Grasp Maintenance’s Bottom Line

Effective predictive maintenance demands more than technology.

By C. Kenna Amos, contributing editor

1 of 2 < 1 | 2 View on one page

Too many chemical companies still undervalue the role of maintenance. “Most CEOs don’t care about how much you do in maintenance, until you cannot deliver — and then they blame you,” says Jay Lee, director of the National Science Foundation’s Center for Intelligent Maintenance, at the University of Cincinnati, Cincinnati, Ohio. Perhaps that’s why he champions precision maintenance, which details how much to do, where and what’s derived from adjustments.

Predictive maintenance (PdM), which focuses on equipment health, plays a key role, he notes. However, this should include more than vibration analyses or condition monitoring — both of which he labels as problem-identifiers. “When you are problem-centric, you don’t forecast — and predictive maintenance is forecasting,” Lee stresses. To him, forecasting means collecting daily equipment metrics for use in plotting deterioration in equipment performance. “With degradation rates — the health map of a machine — you can prioritize maintenance schedules.”

Prioritizing those schedules starts with technologies and people, including non-maintenance staff. For example, operators can do inspection rounds to spot trends in equipment performance, notes Scott Brady, manager of marketing and operator-driven reliability for SKF Condition Monitoring, San Diego, Calif. “And if a trend becomes high, the maintenance guys get notified.”

However, getting information presumes you know what to do with it. “How do you convert data to useful health of equipment?” Lee asks. “[And] how do you prioritize and optimize actions?” For example, if a machine’s performance is degrading, do you repair immediately or examine the impact of failure — and base the decision on the change’s importance and when it’s made? He also asks, “How can you physically convert data into information that can be used by everybody?”

Safety First
Lee believes chemical plants must become safety-centric regarding processes and equipment. That involves comparing data and their frequency to the impact on safety but also looking at costs separately. Superimposing all those data, end-users can identify “the criticality line of safety over cost. As you approach this criticality line, you may need to take more aggressive action,” he says. “You may have to do more than just check but stop operation and do a detailed examination.”

Concentrate on assets, especially those related to safety, having the highest criticality, suggests Bart Winters, solution manager for safety and reliability with Honeywell Process Solutions, Phoenix, Ariz. “It [criticality] can be measured by lost production or from a safety standpoint but safety definitely should take priority.” What’s inappropriate, though, is what he calls the “peanut butter approach,” that is, spreading the effort equally across all assets. Instead, he recommends evaluating where spending has the most impact. This best-bang-for-the-bucks viewpoint pays off. At a typical chemical site, a 1% improvement in overall equipment effectiveness “could be about $1 million annually in increased profit,” he says.

Oxea Chemical Co., Bay City, Texas, clearly rejects the peanut butter approach. Its site has approximately 500 pieces of equipment, including compressors, lots of ANSI- and a few API-style pumps, and even some exotic custom-modified machines. Oxea has contracted with Rockwell Automation to regularly collect and process data on this equipment.

“Some equipment is critical and not spared [so it’s] very important to keep it running,” explains reliability engineer Don Kincer. Some products are sold out, so “you have to keep interruptions minimized,” he adds.

Using Enpac portable condition-monitoring data collectors and signal analyzers, Joey Gramoski, a Rockwell field-services engineer who visits on a preset schedule, provides an exception report that Kincer and rotating-equipment specialist Robert Slaton review. It details what’s becoming unique or abnormal “like a discrete frequency or vibration that was not there before,” notes Kincer, adding: “Most of the time, we’ll go out to the machine he [Gramoski] pointed out and make a maintenance decision shortly after we get his report.”

This new PdM-focused, boots-in-the-field approach to maintenance substantially altered Oxea’s financials. “When I first got here, it was preventive maintenance for bearings. Every year, we replaced all bearings, whether they needed to be or not,” Slaton recalls. But now with the new vibration-focused PdM program, “we’ve removed the work orders and replaced them with the new program. Now, we may be replacing bearings every five years — and it could even be longer.”

Oxea no longer can afford the old way of maintenance, stresses Kincer. “We’re keenly aware of plant behaviors, instead of carte blanche replacing things. We want to know why things broke.” For good reason, too. The plant’s maintenance budget was in the $10-million range, but “now we’re at about $7 million and trending down,” he notes. That’s possible because, as Slaton says, “we’re going to ask equipment how it feels” and have planned versus unforeseen interruptions.

Smarter Monitoring
Oxea’s experience emphasizes how much companies need some intelligent information or mechanism to understand conditions.

Rotating equipment is a good case in point, notes Winters. “With some onset of a problem such as [with] a bearing, several things can be picked up before a failure occurs to prevent secondary or collateral damage.” For instance, vibration data may show a problem “typically months or weeks before failure occurs.” Then there’s also oil analysis and thermography, he adds.

Embedded systems in instrumentation can provide early event detection (EED) and advanced diagnostics for problems such as bearing deteriora-tion and cavitation in pumps, notes Ravi Kant, Chanhassen, Minn., based manager of global refining and petrochemicals marketing for Rosemount Measurement, a part of Emerson Process Management. Depending upon the issue, intelligence can be derived from first principles or operating data.

1 of 2 < 1 | 2 View on one page
Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.


No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments