IF YOU'RE one of the thousands of process automation professionals around the world burdened with an older, maybe even antiquated plant—listen up. We’re not just talking to you if you live in the U.S., the rest of North America, or Europe, either. There are older, out of date plants everywhere. They exist in Japan, India, and China, too. So in this article, we are talking to YOU!
What’s Asset Management?
You may be confused when people use the term, “asset management.” It isn’t an accident. It’s a buzzword like “synergy” or “integrated.” Some vendors use it to mean an operations and maintenance function, like predictive maintenance. Others use it to mean process optimization, using a plant’s control system and field devices. The term is in serious danger of becoming whatever the vendor wants it to mean, when he’s explaining to you why you can no longer get along without it. However, both asset management and asset optimization can make you a lot of money, if you play your cards right. In the next few pages, we’ll show you how to draw a full house with asset management.
The simplest way to look at this is as a series of steps to improving your plant’s availability (what we used to call “uptime”). Even though it’s clear that it’s a very costly practice, many plants still operate on reactive maintenance—when it’s broken, then they fix it. The reason this is so costly is clear from the “after incident reports” we’ve all seen. The failure of a $3,000 level sensor shuts down an entire batch reactor train, costing the company $50,000 per hour in lost production, for days, while the device is repaired or a new one is procured.
This has led companies to try predicting where problems will occur, based on history, and allows failure-prone devices (valve, motor, sensor, controller, etc.) to be replaced during scheduled maintenance shutdowns so they don’t die during a production run.
This isn’t cheap either, and the costs of throwing away perfectly good used equipment because it <ital>might<end ital> fail between two scheduled maintenance periods can run high, especially if you’re talking moderately large control valves or devices of that ilk. Auditors and other corporate financial officials have real trouble understanding the economics of this, too.
Predictive Maintenance—The Heart
The most progressive companies are now attempting to apply predictive maintenance, which seeks to determine when a device is likely to fail, based on its operating characteristics, and replace it just before failure. The idea is that if enough data on the performance of the device is available, it becomes relatively simple to determine when it will fail. Data, including vibration, pressure drop, deviation from accuracy, observable wear, increased power consumption and so forth, can be used to determine when pumps, motors, valves, or sensors need to be replaced. This immediately reduces the number of devices that require replacement every shutdown, which saves money that would otherwise be spent unnecessarily. It also reduces unplanned downtime by spotting and correcting potential unplanned shutdowns before they occur.
This is done by collecting a sizeable amount of data, and looking at the trend plots. In Figure 1 below, you can see where the vibration data being collected (by hand or automatically) from a motor train (pump, gear reducer, and motor) goes rapidly up the chart. This indicates that there’s an approaching failure point. The ability to see failures before they occur is key to predictive maintenance.
FIGURE 1: VIBRATION DATA CAN PREDICT FAILURE FOR MOTORS AND PUMPS
You can see the difference between a worn motor and a new one, and you can see the point at which the vibration trend signals a real problem coming real soon.
Wade Howarth, automation manager for Cargill Health and Food Technologies in Eddyville, Iowa, is saving money with a predictive maintenance program. “We’ve been able to shift our focus to “prevent” from “correct,” and the documented savings are significant,” he says. “We intend to keep exploiting the predictive maintenance environment and avoiding unexpected stoppages.”
So what’s keeping many plants from moving in this direction? Partly, it’s the perceived startup cost and time to implement the system. As Howarth says, “Our work practices have certainly changed.” Sending already overworked operators and maintenance personnel around the plant with clipboards in a sort of “dance of the human dataloggers” is seen as impossible by most, and incredibly burdensome by the rest.
|FIGURE 2: MACHINERY HEALTH MONITOR|
Top-of-the-line Emerson CSI machinery health monitor crunches the data internally, and sends back reduced data, while other systems send raw data back to the host for processing.
Companies have responded by providing optimized dataloggers, and even locating computing power out in the field to automate the process of datataking. For a look at one of the most comprehensive of these, see the Emerson CSI machinery health monitor in Figure 2. However, at $6,000-$8,000 per motor train, this solution is out of reach for pervasive preventive maintenance, even at very large companies. If you have 500 motors in a plant, it’s prohibitive to instrument all of them.