The ideal maintenance reliability strategy is not a one-size-fits-all approach. To have the greatest impact, you must analyze each asset and chart the path that will produce the best results for your organization. Maintenance practitioners are turning increasingly towards predictive maintenance (PdM) and condition monitoring (CdM) techniques to help them achieve success on this journey.
To effectively navigate the maintenance landscape, we first need to use the correct language. Predictive maintenance (PdM) and condition monitoring (CdM) are maintenance techniques and tools within a condition-based maintenance (CBM) strategy. These techniques and tools are used to detect symptoms of potential failures of equipment; they are not strategies themselves.
Condition-based maintenance gathers asset condition information from predictive maintenance and condition monitoring techniques to determine the optimal moment to perform maintenance for mitigating conditions that lead to failure.
The goal of condition-based maintenance is to ensure each asset is available when called upon, and to avoid performing maintenance too soon or too late on the P-F curve. The idea is to move left on the P-F curve by detecting potential failures sooner rather than later.
Therein lies the challenge. Finding the ideal moment for maintenance has been confounding maintenance practitioners for decades.
Until recent years, maintenance strategies were primarily a combination of time-directed and run-to-failure. Maintenance decisions were largely based on human observational-based techniques, and empirically capturing asset health information after the fact.
Facilities would wait until a motor, pump, or conveyor failed before acting. At the time, it was about the only choice they had because the tools and data were not readily available to reliably find adverse conditions quickly enough to avoid them.
The rise of automation and digital data aligned with Industry 3.0 provided maintenance crews with technological observation methods to help them anticipate conditions that might result in asset failure. Enter the world of computerized maintenance management system (CMMS) software. This allowed them to establish a digitized time-based preventive maintenance (PM) strategy for each asset.
Maintenance practitioners would regularly measure key indicators on critical equipment to discover the potential failures, and then schedule downtime to repair or replace components on a set schedule (often specified by the original equipment manufacturer), regardless of the actual condition of the asset. This generally reduced the failure rate but was labor intensive and costly due to added downtime and sometimes unnecessary component replacement.
Now, as we move forward with Industry 4.0 and smart autonomous systems within the Industrial Internet of Things (IIoT), many facilities are using real-time (or near real-time) asset condition data to optimize maintenance activities.
This condition-based maintenance strategy involves capturing data from a variety of sources, including sensors, handheld diagnostic tools, SCADA, and other data acquisition systems. The data is aggregated, analyzed, and turned into actionable intelligence through a CMMS, which then directs maintenance resources where they’re most needed. So rather than putting out fires, maintenance practitioners can work on digitalizing processes.
Predictive maintenance and condition monitoring are similar in that they both use technological observation methods to capture asset condition data. Also, they are both complete participants within those methods in that they immerse themselves completely in the environment.
The differences between predictive maintenance and condition monitoring are minimal; they are actually complementary. So, rather than choosing one or the other, most maintenance programs combine them to provide a more comprehensive maintenance reliability picture.
Together, predictive maintenance and condition monitoring techniques provide the means to capture and record empirical data you need to make better decisions about where to apply maintenance resources to gain the best return.
Predictive maintenance tools and techniques employ ultra-sensory technology, such as thermography, vibration analysis, oil analysis, and ultrasonic sound measurement, to take a snapshot-in-time of asset health. Because you’re capturing a brief period of time, it is important to also record the operational context of the readings, such as how long the component has been running, and other operational data that might affect (or better assess) its performance health.
One caveat about predictive maintenance is that it does not predict the longevity or the exact functional failure point of an asset. Rather, it provides technological observation data to help you more accurately, effectively, and fiscally responsibly schedule maintenance on an asset to provide a higher degree of availability and capacity assurance.
Condition monitoring tools and techniques also provide technological observational data; but rather than just capturing a snapshot, they collect ongoing asset health data on a continuous basis through sensors and other data acquisition systems.
Condition monitoring techniques are ideal for assets that are hard to access or are in hazardous environments. Installing sensors to detect excessive bearing vibration or overheating, or motor power quality issues, provides critical asset condition information without putting employees at risk or requiring production shutdowns.
Depending on the system, data can be collected at intervals ranging from seconds, to hours, to days. The rise of the IIoT, advances in sensor technology, and reductions in cost are causing data volume to grow at an exponential rate. Applying artificial intelligence and machine learning to this increasing volume of trending condition data is helping maintenance practitioners more accurately identify the optimal point on the P-F curve to apply preventive measures.
Random failures are classified as such because they result from unknown causes. One of the areas where predictive maintenance and condition monitoring tools and techniques have the greatest impact is in identifying the root causes of random failures.
This means maintenance practitioners have to look at everything from a technological observation point such as vibration, thermography, and power quality, among others. Performing all those inspections using repetitive time-directed maintenance activities can be a huge drain on time and resources.
Now, predictive maintenance and condition monitoring techniques can often identify conditions leading to random failures, classify them in the proper failure curve, more effectively resolve the cause, and then apply the correct mitigating maintenance strategy.
By helping maintenance practitioners in this way, these techniques can help eliminate many failures formerly classified as random. Simply stated, we don’t know what we don’t know. Using technological observation data derived from predictive maintenance and condition monitoring techniques helps to make the unknown known so we can move the P-F curve further right and thus gain more operational time between failures.
The progression of maintenance strategies tends to move from time-directed, to condition-directed, to (eventually) data-directed maintenance. But not every asset needs to follow that route.
The truth is, most facilities use a combination of these maintenance strategies based on the criticality and the cost of the asset in question. Although the trend is toward condition-based and data-directed maintenance strategies, many organizations develop a hybrid preventive action plan that includes everything from time-directed to data-directed maintenance procedures. The key is finding the right mix.
Although time-directed maintenance has fallen out of favor in recent years, there are cases where the scheduled discard or restoration tasks associated with traditional preventive maintenance are most effective. For mission-critical applications, where a failure could cause widespread physical or environmental damage, such as in oil and gas processing or nuclear power plants, combining condition-based with data-directed maintenance strategies is a preferred approach.
After the criticality of the asset, the budget and the availability of human and technology resources play essential roles in deciding which maintenance technique to use.
You don’t want to “chase a dime with a dollar” by deploying predictive maintenance and condition monitoring techniques when standard time-directed—or even run-to-failure—approaches are more effective. In some cases, it is cheaper and faster to replace an asset than to sensor it, or to conduct regular PMs on it.
Equally important is the time sensitivity and collection frequency of the asset condition data. For example, a bearing failure mode with a P-F curve of six months may not need condition data to be collected every minute, hour, or second. Perhaps weekly is fine. On the other hand, that same bearing in a mission-critical operation, such as a nuclear power plant, may need constant monitoring.
Key Considerations in Designing an Effective Maintenance Strategy
Creating the best maintenance strategy for your organization is both a financial and logistical decision process. Here are some key factors to keep in mind:
Decide whether it is cheaper and quicker to replace an asset every so often than to send a maintenance practitioner to check on it at regular intervals.
Determine which assets are so critical to keeping things moving that the cost of adding condition monitoring is worth it, no matter what the cost.
Remember there is a difference between strategies and techniques and tools:
Condition-based maintenance (CBM), reliability centered maintenance (RCM), reliability-based maintenance (RBM), design out maintenance (DOM), and even run-to-failure (RTF) are all strategies.
Predictive maintenance and condition monitoring describe techniques and tools used within the condition-based maintenance strategy.
Time-directed, condition-directed, and data-directed are procedural classifications. These classify the tasks maintenance practitioners deploy in a preventive action plan to eliminate and/or mitigate failure modes.
Reliability-based maintenance tools, such as failure mode and effects analysis (FMEA) or failure mode, effects, and criticality analysis (FMECA), help determine the most prudent maintenance strategies and procedural steps to identify, mitigate, or eliminate potential causes of failure.
In the end, two ROIs must be considered. One is the traditional return on investment from increasing efficiency with existing resources. The other is your return on integrity. As a maintenance practitioner, your integrity—and that of your entire organization—rides on maintaining asset availability and capacity assurance. The maintenance tools, techniques, and strategies you choose must support that goal.
About the author
Gregory Perry, CMRP, CRL, is a Senior Capacity Assurance Consultant at Fluke Reliability. He is a Certified Reliability Leader with nearly two decades of experience in maintenance and operational best practices and has a broad base of experiences in MRO and storerooms, world-class maintenance principles, and world-class CMMS consultation and leadership. In addition to delivering implementation and consultative services to clients, Perry also presents maintenance best practice sessions at leading industry conferences and has authored several online best practice webinars.