Predictive Maintenance Finally Gets Wide Adoption

With the need to push costs down, plants are implementing condition-monitoring systems.

Aw 2411 0911 Predict

At the LyondellBasell Industries oil refinery in Houston, a condition-monitoring program has changed the way the company manages maintenance. Now, instead of gathering readings on paper reports that never get reviewed, plant operators collect data on handheld electronic devices, aggregate the data and track trends that can indicate a problem.

A faulty seal or broken pump can lead to a costly production interruption. “Now, operators have an actual reading from devices,” says Mark Fisher, operations reliability supervisor at LyondellBasell. “If the reading is above a certain limit, they’re prompted to tell their maintenance folks so they can prevent a catastrophic shutdown. The readings allow us to fix things before they break.”

Before implementing a Wonderware IntelaTrac system, the plant took temperature and vibration readings on paper. “You can’t trend on a piece of paper,” says Fisher. “The supervisor would set up a big pile of run sheets in a three-ring binder. By the time anyone got around to looking at them, it was too late to take any action in a timely manner.”

He notes that the paper reporting didn’t include specific data ranges to indicate problems. “With the paper system, one operator would look at the reading and see something wrong, while another operator would see it as OK.” With the handheld data readings, a note will pop up on the screen when the range is exceeded, prompting a call for maintenance. “We lose the inconsistency,” says Fisher.

Big brother?

Fisher notes that there was resistance to the new system initially. “At first, operators were skeptical. They thought it was Big Brother.” That changed when operators started to detect fans that weren’t working and seals that were plugged—problems that had gone undetected with the paper system.

Plants are turning to condition monitoring to reduce costs and replace the knowledge of baby-boomer engineers who are about to retire. Some are implementing predictive tools in-house, while others outsource it to software companies or original equipment manufacturers (OEMs). Many plant operators are just now getting around to implementing condition-monitoring technology that’s existed in their control systems for years. Ease-of-use portals and dashboards are helping resistant workers switch to monitoring technology. Condition monitoring includes a number of analytical tools, including vibration monitoring, oil analysis, temperature monitoring and infrared imaging. They share one thing in common—collecting data on plant equipment and analyzing the data to see when things are out of whack. Sophisticated condition monitoring can also catch subtle aspects of equipment performance. “It may not be the temperature that’s the issue, but that it’s rising quickly,” says Colin Shearer, senior vice president of strategic analytics at SPSS Inc., an analytic software company in Chicago.

For some companies, predictive maintenance has become a boardroom issue. “There are companies that consider condition monitoring a strategic advantage,” says Tom Alford, product manager, integrated condition monitoring at vendor Rockwell Automation Inc, in Milwaukee. “One power-generation company called out condition monitoring in its annual report as part of the company’s strategic vision.”

The recession has encouraged the use of predictive maintenance tools. The tools are becoming more popular as plants struggle to extend the life of their equipment and optimize equipment operation in the midst of a severe downturn. Plants can no longer afford scheduled maintenance—which often means replacing something that’s not broken—or the costly fix-it-when-it-breaks maintenance strategies.

With the pressure to drive down costs, many manufacturers turn to technology they already have on hand but haven’t implemented. “Since we’re having a slowdown in capital projects, they’re starting to use the products that they bought in the past,” says Rich Chmielewski, manager for PCS7 at Siemens Industry Inc., in Alpharetta, Ga. “We’re getting questions about reporting and diagnostics. Our customers are starting to use technology they’ve had but haven’t been using.”

Making connections

Much of the infrastructure of predictive maintenance systems has been in play for years. Plants already use smart devices that can sense temperature and vibration. They’re also using a fieldbus network that transmits the device data. “You take the data from smart devices and smart control valves, and you send it along the Hart and fieldbus communication and networking systems,” says Stuart Harris, general manager of the Plant Asset Management business at Emerson Process Management, an Austin, Texas, automation supplier. “You take all that data and apply reliability analytics to see the efficiency of the equipment. Then you combine predictive diagnostics with decision support to make the connection with performance.”

One of the biggest hurdles to adopting condition monitoring is getting people to change long-held maintenance practices. “How do you get away from fix-it-when-it’s-broke? A lot of people are stuck in that type of maintenance,” says Emerson’s Harris. “You stop doing some things in routine maintenance that don’t add a lot of value. Then you identify opportunities for bringing in technology that avoids things getting broke.”

The best transition from old-style maintenance to condition monitoring is gradual. “Maintenance is cultural. In many organizations, the effectiveness of maintenance is measured by how quickly they fix machines when they break rather than measuring the overall cost of maintenance,” says Jonathan Hakim, president of Azima DLI, a
condition-monitoring company in Woburn, Mass. “Either that, or it’s measured on ‘Do I perform all my planned maintenance on time?’ To be effective, condition monitoring has to be combined with a move away from planned maintenance.”

Some plants implement monitoring one piece of equipment at a time, so the data collection and analysis in a broad changeover doesn’t stymie the change. “The idea is to give people the right information to collect,” says Jim Frider, manager of mobile solutions at Invensys Operations Management (Wonderware) in Lake Forest, Calif. “You have to know what’s too much data and what’s too little. That helps to get operators on board, which is always challenging.”

Portals, dashboards and benchmarking have helped ease the switchover to predictive maintenance and condition monitoring tools. “Advances in predictive maintenance have less to do with new technology than with new ways to bring the data to the user,” says Bill Polk, research director at AMR Research Inc., in Boston. “It’s how you see the preventive maintenance data that’s important. It’s now aggregated and put into portals.”

Outsourced monitoring

Some plants turn to equipment OEMs, control vendors or software companies to run their condition monitoring programs. These maintenance monitoring contracts can make predictive maintenance affordable for mid-size to small manufacturers. Some control vendors contract with plants to run their maintenance remotely. A program designed and run by Houston-based vendor ABB Inc. came in handy for Vale Inco’s Voisey’s Bay nickel mine in Labrador, Canada, a site that is more than 150 miles from the nearest road. The Toronto-based company needed to know in advance when a part might fail, because turnaround on spare parts shipments is counted in days, not hours.

To make things more difficult, the mine was a greenfield site, so there was no historical data to indicate acceptable data ranges on the equipment. ABB also had to teach the condition-monitoring program to plant engineers. “That was relatively easy, since it was greenfield,” says Jeff Vasel, global asset optimization manager at ABB. “Change management is easier when you start with people who are new to maintenance.”

ABB monitors the site’s equipment remotely. “I can access their site right now.  We monitor heat exchangers, control loops, even the electric flow. We also track vibration and ultrasonics,” says Vasel. “We are able to predict when a pump or motor will fail within 40 hours. Since we can’t get equipment up there quickly, we have to know when the parts will fail so we can have them at the site when it happens.”

The outsourced model has delivered tangible benefits to small and mid-size companies that can’t afford pricy in-house monitoring and analytical systems. “The small manufacturers are getting the same benefits the large end-users are getting,” says Shaun Kneller, account manager at vendor B&R Industrial Automation Corp., in Roswell, Ga. “So the small guys are able to avoid scheduled maintenance. Instead of replacing a bearing every month, they’re replacing it every six months.”

Brain mapping

The brain drain has also prompted adoption of condition monitoring. The most knowledgeable plant engineers are retiring over the next few years. One way to capture their expertise is to program it into plant technology. “The gray hair brigade is reaching maturity and it’s tough to get new blood in because plants aren’t as sexy as high tech,” says Barry Lynch, a manager with the Proficy Maintenance Gateway at GE Fanuc Intelligent Systems, a Charlottesville, Va., automation supplier. “We’re basically taking the knowledge of the mature workers and pouring it into the IT (information technology) rules. You capture their intellectual property and digitize it.”

As well as digitizing the expertise to keep an individual plant’s equipment running efficiently, the best practices derived from knowledgeable workers can be converted into benchmarks or best practices. “The expertise on how to solve equipment problems lives with a field force of people nearing retirement,” says Brian Anderson, vice president of marketing at Axeda Corp., Foxboro, Mass., which provides remote monitoring services. “You can capture their knowledge and turn it into rules. Then you can use that expertise on a global basis.”

Whether it’s on the control dashboard or outsourced to a vendor, predictive maintenance and control monitoring is seeing widespread adoption in the last couple of years. The economic downturn has prompted tough cost-cutting measures, which means the end of the old and costly “fix-it-when-it-breaks” maintenance mentality. Those who produce predictive maintenance tools have overcome resistance to adoption with easy-to-use dashboards or by taking on the chore themselves.

Related Sidebar - Gerber Scientific Monitors Cutting Equipment Remotely
To read the article accompanying this story, go to www.automationworld.com/feature-6216.

Subscribe to Automation World's RSS Feeds for Feature Articles

More in Control