They were of concern and evident on subsequent inspections. The third elevated reading that was part of an upward trend instigated a planned work order in the computerized maintenance-management system (CMMS) to investigate and take further action. This equipment could not be taken out of service lightly, as it was at the time constrained by high production demands. Experience with a sister line’s previous planned bearing change also played an important part in the escalation of the risk.

The input side of a transfer gearbox is shown with the output bearing node point on the left-hand side.
On more detailed inspections, it was determined that the bearing with the highest distress was indeed the suspected output bearing. Audible clicks were loud and clear at the output end bearing. The two bearings at the node locations were on the drive line of the motor at the input and output ends of the transfer gearbox. The adjacent bearings on the large, helical step-down gear were still reading low and had no audible clicks. The engineering manager was advised that there was an anomaly on one of the input bearings, that the others were in good condition and that production could continue with targeted condition monitoring. The routine oil sampling was increased from monthly to every two weeks. AE inspections were increased, with spectrum readings now on a four-day cycle. This would give some comparative evidence when the new bearings were eventually fitted.
The planned change of the bearing set was arranged with the production planner, maintenance manager and product specialist. It became clear that the equipment would have to operate for at least another six months until it became available. Contingency plans were formulated for an emergency change if any of the AE readings or oil samples showed advances toward failure. Warnings were issued that this could occur rapidly if the bearing failed. A new bearing set was purchased, and a meeting was scheduled with the bearing manufacturer to examine the used bearings when they were eventually changed in early 2011.
The AE readings stayed at an elevated level during this long waiting phase, and oil sample results showed no elevated readings in the key elements associated with roller bearing failures. During the weeks before renewal, many spectra were taken from all points of the gearbox for future evaluation. This would rule out frequencies from the oil pump and other components around the assembly. An AE envelope spectrum graph before the bearing change is shown below.

As can be seen, there was something creating a spike at 73Hz, which happened to match the frequency of the bearing race. This provided a clue that there was a race surface defect of some kind and not an element breaking down or the cage disintegrating.
The bearing change finally took place, and the production plant was turned around within 12 hours so that the equipment did not incur any unplanned downtime. The used bearing set was returned with the transfer gearbox, and the two units were degreased. On first inspection, they both looked similar and in good order. The elements and cages were then dismantled from the outer and inner races, with care taken to keep them in order and in the correct aspect for reassembly later.
It became clear that on the suspect output bearing, a major spall on the inner race had developed, and every element was pitted with the debris that had emitted. At this point, a representative from the bearing manufacturer was invited to visit and examine the bearings. He concurred that the bearings had lasted very well considering the atmosphere and heat in which they had operated for almost 10 years. This would be considered an end-of-lifetime mode of failure. It may have lasted many more months or could have accelerated to failure within days or weeks. The photograph below is of the spall that measured approximately 10 mm in length and 2.5 mm wide.

Spall damage to the race is shown above with feathered edges and surface pitting in the loaded area of the spalling. Note the next layer of material on the right-hand side that would have given way.
When the remedial work had been completed, additional spectrum samples were recorded and monitored to learn more. Carpet noise levels were lower, and the decibel scale was a third of the previous graph example. The maximum peak was now less than 0.4 decibels, while the carpet level was less than 0.2 decibels.
In conclusion, the systems and tools relied on every day proved effective in capturing this anomaly before it turned into a major event. The key to this was the full involvement of engineering with operations to plan in the remedial work with as little disruption as possible.
Among the lessons learned were that the inspection frequencies at seven-day intervals were correct for this critical plant, the preventive action was started at the earliest opportunity, the equipment enabled the pinpointing of the bearing fault, the audio facility allowed a second reference that linked rpm with the audible clicks and that this all gave sufficient evidence for the planned work to commence at the earliest opportunity.
Root-cause analysis was carried out directly after the bearing change was completed to investigate any future recommendations for servicing this equipment. It was decided that as the bearings had reached their end-of-life cycle, there was no need to alter any future planned maintenance. Condition monitoring with AE had provided the confidence to pick up any anomalies at a very early stage in the curve.

LV Motor Maintenance

Motor maintenance are split into Predictive and Preventive.

Preventive:
1. Motor bearing lubrication. Typically motor frame size D180 inclusive and below are sealed for life bearings. Anything larger are greased as frequent as every 1000hrs. Refer to their respective manual for accurate greasing requirements.

2. Cooling fins and fans clean up. Frequency depends on operating environment. In generally clean environment, 4-8 yearly is acceptable. Increase accordingly to its operating environment. Note that in particularly post OH/ outage, plant will have bits and pieces of wrapping papers and plastics. That tend to get suck onto the ventilation guard of the fans blocking up cooling path. Thus I recommend performing this work post OH.

3. Alignment check should be perform on annual or 2 yearly basis particularly when new as the foundation sinks into place. It can later be extended to 4 yearly or as required based on VA.

4. VA is to be performed monthly, 3 monthly or 6 monthly depending on criticality of the motor. That should also pick up a range of other failure modes.

5. Terminals and glands is to be inspected 2 yearly in extreme processing operating environment, 5 or 10 yearly in a clean factory environment.

6. Thermography to be carried out monthly to check for anomalies and hot spot on motor and connections.

Predictive:
1. Measurement and trending of insulation resistance. For motor rated below 1000V, it is measured with a 500Vdc megger. (TECO manual extract)

2. In accordance with IEEE-43 clause 9.3 standards at the time of writing, the formula to calculate an acceptable limit is:

                      R => [ (Rated Voltage (V)/ 1000) + 1 ] x 10MegOhm

3. I recommend performing Motor Circuit Analysis (MCA) commencing with 2 years interval until sufficient data for trending (3-5 points), before extending to 4 years interval or more. It is a quick easy NDT that gives fairly good reliable indication of motor health whether it will survive another 2 or 4 years, or till your next available replacement opportunity.

4. On HV motors, I also recommend installing partial discharge monitoring equipment. Especially critical motors, it is better to go online PD monitoring. Trending data should indicate the deterioration rate of PD. Maintenance personnel should start taking concern and planning shall the PD deteriorates at a rate of double every 6 months or any anomalies.