by Matt Migliore
The words "maintenance" and "headache" are oft used in the same sentence, typically one after the other, for good reason. Maintenance, after all, is indeed, generally speaking, a headache. However, as monitoring technologies continue to evolve, and plant- and field-based instruments and systems grow more capable of providing diagnostic information about their operating conditions, the worst-case maintenance scenario — the dreaded "unplanned downtime" — is becoming less and less of a concern. In fact, with the price and size of sensors continuing to fall, and processing power on the rise, predictive maintenance technologies are providing more robust failure-mode monitoring than ever before.
Types of Maintenance
To fully understand the value of predictive maintenance for certain applications, users must first grasp how maintenance procedures are classified and how they differ from each other. In general terms, maintenance can be classified into three different categories: corrective maintenance, preventive maintenance, and predictive maintenance.
Corrective maintenance is generally characterized by a system that is run to failure before it is taken off line for maintenance. This form of maintenance, sometimes referred to as "repair" maintenance, is the most inefficient method of maintaining systems. It generally results in unplanned downtime for the process and requires the system to be pulled out of service without prior preparation or planning. Very few modern-day processes would be able to survive if operated on a corrective maintenance program. Further, depending on the process conditions of the application, a corrective maintenance program may compromise the safety of plant personnel and the environment.
Preventive maintenance has been, and remains, standard practice for many fluid handling systems. Preventive maintenance, in general terms, is any time- or count-based maintenance procedure. The aim of a preventive maintenance program is to maintain the operation of equipment without incurring unplanned downtime. Preventive maintenance activities may include tests, measurements, adjustments, and/or parts replacements — all of which would be performed on a scheduled and/or planned basis in an effort to ensure the operation of the process without unplanned downtime. While a more effective approach than corrective maintenance, the key limitation of preventive maintenance is that it occasionally results in maintenance being performed on systems that may not actually require maintenance. Thus, preventive maintenance can be labor intensive, ineffective in identifying problems that develop between scheduled inspections, and costly.
Predictive maintenance is a data-based method of maintaining the operation of systems. So, instead of waiting for a system to fail or performing maintenance procedures on a time-based interval, predictive maintenance programs use key pieces of process information in an effort to more precisely determine when and what process systems actually require maintenance. This approach, in theory, offers a more cost-effective and efficient platform for systems maintenance, as it strives to perform maintenance only when warranted and when maintenance activity is most cost-effective and before equipment falls out of its optimum performance range. However, diagnostic information is required for users to employ predictive maintenance, which generally calls for the implementation of advanced technology. As such, end-users would be wise to carefully consider the cost-benefit, as well as devise an implementation strategy, prior to investing in predictive maintenance technology.
What’s Your Maintenance Makeup?
Given the upside potential for more effective and efficient maintenance handling, predictive capability may be something end-users want to consider as part of the decision-making process when evaluating new equipment. For existing and/or lower-end systems, however, predictive maintenance generally won’t be a realistic option.
"The problem with predictive maintenance on low-cost equipment is that there is often no automatic way of extracting data or trends over time," says Andy Marsden, Americas services manager for Thermo Fisher Scientific (www.thermo.com), "In contrast, more expensive equipment may have remote diagnostic capabilities in which the user not only can extract data, but can often remotely take corrective action."
For older systems where process data is not readily available, Marsden says preventive maintenance is likely going to be the best option. For an older generation ultrasonic flow measurement system, for example, Marsden says a simple and effective preventive maintenance practice would be the regular removal, cleaning and lubrication of the transducers that attach to the process pipe.
On the other hand, according to Marsden, newer systems on the higher end of the spectrum are not only being designed with predictive maintenance capability, but also with fewer moving parts in an effort to increase the mean time between failure (MTBF). He says these two trends together are helping to reduce the amount of maintenance required.
Whatever the approach, be it predictive or preventive maintenance, Marsden says the end-user should strive to prevent and/or mitigate the consequences of a system failure. He says the possibility of human failure must also be taken into consideration. "Competent and trained personnel should be utilized and, if available, OEM training is beneficial," says Marsden. "Existing maintenance practices that have not worked well in the past should be avoided."
Predictive Maintenance In Action
In general terms, while preventive maintenance is based on a routine, predictive maintenance is based on information. "Predictive maintenance watches the equipment and quantifies whether the equipment is healthy or not and determines if, and what, needs to be done to bring the equipment back to a normal state," says Sandro Esposito, global product marketing manager – Smart Products for Dresser Masoneilan (www.masoneilan.com).
Esposito cites a smart digital positioner on a control valve as a prime example of predictive maintenance in a real-world setting. He says current-generation smart positioners are capable of monitoring such key control valve performance indicators as the ability of the valve to yield the appropriate position; the valve’s ability to prevent fluid from leaking to atmosphere; and the valve’s ability to effectively throttle and shut off the fluid. Using inherent sensors to monitor key performance indicators, Esposito says smart positioners can determine if the process is being operated in the prime operating range and, if it is not, send an alarm to a resource management system that provides determinative information to help the user identify the root cause of the problem.
Esposito says smart sensors combined with monitoring software can give end-users a real-time view of the process and provide valuable information that is truly helpful in identifying performance degradation and pinpointing when equipment failure may occur.
One specific application where smart positioners have proven to be valuable, according to Esposito, is in partial-stroke testing of emergency shutdown valves (ESDs). In such a scenario, Esposito says smart positioners not only ensure the ESD valve will perform its duty when needed, but also provide a ready source of archived data when required for regulatory auditing purposes.
Predictive Maintenance Interoperability
When developing a predictive maintenance program, integration is a key consideration. A device that has the ability to monitor performance variables and capture information must also be able to easily share that information with a control system in order to be truly useful to the end-users. According to Esposito, many early generation predictive maintenance control systems were proprietary in nature. This created a situation where, as Esposito describes it, "users had to struggle with islands of automation to get precious data from the field."
Today, Esposito says most host system providers are providing solutions that have the ability to read data from a variety of different field devices. Further, he says the movement toward open standards for wireless systems is allowing users to capture more data without adding complexity to their infrastructure.
On the device side, Esposito says the key evolution in terms of integrating diagnostics has been in the emergence of Device Type Manager (DTM) technology, which is essentially a software device driver for interfacing with the host system. According to Esposito, DTM gives end-users plug-and-play capability when implementing new devices into their predictive maintenance architecture. "It opens up all the possibilities of the device in an open-frame architecture," says Esposito. Ultimately, he says DTM enables the user to choose the device application without worrying about the integration of predictive diagnostics. As such, Esposito says end-users should be looking for DTM-based devices to ensure interoperability into the future.
Predicting the Future
The pinnacle of predictive maintenance is 100 percent failure-mode monitoring. However, in order to provide this level of monitoring with today’s generation of technology, the devices would either be far too expensive for practical application or the power requirement of the device would be too high or the devices would be weighed down with so many sensors that the diagnostics would disrupt the process (or all of the above).
The current state of predictive diagnostics is that smart devices with inherent sensors are available. These devices are available in a form that is relatively integratable with open standards-based host systems. Using the smart devices and the software together, end-users can gain added insight on the operating conditions of the process on a real-time basis. With this information, end-users can perform root-cause analysis of possible failure modes. In turn, this information can help more effectively and efficiently identify where failures may occur based on process conditions. This is not to say that today’s predictive maintenance systems can identify with utmost precision where and when a failure may occur.
"[Today’s predictive maintenance technology] can identify impending failures of various sub-assemblies in a control valve," says Esposito. "But to be able to say precisely that the third packing row around the stem housing is leaking, that’s kind of pushing the envelope at this point."
However, Esposito says predictive maintenance will become more determinative as sensors continue to get smaller and less power hungry so they can be integrated and packaged with smart positioners. "Energy’s always the constraining element to work around without adding more wires to the field," says Esposito. "Once you have the power available or can harvest energy, the sensors are available, and they can be packaged in a way to diagnose close to 100 percent of possible failure modes."
More importantly, Esposito says he sees the linkage of information throughout the lifecycle of equipment as a key enabling trend for predictive maintenance technology. He says he envisions a "Valve-Opedia," similar to the popular online encyclopedia, Wikipedia, which would provide a repository of fault models for different pieces of equipment. In turn, end-users would employ such a repository to identify the possible root causes of their own equipment failures. Using RFID (radio frequency identification) tags, Esposito says equipment failure information could be automatically logged in a central database to be shared with other users around the world.
"Imagine a technician at a plant in Los Angeles fixes a valve; that equipment is fixed and the root cause is identified because of the fix," says Esposito. "At the same time, a person in Singapore is experiencing similar symptoms, but the root cause is different. All of this information is gathered and recorded in a repository where another user can leverage it to help identify what the possible root causes of a failure are," he says. "In essence, it is like linking the life experiences of all users with the data from smart devices."
While Esposito says he sees a system of this sort being employed on a manufacturer-to-manufacturer basis, he also see the possibility of an ISA- or IEC-based standards movement that would establish some common sets of symptoms and remedies for control valve failures, for example.
Meanwhile, Thermo Fisher’s Marsden sees modular systems design as a key element of predictive maintenance in the future. He says that by establishing commonality of parts between systems, end-users will be able to more efficiently perform root-cause analysis of failure modes, as well as eliminate the need to keep many parts on hand to maintain various instruments within a given process. "As instrumentation evolves and new technologies become available, manufacturers [will] seek to eliminate failure and minimize maintenance requirements," he says "Remote capabilities such as alert level alarms, automatic calibration routines, built-in redundancy and plug-and-play modules are examples."
Matt Migliore is the editor of Flow Control magazine. He can be reached at [email protected].
