Dow Corning Corp., in Midland, Mich., started using condition monitoring 15 years ago. The company— which produces thousands of non-finished-goods products, monitors vibration at its 24-hour-per-day process plant.
By monitoring vibration, plant technicians were able to replace bearings and other parts before they failed. The plant was also able to avoid scheduled part replacement when the part showed no deterioration. In recent years, plant engineers have extended condition monitoring to go beyond simple vibration. “Over the last four or five years, we’ve added other technology and we’re absolutely seeing a payback,” says Dan Warren, condition-based monitoring analyst at Dow Corning.
The plant now monitors for heat using infrared technology to identify hot spots. Some problems can be recognized and understood more quickly by monitoring heat build-up than by monitoring vibration, says Warren. “So now we can see a hot spot.” Oil is another parameter to check for machine wear. “We monitor the integrity of the oil and fix the oil before the machine quits working,” Warren notes. The plant also added acoustic measuring devices that measure the sound of the machine and compare sound data to the historic sound data of the equipment. Any sound discrepancy is sent as an alert to the control system, supplied by Invensys Process Systems.
The goal of the monitoring system is to alert plant operators when there is any change in the plant’s processes. “Here’s the litmus test I use—at 2:00 a.m. on a Saturday, is the deviation going to be clearly read by a technician?” says Warren. “We have several pieces of equipment [on which we] can see that something has changed. This is not necessarily a system failure, but it’s a process change.”
The goal of monitoring the equipment is not just to send an alert when something is about to fail, but to determine why something fails. Warren wants his technicians to find out the source of the potential failure so they don’t have to endlessly chase near failures. “We’re done a lot of work to understand the root causes of equipment failures,” says Warren. “It’s cheaper to fix the cause of the failure than to fix the failure itself.”
As the Dow Corning story indicates, the shift from preventive maintenance has been going on for two decades. But recently, it’s been gaining steam. Plants are using new technology to automate the collection and analysis of equipment data. That technology—often found in smart devices already in the plant equipment—reduces the cost of monitoring and gives plant managers more useful data. Because the data can point out any change in the process, it can also be used to improve quality. “When the data comes in, it forces us to look at how the machines are changing,” says Warren. “That improves our efficiency on where we’re spending our time.” That also means machinery can be corrected before product quality starts to slip.
The time to examine maintenance budgets for savings has arrived. Once plant managers get their control systems in place, they start looking for new areas in operations where they might find savings. Many are looking at maintenance budgets as an area to reduce costs, increase uptime and improve quality. Predictive maintenance and condition-based monitoring are not quite interchangeable terms, though they often get meshed together out in the field. Predictive maintenance is simply the ability to determine the health of a part—or fluid—that wears with time. Preventative maintenance is usually on a schedule—like a car’s oil change. Once the part has exceeded its estimated healthy life, the plant shuts down on schedule and the part is replaced. But what if that part still had a year or two of healthy life left? Predictive maintenance data lets plant managers put off planned maintenance until the part is about to fail.
Condition-based monitoring can also determine the health of machinery. But its possibilities go further in measuring the health of the plant processes as well as individual parts. “Predictive maintenance is done when something needs to be monitored for eventual repair rather than for operational health and safety,” says Rick Schilpz, business manager for condition monitoring at vendor Rockwell Automation Inc., in Milwaukee. “Condition monitoring can be used for quality assurance, since vibration analysis can help determine whether the product being produced is good or bad.”
The ability to see if the plant’s machinery is working well—before it quits working altogether—can bring considerable savings. “If you run a bearing to failure, you have to replace other equipment such as shafts. In a cement plant, if you run a gearbox to failure, you have to replace everything around it,” says Michael Bensema, P.E., FIS Condition Monitoring Specialist at FAG Industrial Services, a provider of condition monitoring products based in Danbury, Conn. “That could be $50,000 to $70,000, where a bearing is $5,000.”
Adding it up
At Carter Holt Harvey, a pulp and paper mill in New Zealand, plant managers were able to add up a number of measurable improvements over a four-year period when it implemented condition monitoring as part of its ABB control system. Barry Kleine, who works in ABB’s Process Automation Reliability Services unit, was the head engineer for the first two years of the Carter Holt Harvey project. He points to these results:
- Availability increased from 89.2 percent to 94.5 percent
- Overall equipment effectiveness (OEE) increased by 22.4 percent
- Employee satisfaction rose from 66 percent to 78 percent
- Maintenance costs were reduced by 22 percent
l Maintenance cost per ton of product was reduced by 35 percent.Whether plants are implementing predictive maintenance or the broader condition-based monitoring, the savings tend to be measurable, and they come with a clear return on investment, usually within the first year. “When you move to predictive maintenance—where you fix things only when they’re about to break—you can cut your maintenance budget by about 20 percent,” says Tim Sweet, product manager at Honeywell Process Solutions, a Phoenix-based process automation vendor. “That can add up, since the average refinery has about 30 to 40 maintenance people and a multi-million dollar maintenance budget.”
Access to information
Predictive maintenance and condition-based monitoring can be done either manually or through the plant’s control system. “We perform condition-based monitoring by using intelligent devices on a fieldbus network,” says Moin Shaikh, manager of fieldbus technology and networking at vendor Siemens Energy and Automation Inc., Alpharetta, Ga. “We bring all of the data from the devices to the control level so the maintenance team can have access to the information.” With data coming into a central place—the control system—technicians don’t have to wander the plant collecting data. “The savings come from reducing unplanned downtime, since you’re only doing maintenance when it’s required,” says Shaikh. “Also, with the data coming in from the network, you don’t have to make multiple trips in the field.”
Control-based monitoring first gained traction in the oil and gas, and power industries, but in recent years, it has expanded to other large-machine industries such as pulp and paper, steel, even automotive. “Condition monitoring is most popular in heavy equipment-based industries such as energy plants and metals where there is a lot of wear and tear on the equipment and a failure will be disruptive,” says Houghton LeRoy, research director for asset management at ARC Advisory Group Inc., in Dedham, Mass.
The automation aspects of predictive and condition-based monitoring have the additional benefits of replacing knowledge with technology. Industry in North America is experiencing a “brain drain” as knowledgeable workers retire. “We are seeing a lot of knowledge going out the door at plants,” says Preston Johnson, sound and vibration segment manager at National Instruments, the Austin, Texas-based control and instrumentation supplier. As companies lose that expertise, they replace it with technology, but gather data on machines that used to be monitored by technicians who could tell when equipment was not operating effectively.
However a plant implements and uses predictive maintenance and condition monitoring, the return on investment (ROI) is usually an easy calculation. “The ROI can be much less than a year since it comes from a number of areas,” says David Ochoa, director of strategic planning and asset optimization at process controls vendor Emerson Process Management, also in Austin. “The direct savings are in reduced maintenance costs, avoiding overtime and efficiencies in maintenance, which reduces labor and materials costs. The soft savings come in the avoided cost of unplanned shutdowns.” The savings from unplanned shutdowns are hard to measure—how do you measure something that doesn’t happen? But industry experts estimate that unplanned shutdowns can be reduced by 20 percent to 40 percent when a plant switches to predictive maintenance and condition-based monitoring.
The wireless future
At some point in the next few years, plants will begin to use wireless devices to run predictive maintenance and condition-based monitoring. The idea of collecting data without wires is too attractive to pass up. The technology to support wireless data collection already exists. The hold-up is security. “Everybody wants to talk about wireless and wants to know when that’s going to come out,” says John Wenzler, corporate account executive at vendor Bosch Rexroth Corp., in Hoffman Estates, Ill. “I haven’t seen anyone implement wireless in a real factory setting yet. That’s for companies that are on the bleeding edge. Wireless will eventually catch on, but that won’t come until plant managers and IT (information technology) departments are convinced that wireless security is bullet proof.”
To see the accompanying sidebar to this story - "Changing the Maintenance Focus" - please visit http://www.automationworld.com/view-3641
For more information, search keywords “predictive maintenance” and “condition monitoring” at www.automationworld.com.