- Predictive technologies, whether hardware or software, vary in price and analytics, but which tool to use typically depends on the type of failure being looked for and then criticality of the equipment.
- The first step toward embracing vibration monitoring and analysis in an appropriate way is to understand how to assign vibration responsibilities within the plant.
- To analyze vibration data requires software, hardware, training, significant asset information infrastructure, and a disciplined schedule.
Vibration analysis is only as good as the data you collect and what you do to correct problems. Just as a healthcare professional might use a stethoscope or an electrocardiograph to measure heart health, maintenance and reliability professionals have wide range of tools at their disposal. These technologies, whether hardware or software, vary in price and analytics, but which tool to use typically depends on the type of failure being looked for and then criticality of the equipment.
“It takes essentially the same amount of money and effort to collect poor data as it does to collect quality data,” explains Sam McNair, P.E., CMRP, senior consultant at Life Cycle Engineering (LCE, www.lce.com), headquartered in Charleston, South Carolina. “Taking RMS vibration levels at one or two frequencies, such as with a handheld vibration pen, is not vibration analysis. To equate it to full-spectrum vibration analysis is to equate a doctor with a stethoscope to one performing an EKG. Both have their proper place in the scheme of things, but they are not the same by any means. Just as an EKG costs more to use, requires more knowledge to analyze, and gives more accurate and definitive results, so does performing full-spectrum vibration analysis.”
Figure 1. Once the site’s machine population is categorized by the consequence of its failure, the overall level of risk is more clearly defined and monitoring and analysis can be assigned from a financial perspective. (Source: Robert Perez)
The quality or degree of the vibration testing is dictated by risk profile or criticality of the machinery. “Reviewing the overall risk profile of a site should always be the starting point for justification of a machinery condition monitoring program,” says Robert X. Perez, author of “Is My Machine OK?” and staff reliability engineer at Enterprise Products in San Antonio, Texas. “The machinery population can be broken up into the broad categories. Low consequence includes small to medium horsepower process machines with low levels of secondary consequences if they fail unexpectedly. This group usually represents the 90% or more of the site’s machinery population. High consequence includes process machines associated with high levels of secondary consequences. If these machines fail unexpectedly, bad things, such as fires, product releases, production outages, and costly secondary mechanical damage, happen. This group can represent about 10% or less of the sites’ population (Figure 1).”
Once the site’s machine population is categorized in this way, the overall level of risk is more clearly defined, explains Perez. If the site is composed only of smaller, low-risk machines, then analysis performed by a mechanic or operator using a vibration data collector or vibration meter is adequate. This level of analysis is limited to trending and simple spectral analysis.
However, if the site comprises a large population of high-consequence machines, then Perez recommends analysis typically performed by a certified vibration analyst, using advanced vibration data collection and analysis hardware and software. This level of analysis involves spectral analysis, phase comparisons, Bode plots, and correlation analysis, he explains.
Mission-critical, complex machines with many variables require a vibration expert to watch them closely, agrees John Bernet, product & application specialist at Fluke (www.fluke.com). “These machines impact plant production,” he stresses. “Companies with this application have no option but to hire a reliability team or outsource to a service consultant. Industries that find they need to do this are petrochemical, pulp and paper, power and others that can’t afford to have any stoppage in production. The maintenance staff doesn’t have the resources, training required, or time needed to trend and analyze the hundreds of rotating machines not supported by the reliability team. The maintenance staff needs easy-to-use vibration testers where extensive setup, trending, analysis, and on-site experts are not needed to get machine-condition answers.”
Whose vibration is it anyway?
The first step toward embracing vibration monitoring and analysis in an appropriate way is to understand how to assign vibration responsibilities within the plant. “If your organization is large enough to have a dedicated reliability team, then vibration monitoring and analysis should be part of that team,” recommends Wayne Vaughn, P.E., CMRP, principal consultant at Vesta Partners (www.vestapartners.com) in Stamford, Connecticut. “It’s valuable to cross-train a person or more on monitoring who work in the maintenance group, both for scheduling flexibility and to develop the career path of craftspeople. If your site doesn’t have a reliability group and is small, then I like the idea of a partnership with outside help for analysis and initial setup, but using maintenance personnel to take the readings and do quick reviews of data.”
If an organization is starting a vibration analysis program, then chances are high that the maintenance department will be in reactive mode, explains Jason Tranter, managing director and founder of Mobius Institute (www.mobiusinstitute.com). “They will be in fire-fighting mode, and therefore their priorities will be associated with fixing today's breakdowns,” he says. “If the vibration analysts don't have 100% focus on the vibration monitoring program, then they will be constantly distracted; their data collection routes will be delayed more and more and thus the program will not be very effective.”
If the organization has a reliability group, then hopefully that means that the organization is thinking about improving the future, continues Tranter. “Condition-based maintenance is all about fixing tomorrow's problems so they don't become today's problems,” he explains. “The maintenance department will hopefully be better able to deal with a condition monitoring program if they believe in reliability improvement. But, if the maintenance department is in fire-fighting mode, it isn’t a good place for a condition monitoring team, unless they have the independence to do the job properly and the support to have their recommendations acted upon.”
In an ideal world, the vibration analysts would have enough time to test all machines with sufficient frequency to detect faults at the earliest stage and monitor the progression of the fault. Spending a lot of time with the machines has numerous benefits, says Tranter. “But, if the budget doesn’t support this effort, then a very good solution is to involve operators in the program,” he suggests. “Operators have a much better connection with the machine and the process, and it’s in their interests to detect faults before they become potentially dangerous. If the operators detect a change in vibration by using a simple meter or ultrasound, then they can call for the vibration analyst to come and perform a complete analysis.”
Involving the people closest to the machinery is a prescription for machine health, agrees Mickey Harp, vibration application engineer at Ludeca (www.ludeca.com). “Many organization stress ownership of the machines and the process across the entire staff within a plant,” he says. “The best vibration program will fail if not enough time and effort are given for the team or group to accomplish their goals. If the vibration program is contained in-house, it gives the organization the ability to learn and trend the machines. It allows for all teams to identify issues, find root causes, and schedule repairs. It also allows the organization to allow their own people to better understand their machines and process.”
Another way to divide the responsibilities is to think in terms of the equipment, with maintenance staff handling the individual asset and reliability professionals looking at the system all the way upstream.
In a plant with a mature PdM program, the maintenance staff might handle specific pieces of machinery, while the reliability team might analyze asset health from more of a systems approach. (Source: Mobius Institute)
“In a mature program, the maintenance staff probably is more focused on the equipment that is in the plant, explains Bill Slonaker, business manager at Mobius Institute. “As they progress, they will probably institute more of the precision maintenance techniques such as precision shaft alignment and dynamic balancing to wring out additional savings. They will also be getting the maximum life out of rolling element bearings by closely monitoring and lubricating the bearings properly, doing pre- and post-repair tests. They may also take the next step and work with bearing manufacturers to find better bearings, couplings, and seals for their particular applications.”
The reliability team, on the other hand, is probably going to be looking at things in a more holistic approach to improve reliability and reduce costs throughout the plant, explains Slonaker. “For example, they should be looking upstream at the applications and types of machines that are going into the plant, doing acceptance testing on them at the time of manufacture and again pre- and post-installation to make sure that bearings have not been damaged in shipment, that machine trains are properly installed, aligned and balanced before going into production,” he explains (Figure 2).
Assigning the vibration duties is similar to planning any other predictive maintenance (PdM) strategy, and it includes three minimal necessary phases — collection, analysis, and corrective action, says LCE’s McNair. “Concerning the collection of data, it doesn’t matter who does this, as long as they have the proper data collection tools, the correct location and number of sampling points, and use the proper technique,” he says. “They either use the right tools, placement, and method, or they don't. Anything less than correct will give partial or erroneous data. Anyone who has the proper training, tools, and engineering support is able to collect the data.”
Collecting data on a consistent schedule depends upon the discipline of the organization, explains McNair. When maintenance personnel are assigned to any scheduled PM or PdM task and an emergency occurs or backlog gets too big, the first thing that gets dropped is the PM or PdM. “That is unacceptable,” he says. “However if the organization is disciplined and has high PM compliance, then it doesn’t matter to which part of the organization the properly equipped, trained, and supported data taker belongs. Typically, in the real world where most companies are pretty reactive, having the PdM tasks administered by the reliability group isolates them somewhat from the world of emergencies, and therefore they stand a better chance of getting them done.”
To analyze vibration data requires software, hardware, training, significant asset information infrastructure, and a disciplined schedule. And it’s usually done by an engineer or equivalent by experience, says McNair. “Most maintenance organizations tend to use their maintenance engineering talent to assist them to figure out emergency repairs, design work aids to support immediate maintenance needs, buy special tools, chase parts, plan jobs, do small projects, and sometimes to administer programs such as welder certification,” he explains. “The work tends to be highly variable, often urgent, and relatively short-term-focused. All of this detracts from the highly regular, repeatable work habits and long training investment needed to support a good PdM program of any kind. The PdM work is generally given to the reliability group for the same reasons the data-collection part is.”
What do you do with a good analysis once you have it? “Regardless of which department originates a high-quality vibration analysis, that is only half the battle,” says McNair. “Even programs that are highly effective at discovering discrepancies are rarely more than 30% effective at getting the corrective action implemented in an appropriate time and manner.” The appropriate time is when the issue is on the left-hand side of the P-F curve, and the appropriate manner is planned, scheduled well in advance, and executed precisely and with bare minimum operational impact, he says.
Let’s work together
Reliability-centered maintenance programs are most effective and most profitable when a variety of appropriate PdM technologies and tools are used to complement one another. And vibration analysis and ultrasound are as complementary as two sides of the same coin.
“When performing PdM, you should always verify your finding, preferably using a different technology than used for initially detecting it,” explains LCE’s McNair. “You should consider your PdM choices based upon the primary failure modes you are protecting from, the window of detection required, the ability to independently confirm, and the maximum revenue impact at an acceptable ROI.”
One of the first characteristics of an effective PM or PdM is that it is failure-based, says McNair. “To have an effective program, you need to choose the technology and sampling interval that will, with a high degree of repeatability, detect the specific failure modes you are looking for,” he explains. “And you must be able to detect them in sufficient time to plan and execute a timely corrective action. Sampling interval is a function of the window of detection (WOD) for the technology applied. The WOD is the time interval between when you are able to detect a defect and the terminal failure of the asset, and, to be effective, the sampling interval is generally one-third or less of WOD. This only works for degradation, or trendable, types of failure modes.”
If ultrasound works and gives you an adequate WOD, then use it where it’s effective, says McNair. “If a less-expensive technology results in adequate detection and the lower cost enables you to apply it to a wider range of assets with an acceptable ROI, then do so,” he explains. “Other than specific cases of safety and compliance, PdM is not about what technology you use; it is all about using the technology for making money. If all your budget will permit is the least expensive unit out there, you can still have a good ROI. But don’t expect world-class results.”
Repeatability of measurements is the key distinction. Ultrasound is useful in certain situations where it is extremely difficult to acquire vibration data, but there are limitations, explains Mobius’ Slonaker. “It can be used to quickly scan a conveyor belt with rollers and rolling element bearings,” he says. “A bad bearing will be considerably louder and easy to detect. For large machines, the transmission of high frequency vibration is greatly attenuated through the machine’s structure and might be difficult to detect without making contact with the machine with a sensor. Ultrasound is not as repeatable as vibration data unless you have a fixed measurement location and can control some aspects of the surrounding environment, so trying to use ultrasound as a quick way to walk through a plant instead of taking vibration data is going to incorporate some risk. An equally failing bearing in a conveyor would be easy to detect but wouldn’t be noticed in a massive machine using ultrasound detection.”
Vibration measurement systems are often set up with installed sensors or at least a mounting pad that the sensor is attached to for data collection. “This provides a high degree of repeatability,” says Slonaker. “High-frequency sound is fairly directional, but that decreases as the frequency goes down. If you weren’t in exactly the same location each time, the sound levels could vary considerably. Most people agree that ultrasound is a good complement to vibration data collection and analysis. It can also be used for a wide range of applications where vibration wouldn’t be effective, such as detecting compressed air leaks in piping.”
Ultrasound is a useful monitoring tool, but isn’t effective in detecting all types of machinery malfunctions, warns Enterprise’s Perez. “For example, ultrasonic methodology is capable of detecting failing rolling element bearings and over- and under-lubrication conditions,” he says. “However, ultrasonic methodology is ineffective in detecting an early vibration issue in a fluid film bearing. It’s important to match the right monitoring tool with the failure modes you expect to see. Before starting an ultrasonic monitoring program, list your most important failure modes and determine which of those are best detected with ultrasonic methods. The best overall machinery monitoring program is one that utilizes multiple, integrated monitoring technologies that are well-suited to detect your expected failures modes.”