0
Select Articles

# A Field bus that ChoosesPUBLIC ACCESS

A Company Introduces a Different Approach to Machinery Monitoring: This One Sorts the Information.

[+] Author Notes

Associate Editor.

Mechanical Engineering 127(08), 28-29 (Aug 01, 2005) (2 pages) doi:10.1115/1.2005-AUG-3

## Abstract

This article focuses on the process involved in machinery health monitoring. Operators need know only about certain things for which they can take immediate action: a cavitating pump, an overheating motor, a severely vibrating train, a failing bearing. They do not need knowledge of longer-term concerns like imbalance or misalignment, the domain of maintenance gurus. Likewise, management needs information about overall plant health and the priorities of various repairs, but it does not want to know that a circulation water pump is exhibiting mild axial misalignment or another pump elsewhere is beginning to cavitate.

## Article

When it comes to machinery health monito ring, "It's not really about the machine," according to folks from Emerson Process Management. Instead, they tell you," It's about the process."

That's one way they justify the expenditure for a new $5,000 to$7,000 Machinery Health transmitter for monitoring an ac induction motor/centrifugal pump train whose worth might be only twice that. The idea is to provide continuous machine informationnot only to the main ten ance department but to plant operations as well. The company is marketing a system, dubbed CSI 9210, that uses Foundation Fieldbus, the digit al two-way communications protocol. Armed with information, operators can keep a process within an optimum operational range.

According to the marketing manager at Emerson, Todd Reeves, the CSI 9210 relies on intelligence embedded within the field device to avoid congesting the fieldbus line with reams of spectral data. After gathering data from six vibration transducers, four temperature monitors, a tachometer, and a flux sensor, the transmitter runs the information through a fuzzy-logic neural network—a kind of" expert in a box"—then dispatches the findings about the machine train to operations, maintenance, and management, Reeves explained.

Operators need know only about certain things for which they can take immediate action: a cavitating pump, an overheating motor, a severely vibrating train, a failing bearing. They don't need knowledge of longer-term concerns like imbalance or misalignment, the domain of maintenance gurus. Likewise, management needs information about overall plant health and the priorities of various repairs, but it doesn't want to know that a circulation water pump is exhibiting mild axial misalignment or another pump elsewhere is beginning to cavitate.

One chemical plant told Emerson that its operators needed updates within 60 seconds if they were going to act upon the results of any analysis. Another plant said that a 4-20 mA sensor couldn't act fast enough to detect a pump tearing itself apart, which takes only seconds.

The system is an extension of Emerson Process Management's PlantWeb architecture, president John Berra said. That system puts predictive intelligence in a plant in the form of a number of familiar brands, such as Rosemount measurement devices and Fisher digital valve controls. The new machine monitoring grows out of Emerson's 1997 acquisition of Knoxville, Tenn.-based vibration monitoring company Computational Systems Inc.

Taking advantage of PlantWeb and its approximately 4,600 installations is a sensible approach to capitalizing on established fieldbus installations. But fieldbus is not equipped to handle the large data sets that make up a series of machinery vibration spectra. That's the reason the CSI 9210 transmitter crunches the numbers at the machine itself, reporting only the equipment's status to operating, maintenance, or management personnel.

Once status is determined, a report is delivered in the form of a diagnosis and a machinery health value—essentially, a new process variable—a new measurement set overwrites any earlier spectral data. The transmitter's neural network then updates the diagnosis, according to Mike Sander, the director of marketing.

The diagnoses themselves are based on more than 20 years of machinery analysis experience, the optimization division's president, Doug Llewellyn, said. All problems have known failure patterns, he added.

Right now, the system is directed specifically to watch ac motor-powered centrifugal pumps. Such machine trains make up a whopping 60 percent of the typical process plant's 2,500 machines, said Brian Humes, vice president and general manager of Emerson's Machinery Health Management unit.

There's plenty of opportunity left out there, though. Emerson's executives said they have plans for centrifugal fans, rotary blowers, geared pumps, and so on. That is, all the rest of a plant's rotating equipment.

New in the architecture: Emerson's Machinery Health transmitter (shown in blue at left) is designed to fit into the company's PlantWeb system. The device crunches numbers remotely and reports a machine's condition.

On the floor: Billed as "an expert in a box" (the blue box on the left, in fact), the CSI 9210 puts streams of data through a fuzzy logic neural network before transmitting it.

The idea is to provide continuous machine information not only to the maintenance department, but to plant operations as well.

View article in PDF format.

## Discussions

Some tools below are only available to our subscribers or users with an online account.

### Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related Proceedings Articles
Related eBook Content
Topic Collections