Predicting the repair needs of critical components in lathes, mills, turbines, fans, drills, pumps, and any other rotating device, allows today’s MRO executives to control the destiny of their plant’s operations by preventing unexpected shutdowns in the first place. This ability to plan for the future is no longer the domain of Fortune 500 companies, as PdM-software’s power to deliver answers, not merely data, has even trickled down to small and mid-sized businesses. However, it wasn’t always that way.
It wasn’t that long ago that plant engineers relied on run-to-fail (RTF) maintenance schedules—a method that almost guaranteed unscheduled process-line shutdowns. Hardly satisfactory, the maintenance industry welcomed advances in computerization as a means to help prevent expensive machine breakdowns.
When SymphonyAI Industrial started out doing machine condition analysis for the U.S. Navy back in the 1970’s, the vibration work was all manual. You’d have a group of engineers that went out on the ship for up to 10 days to gather all the measurements via analog instrumentation tape recordings. It would then be played back by technicians through a processor to produce a five-foot high stack of graphs. For the next three weeks, a group of analysts would go through all the data, manually categorizing all the machinery and making specific repair recommendations.
Even the private sector “made do” with rudimentary results while absorbing huge manpower expenses. Automakers and power companies, for example, would maintain a staff of vibration experts for the sole purpose of tracking machine life.
Perhaps the advent of big data possibly accounts for the reluctance of some individuals to believe that today’s basic PdM software can run on nothing more than a common desktop computer. Yet, given the comparative low cost of entry and the gains to be had, maintenance managers of even small manufacturing facilities are now turning to full featured PdM systems. Such are the advances that have been made over the past several years.
Overly cautious, prevention-minded systems of the last decade wasted valuable maintenance resources by manually analyzing each piece of new data, whether the machines needed attention or not. 80% of the machines in a typical plant will have no serious mechanical faults, so why waste time on machine parts that have no need for replacement? State-of-the-art predictive software employ expert systems to weed out machine test-results that look acceptable; thus allowing analysts to focus only on those machines that may have faults. The time saved by not manually reviewing the data from every single machine in a plant is significant.
As an example: Some thirty years after its initial work for the Navy, SymphonyAI Industrial deployed its expert system of PdM software to analyze data for thousands of machines within the US Navy Sealift Command. A subsequent study of 332 such machines confirmed savings of between tens of thousands per month, as opposed to manually calculating the data with in-house engineers. The accuracy rate proved to be 8% greater with the PdM system, as well.
Spectacular gains in maintenance efficiency and dollar savings have resulted from integrating modern PdM systems into existing facilities. Yet, the use of PdM to predict the useful life of brand new machines promises to save organizations even more money.
Recently, Chrysler’s Toledo North assembly plant avoided costly downtime by accurately predicting maintenance failures in newly purchased equipment. “During the launch period of the plant we requested evaluations via vibration analysis and IR analysis as one of our buy-off criteria before we signed off on the equipment and took ownership,” recalls Terry Kulczak, the maintenance advisor for the plant. “We had already settled on the SymphonyAI Industrial ExpertALERT system for this task because we had good success with it at our Durango plant in Newark, Delaware.”
Kulczak explains that his team used their PdM system to evaluate over 600 pieces of new equipment such as regenerative thermal oxidizer motors, water pumps, cooling fans, and gearboxes. “Using the SymphonyAI Industrial software, we found that there were some machines out there that weren’t up to spec,’” says Kulczak. “Some had bad bearings, alignment problems, and improperly sized shims, which led to excessive vibration. These had to be changed out, and it was all done under warranty.” At least 106 pieces of equipment needed adjustment or new parts. In his report to management, Kulczak estimated that the maintenance costs to repair these defects—had they not been detected in advance—would amount to at least $31,000, with a possible maximum cost of $112,000. Production losses due to failed machinery would have resulted in even greater losses to the plant’s bottom line.
“The software analyzed the data and spit out the summary sheet, so you can go back into the data and decipher it a little more closely if you want,” says Kulczak. “From what I understand, older software out there did not have these features. A lot of the contractors questioned our calls, and they didn’t use SymphonyAI Industrial. But we
showed them the data and it turns out we were right.”
Gains made possible through improved data collectors. The above-mentioned, bottom-line savings earned by investments in maintenance management have occurred through the continuous evolution of each element of the PdM chain. Progress has been achieved at every step, beginning with technological advances in field instrumentation that measures torsional vibration, torque and horsepower, motor current, structural vibration, shaft alignment, and acoustics with every increasing accuracy and fidelity.
Even more important, an emphasis on portability has placed condition monitoring into the hands, literally, of production floor personnel. As proof, SymphonyAI Industrial’s data collector can host a replica of an entire condition monitoring database. With analysis tools and automated diagnostics onboard, the user enjoys a tremendous advantage in obtaining immediate answers while in the field.
With today’s versatile and portable data collection systems, you are basically doing EKG’s on machines. Just like when you go to the doctor, an EKG will detect any early problems. Similarly, machines will give off precursors, or early indicators, of future problems when using this type of solution.