Predictive maintenance equipment

Consider the possiblities for the next generation of predictive technologies and applications

Sometimes we take things for granted. Last fall, I participated in a roundtable on equipment reliability. As one of five experts, my introductory statement included a comment that predictive
maintenance was not working. Needless to say, this comment stirred quite a heated discussion.

Not only did the other members of the panel disagree, but also many of audience. Many in the audience assumed my comment was directed at the relatively stagnant progress of predictive maintenance systems over the last two or three years. Instead, my view is that we are not using the technology effectively. Current predictive maintenance technology, while not perfect, provides much of the data that is required to improve the performance of plant manufacturing and process systems.

If you stop and think about the advancements that predictive technology has made over the past ten years, you will discover that it boarders on miraculous. Vibration monitoring instrumentation for example has advanced from analog tape recorders and real-time analyzers that weighed hundreds of pounds to microprocessor-based instruments that can be held in one hand. It has not been that long ago that we were limited to time waveform data. The ability to transform time-domain data quickly into frequency-domain signatures is a relatively recent development.

In 1981, Technology for Energy Corporation introduced the very first microprocessor-based vibration meter. This 25-pound beast started the evolution of predictive technology. If you do not think that we have advanced, this first vibration meter could acquire 100-line frequency-domain signatures, store a maximum of 15 data points, and could not save time waveforms. Now, vibration meters weigh less than five pounds, can acquire both time- and frequency-domain signatures. They also have unlimited memory , and can capture 12,800 or more lines of resolution.

The impact of this new technology was that this smart meter permits plant personnel, with little or no training, to acquire and use vibration data. This development, along with its accompanying software program that automates data acquisition, management, and reporting, reduces the skill level required to acquire and analyze vibration data.

Today, smart meters are capable of handling most of the monitoring and analysis needed for optimization of mechanical and process systems. Unfortunately, we are still using this advanced technology in the same way as the 1981 smart meter. In most programs, we continue to limit the use to simple, rotating equipment.

If you stop and think about the advancements that predictive technology has made over the past ten years, you will discover that it boarders on miraculous.

Most of these data collectors also have the ability to record a tachometer input along with vibration data. This feature permits direct data acquisition and analysis of both rotating and non-rotating machinery. They can be used to analyze reciprocating, linear motion, and almost any dynamic machine or
system successfully.

We also continue to analyze vibration data in the same way as in 1981. Many programs are still limited solely to trending data. Trending raw vibration data has never been an effective tool. Variations caused by changes in load, speed, or process instability always have distorted the shape and amplitude of the vibration recorded in these trends. Unfortunately, this is one area where technology advancements have not kept pace with plant needs.

Even those programs that include analysis, the logic used today is identical to ten years ago. Most programs almost exclusively rely on failure mode charts or other types of simplified diagnostic tools. To put this limitation into perspective, I recently re-read one of my earlier books on the use of vibration as a predictive maintenance tool. The book, released in 1989 now reads like McGuffie's Primer--the first grade book for reading. While the techniques expressed in the book on predictive maintenance were cutting edge in 1989, they are totally inadequate for today's technology.

Thermography
The same level of advancements has been made in thermography. Ten years ago infrared cameras were extremely limited. They had the ability to capture black and white thermal images, but few could store them. Color scans were not even within the realm of possibility. These thermal cameras, like those made by Hughes, were huge, extremely expensive, and required bulky nitrogen tanks for cooling.

Now, high resolution, color cameras weigh almost nothing and are about the size of a small camcorder. Nitrogen cooling has gone the way of the dinosaur. Perhaps the greatest advancement in thermography is in the software that permits long-term storage and retrieval of full-color thermal images.

As in the case of vibration monitoring, the application of infrared scanning is still being used in exactly the same way it was ten years ago. The technology is almost exclusively limited to simple electrical equipment, like fuse boxes, switchgear, and similar equipment.

With current technology, infrared scanning can be used effectively for almost any system in which surface heat distribution is a valid indicator of operating condition. It has been successfully used for a variety of applications ranging from overloaded bearings to boiler condition monitoring. Frankly, we have not yet scratched the surface of the potential applications for this powerful tool.

Tribology
Tribology has not grown as fast as other predictive maintenance technologies. Ten years ago, analysis of oil and grease samples was strictly a manual, laboratory process. Trained technicians prepared each sample and use special laboratory equipment, like electron microscopes, for evaluation.

Currently, samples are still gathered in the same way that they have always been, but now there are computer-based instruments that can automate much of the analysis. These new instruments provide most of the sample preparation and analysis without the need for a trained laboratory technician. They operate on the same basic principles as the blood testing instruments used in hospitals. A sample of oil is placed in a test tube that is inserted into the analyzer. The analyzer centrifuges the sample, analyzes the data, and generates a report.

The future of predictive maintenance appears to be good, but one cannot expect to reap the full benefits unless they are willing and able to use the technology correctly.

Each of the major technologies--that is, vibration, thermography, and tribology--must function as a discrete, stand-alone system. If you use all three in your plant, this restriction forces the use of three separate software programs and databases.

While this limitation does not preclude reliability analysts from using all of the data, it does substantially increase the difficulty. Data must be exported to a common program that can handle the three data formats or the analyst must print hardcopies and manually compare data.

Hopefully, vendors of predictive systems will resolve this problem. If they would adopt common, open-architecture database and operating system format, this problem could be resolved easily. This change also would permit direct interface or integration with other plant computer systems.

Now, the technology does not permit direct interface with critical systems, such as CMMS. As a result, the recommendations generated by the predictive maintenance teams must be manually converted into a work order and entered into the CMMS.

Open architecture would eliminate the need for this manual interface. Work orders could be transferred into the CMMS automatically. The converse is also true, maintenance histories, operating parameters and a variety of other data that the analyst needs for accurate analysis readily would be available without the necessity of accessing another plant computer system. Technology supports this much-needed upgrade, but vendors must be willing to standardize database and operating system architecture.

While it cannot be classified as a technology advance, the recent trend of mergers, acquisitions, and joint ventures with the predictive vendor community may help take these technologies to the next level of usefulness. Ten years ago, there where hundreds of small, cash-starved companies that provided numerous approaches to predictive maintenance.

Now, the number has been reduced to a handful of large, highly profitable companies. Coupled with the marriage of these predictive companies and large instrumentation and control companies should provide the momentum that creates the next generation of technology.

The future
Yes, technology advances in predictive maintenance systems have slowed over the past few years. But, during its cycle of advancement since the first microprocessor-based system was introduced in 1981, it evolved into perhaps the most power plant optimization tools. With these systems, we can monitor and evaluate the operating condition of almost any critical production and manufacturing system cost-effectively.

However, to gain maximum benefits, these tools must be fully exploited. The future of predictive maintenance appears to be good, but one cannot expect to reap the full benefits unless they are willing and able to use the technology correctly.

We must keep pace with technology and learn how to use these tools effectively. The quality and quantity of training that a predictive analyst or reliability engineer receives must keep pace.

Now, as it was ten years ago, analysts receive on average of ten days of vendor training in each of the technologies. Even though I am sure that they would disagree, this is one area in which predictive maintenance system vendors have failed to keep pace with their own technology. Little change has taken place in their training courses. If you were to compare course content from ten years ago to their current versions, little has changed. Most still teach analysts to the same overly simplified methods to interpret data. Like McGuffie's reader, it is still valid, but is also outdated. We cannot expect these employees to effectively use these tools without adequate training. The tools are there, are you using them properly?