Predictive maintenance programs have become the standard maintenance strategy for achieving optimal reliability and efficiency throughout the manufacturing world. A successful PdM strategy requires incorporating and coordinating numerous electrical and mechanical testing technologies.
Some companies outsource the testing to qualified consultants; other opt to purchase their own test equipment. The selling point that “the equipment will make you an expert” by performing the diagnostic dirty work and providing you with a clear pass/fail response as to an asset’s health is attractive and appealing. So predictive maintenance test equipment is purchased with the buyers buying into the promise that it will eliminate unplanned downtime and the associated costs, but little thought is put into the requirements of actually gathering data and turning it into useful statistics. In many cases, that means adding use of the technology to the responsibilities of a technician who already has a full workload.
Technology advancements have made it possible for instrument manufacturers to create extremely sophisticated test equipment with computers that make data collection and trending easier and more reliable. However, the reality is that only experience can make you an expert. It takes technicians years to be able to successfully read between the lines, and getting a green flag does not always mean all is well.
Far too often, subtle indicators are missed because the collected data falls within the expected parameters and thus isn’t flagged, even when a real problem is developing. This causes two major concerns. One, the developing issue does not get the attention it needs, so the problem continues to worsen—defeating the real purpose of predictive maintenance—and two, management begins to lose trust in the technology and eventually denies needed support when failures continue to occur.
Vibration analysis is a prime example of the need for experience to ensure success. Modern equipment is user-friendly, and a technician can quickly learn how to gather good data, but it will take three to five years of continuous involvement and training for an analyst to get to a comfortable level of proficiency.
Electrical testing, while not quite as complex as vibration analysis, still requires knowing how to collect good data and what to be aware of within the information. Take, for example, a case in which a company had purchased cutting-edge test equipment and maintenance personnel had been collecting data routinely. The maintenance team’s static testing had indicated that arcing was occurring within the turn insulation of a critical machine, but the level of arcing was well below the pass-fail threshold, so the conclusion was that all was good.
Within a few months, the motor suffered a catastrophic failure during a busy season, causing a huge loss in production and resulting in substantial emergency repair costs. The equipment did its job as designed, but because the results summary showed all green flags, no action was taken. If the technician had known what the data was telling him, perhaps the call would have been made to take this motor out of service, allowing the company to avoid expensive losses and repair costs.
While it is true that the state-of-the-art equipment available today takes most of the guesswork out of predictive maintenance, the fact remains that numerous minor issues that could result in a major failure will be overlooked by the inexperienced eye. Don’t fall into the trap: You will not instantly become an expert, and the equipment alone will not eliminate downtime. Be ready to either hire experienced personnel to manage the process, outsource the work to a qualified industrial solution provider, or accept the fact the program will only be partially successful in its early stages.
Predictive maintenance programs have proved to be the most economical maintenance strategy, and equipment available today makes collecting useful, valuable data easy and quick. However, a successful PdM program depends on experienced diagnostic interpretation of the data.