Risks rise as factory nets go wireless

CHICAGO - Wireless technologies for harvesting real-time data off factory networks - and strategies for putting such information to use - were hot topics at last week's National Manufacturing Week conference.

However, some manufacturing IT professionals highlighted the risks involved in deploying wireless technology in factories, or converging plant control networks with IT systems and back-office data centers. Others voiced concerns that gleaning too much data from manufacturing processes could complicate rather than streamline a manufacturing operation.

"Manufacturers are getting closer to where demand and supply are more synchronous," says Bob Parker, an analyst with IDC. "We're seeing the emergence of RFID and sensor networks as the means for collecting the kind of data that is needed to make this synchronization happen."

Among those moving in this direction is General Motors, which has deployed RFID, 802.11 and sensor networks in several plants.

"The goal isn't just to merge manufacturing plants with IT [infrastructure]," says Pulak Bandyopadhyay, group manager for plant floor systems and control group in General Motor's Manufacturing Systems Research Lab in Warren, Mich. "It's the collection of real-time data and what you do with that data you've collected."

Part of General Motors' move toward IT/shop floor integration has been the deployment of standard Ethernet and IP as a backbone technology for connecting most factory equipment and assembly line systems.

"We've been trying to get rid of separate control networks for everything," Bandyopadhyay says. "Everything now is connected to an Ethernet backbone."

Now the automaker is looking toward 802.11 technology in factories to support technicians using wireless PCs and PDAs on the shop floor. GM is also deploying sensor networks, which use tiny nodes that can monitor a process or device and send data to a computer via a wireless mesh - in which each sensor point is also a mini switch that can send data to and receive it from any point on the plant floor. Deploying sensors on machinery could help GM technicians get more data on why a machine broke down and to repair equipment faster. Tying these sensor networks to back-end systems also could help predict when a machine is likely to fail.

"We spend over $1 billion a year in maintenance on our factories," Bandyopadhyay says. "If we have real-time systems that can track mean time to failure on a piece of equipment, that can help us."

GM is also using a mix of wired Ethernet, 802.11 access points and RFID to track parts in some plants - all the way from suppliers to each stage of manufacturing. This gives the company a more exact picture of how far along a car is in the production cycle.

For GM, the ultimate business driver for integrating real-time data from plants, logistics and other areas of the supply chain is a bit more radical. Bandyopadhyay says he wants GM's networks to be integrated to the point where communications with dealerships, logistics and manufacturing can turn the traditional model of manufacturing on its head; instead of making cars based on forecasts of what buyers want, GM will make cars based on what auto consumers want now. But this is still a way off.

"Getting information on what people want out of the dealerships and making changes to the manufacturing process on the fly is still a work in progress," Bandyopadhyay says. "When you connect all of this together, that's where the big payoff is."

Other manufacturing IT pros question how far a plant should go in putting critical processes on the same network, or cutting the cord and going wireless.

The efficiencies of a single IP network and wireless are attractive, says Scott Buettner, project manager of information services for Honda of America Manufacturing, in Anna, Ohio. But there are concerns, such as a failure of an IP switch or router causing stoppages on an assembly line.

And with wireless, things might get even scarier.

"If an application is mission critical and relies on a wireless connection that is susceptible to interference, then that is a safety issue," he says. In a scenario where a piece of heavy equipment is being operated via a wireless control, "maybe a 1,000-pound die is being moved by a crane . . . then someone comes along, working on a spreadsheet on their wireless laptop, and the signals interfere with each other. Best-case scenario, the person with the laptop loses their work. . .worst case, maybe the [crane controls] fail and the die falls on that person's head."

The collection of real-time data - whether through wired or wireless network infrastructure - poses another dilemma for manufacturers; the sheer amount of data collected from factories, warehouses, suppliers and distribution facilities can become unwieldy and difficult to parse.

"You've got massive amounts of data coming into management these days," says Gary Matula, CIO at Molex, a manufacturer of electrical plugs, cabling and electronics components. Beyond the challenges of collecting factory data are the issues of making sense of it and putting the information to use.

"We're pushing them to make more decisions based on data," Matula says. "But when you deal with people who have been in the industry a long time, they think they have all the experience in the world."

Even smaller companies are becoming more efficient with integrating back-end IT systems with plant floor operations. Weil-McLain, a maker of home heating systems, recently moved its plant from Michigan City, Ind., to Eden, N.C. The company moved from a plant with older, inflexible production lines to a modernized factory. At the same time, it rolled out software from J.D. Edwards (now owned by Oracle) to revamp its back-end processes.

"We're basically moving from the dark ages of technology on the plant floor and spreadsheets to manage everything, to demand-flow manufacturing," says Tony Bauschka, director of IT.

Instead of building boilers based on forecasts computed in spreadsheets, the software ties the company's sales order system into the factory production schedule; the only boilers that are built are ones that have been sold already. The software also streamlines how bills of materials and assembly instructions are passed from the design applications to the plant floor. For example, Weil-McLain has gone from text-based work instructions to visual-based assembly guides for workers. All of these improvements have allowed the company to reduce lead times from three weeks to eight days, Bauschka says.

Learn more about this topic

How GM saved a billion dollars

Network World, 07/12/04

Food maker mixes simple ingredients in IT shop

Network World, 10/20/03

Mixed net brings smarts to factory floor

Network World, 11/15/04

Editors' Picks
Join the discussion
Be the first to comment on this article. Our Commenting Policies