"Cheaper Watches," Feature Article, February 2005

cheaper watches

Research looks at wireless networks to lower the cost of monitoring plant machinery.

by Paul Sharke, Associate Editor

Last spring, the U.S. Department of Energy earmarked $18 million in energy conservation funding to be split among General Electric, Eaton, and Honeywell for developing distributed wireless multisensor technologies. The three projects are expected to find inexpensive ways of monitoring industrial equipment. The energy savings resulting from cheap and pervasive motor monitoring alone could reach 122 trillion Btu by 2020, according to a DOE market report.

GE, through its Global Research arm in Niskayuna, N.Y., will investigate wireless motor monitoring; Eaton Corp. of Cleveland will examine wireless data acquisition from electrical distribution gear; and Minneapolis-based Honeywell Inc. will look at wireless process control. The projects, now completing their first phases, will produce demonstration systems when they wrap up two years from now.

In order to uncover energy wasters, industry needs inexpensive ways to monitor machines and move the collected data over to central processors. Many plants do this already for their large machines with wired sensors and continuous monitoring. But wiring in plants is expensive—mainly because of the high cost of labor—and falls between $160 and $4,000 a foot, according to one expert. Wireless mesh networks are the latest hope in eliminating this costly barrier to abundant information.

A mesh network is a network of networks, according to Sensicast Systems of Needham, Mass. Sensicast, a GE partner in the DOE project, builds low-power radio networks that operate at frequencies where licenses are not required.

Traditional networks, like those for cell phones, use star topographies in which end devices—the phones— communicate through single nodes. All communication to and from each end device passes through the node. Mesh networks turn those end devices into nodes as well.

A plant technician makes his rounds. Major industrial equipment makers are exploring wireless ways of extending his reach.

Redundant links between nodes provide alternative paths for data traveling from sensors to processors, and vice versa. A system can reconfigure itself automatically if nodes move around or connectivity lapses. The network also conserves power by synchronizing the clocks of transmitters and receivers. Individual transmitters turn on only to record a measurement and report data. This makes battery-operated nodes possible.

Critical machinery is usually monitored. Enlightened plants today periodically survey vibration and other predictive maintenance data on their balance-of-plant machines, as well, and recognize the value of such programs despite the sporadic timetable by which they are usually conducted. Such plants "are dying to have programs with more consistent data collection," according to Daniel Sexton, GE's principal investigator on the project. Acquiring data every five minutes rather than every five months is seen as a major benefit, he said.

For GE, at the heart of the problem is figuring out a way to deliver data inexpensively. Once plant personnel possess the data, they can run it through conventional analysis systems to evaluate equipment performance.

Investigators working on the project have so far surveyed existing technologies and examined the radio frequency environment, Sexton said. They've looked at ways of pulling some of the cost out of transducers, which, even as wireless versions, will represent a major expense in any eventual system.

Mounting a transducer to a motor, for instance, should be no more complicated than tapping a hole, he said, because a more difficult mounting repeated over many machines would quickly inflate an installation's cost. Epoxy or magnetic mounts might be preferred for the same reason.

Energy for powering the transducers and transmitters will likely come from long-lived batteries. Researchers have looked at harvesting energy from the motion of the motor as a way of extending battery life. To keep installation costs down, transducers will have to mount to the motors and be able to come online without disturbing the machines they are expected to watch.

Checking Power Quality

Distribution product maker Eaton estimates that 90 percent of the electrical switching equipment the company sells today can monitor the quality of the power running through it. The company further estimates that 90 percent of the buyers for that equipment don't wire it up to take advantage of the data's availability.

Thus, Eaton faces a slightly different business problem than GE. It has to convince its customers that there's value to be mined from all that accessible data. With its recent acquisition of power-quality software maker Powerware from Invensys, the company hopes that it can demonstrate the value in power data to its users.

The company's future customers may get "wireless monitoring by default," according to principal engineer Jose A. Gutierrez, who heads up the company's DOE efficiency project. That's because the company believes that wireless technology will help many applications to run better, he said.

Today there are two "free islands" in the radio frequency spectrum available for unlicensed wireless communications, at 900 MHz and 2.4 GHz, Gutierrez explained. The latter band is considered the wider highway and remains largely uncrowded in industrial settings.

Bluetooth, for example, inhabits the 2.4 GHz band and is used in plants as a way for wireless tablets, personal digital assistants, or laptops to retrieve data from distributed monitoring systems. Originally developed for wireless keyboards and mice, Bluetooth's range is limited to about 40 feet.

Some motor starters (above and below) already have the capacity for monitoring power quality. Convincing plants of that data's value remains a challenge.

Newcomer Zigbee, or IEEE 802.15.4, also resides on the 2.4 GHz band, but occupies the 900 MHz band as well. Developed specifically for sensors and control, it has a range up to 300 feet. Its low transmission rate equates to lower power requirements, Guiterrez said, a benefit for battery-powered sensors looking to conserve energy. Mesh networks increase Zigbee's otherwise limited range.

These two standards join old-timer IEEE 802.11 and its several extensions, known by their vernacular names of wireless LAN, Wi-Fi, and wireless Ethernet.

The new standard for industrial monitoring can be simpler than what's required for cell phones, Guiterrez said. There are no requirements for roaming, for example. Compared with cell phone requirements, 802.15.4 is a "toy," he said.

Although Zigbee is clearly Eaton's choice for a new standard, Guiterrez does not promote it exclusively. Indeed, WINA, the Wireless Industry Networking Alliance of San Ramon, Calif.—which counts Eaton among its members— promotes industrial wireless in general, be it Zigbee, WiFi 802.11, or IEEE 1451.5, a smart sensor standard. WINA focuses on end users with goals of increasing both an understanding of the technology and a confidence in it. The group also aims at enlarging the industrial wireless market.

Keeping Processes on Track

Honeywell is directing its DOE project research at petrochemical manufacturing automation, according to project manager Steve Huseth. Part of that effort will look at steam trap monitoring, he said. Steam traps, which drain condensate from steam lines, begin leaking steam as they wear. With many, many traps occupying the average process plant, such leaks, although singularly small, add up to more than a sliver of wasted energy.

But the question looms: How much monitoring is cost effective? A majority of plants are already quite efficient. Better information could help in wringing more from them, Huseth said, pointing out a popular formula that equates a tenth of a percent of efficiency gained to millions of dollars in energy saved by the typical plant. Due to the cost of sensors and their wiring, however, the tradeoff between more monitoring and better efficiency "is currently about a wash," he said. So the company is working on delivering wireless systems that undercut the costs of hardwired devices.

Honeywell's interest in wireless technology leans primarily toward process control. "A drug or paper manufacturer can throw away a large quantity of product because of a process upset," he said. Pervasive sensing could be quick in detecting shifts in process parameters, providing faster alarms and greater control. The company has already developed its first-generation wireless systems for temperature and pressure monitoring. The goal now is to "push down the cost of instrumenting," Huseth said.

One path to this goal is through sensor design itself. In addition to looking at steam traps, the company is investigating micro machine gas chromatographs. Cheaply made and miserly with power, tiny benzene and hexane detectors could be used throughout a plant.

Inexpensive process monitoring could help reduce waste in plants. A pipe losing heat, for instance, will often change the characteristics of the material flowing through it, Huseth explained. That could make the product useless. Many low-cost temperature sensors, monitored wirelessly, could guarantee a different outcome.

Unlike the monitoring strategies of GE or Eaton, both of which would tolerate intermittent data and even interruptions without drastic consequences, real-time control for Honeywell means the wireless system needs to throw robustness and reliability into the bargain. The Honeywell business problem is one of building confidence that transmitting critical process data over short-range radio networks can happen without interference.

Personnel in a typical chemical plant carry walkie-talkies and wireless tablets as a matter of course, Huseth said. Successful wireless monitoring will co-exist with these other inhabitants of the RF sphere.

Wayne Magnes, who manages the industrial wireless program at Oak Ridge National Laboratory, expects wireless networks to meet resistance as they make their way into the industrial environment. He can recall 20 years ago when naysayers were claiming that Ethernet would never make it in industrial settings. It's everywhere today, he said, but it took a long while getting there.

Today, the United States leads the world in wireless monitoring, Magnes said. He characterizes most current offshore forays into the technology as being of the "me, too" kind. That doesn't mean the stats can't shift. That's why the government is pushing the technology hard onto the private sector.

Put it this way, Magnes said, reflecting on Ethernet's long time coming: "Do we want to wait another 20 years for wireless to get to the factory floor?" Magnes and his colleagues firmly believe that the answer is no.