From on_edges@hotmail.com Fri Jan 26 13:35:44 2001 Date: Sat, 20 Jan 2001 05:42:56 -0800 (PST) From: on_edges@hotmail.com To: hroberts@uic.edu Subject: Forwarded article: SOFTWARE SHOWS WHEN TO PUT FIX IN The following article was selected from the Internet Edition of the Chicago Tribune. To visit the site, point your browser to http://chicagotribune.com/. ----------- Chicago Tribune Article Forwarding---------------- Article forwarded by: Jerry Roberts Return e-mail: on_edges@hotmail.com ---Forwarded article---------------- SOFTWARE SHOWS WHEN TO PUT FIX IN By Jon Van As much as passengers hate a flight cancellation for mechanical problems, airlines hate it more, which is why Delta Air Lines is testing a technology that predicts engine glitches before they spring up. Delta is one of several companies evaluating software supplied by a tiny Lisle-based firm called Smart Signal Corp. that is intended to reduce unexpected equipment failures. If the product delivers as promised, they expect big savings in maintenance costs. Why the interest? Simply put, surprises cost money. Say Delta learns an engine needs to be replaced on a plane that has landed in Japan before it can fly back for service at the carrier's Atlanta headquarters. In such a case, it could cost 10 times more than if the work were done at home, what with shipping and lost travel time. "Our goal is to avoid surprises because they're no good for our passengers and no good for us," says Udo Rieder, Delta's vice president of engineering and quality. Airplane engines already are loaded with sensors that measure mechanical characteristics like vibration. When those measurements drift outside the normal range, Delta must replace the engine. Smart Signal's software analyzes engine sensor information, makes a model of normal operations, then uses subtle variations in the data to predict when trouble is coming before any measurements go outside the normal range. Depending on the complexity of the operation, Signal's software, say, for a factory process, would start at $50,000 for a license. "It definitely works," Rieder says. "We're still in the prototype mode now, but if this pans out as we expect, we'll expand this technology to cover all engines in our fleet. We'll probably also use it on (airplane) frame data." Smart Signal uses technology developed at Argonne National Laboratory that was designed to make nuclear power plants operate more efficiently. But Smart Signal sees a broader market, said Gary Conkright, Smart Signal's chief executive. Any firm with significant capital invested in assets wants to keep those assets operating, he said, and Smart Signal's products can cut the costs of downtime by enabling a customer to perform maintenance just before it's needed instead of after trouble strikes. While many products claim to deliver "predictive maintenance," none are in the same league as Smart Signal, which holds more than two dozen patents, Conkright says. Industry analyst Kevin Prouty agrees. "There are lots of packages out there, but few have any decent analytic tools," said Prouty, manufacturing research director at AMR Research in Boston. "Finding a pattern in them is just an exercise in eye strain." The only problem Prouty sees with Smart Signal's technology is its complexity. "It's basically a black box," he said. "You don't know what goes on in the box, and that's difficult for people to accept. What that means is people want a fairly lengthy validation period before they're sure it works and they know how to use it." Smart Signal's executives say the beauty of their technology is that it detects subtle patterns in any data based on processes that follow the laws of physics. It may take time to convince customers that's true, said John Moore, vice president of the ARC Advisory Group, an industry analysis firm in Dedham, Mass. "They claim to have a generic model," Moore said. "But if you're operating a refinery with hundreds of different pumps made by many different manufacturers, do you have to make a different model for each pump, or will one fit all?" Getting several customers to build experience with the technology should overcome that problem, said Moore, and the company is making deals intended to do that. Smart Signal is working with Rohm and Haas Co., the giant specialty chemical firm, to demonstrate its technology's versatility. Chemical processing is especially difficult to track because as time passes, the quality of a catalyst can change, altering the process output, said Lynn Richard, process control manager for Rohm and Haas. The challenge for Smart Signal is to incorporate these catalytic changes into its modeling software. "We see that their model works very well for about a week, which is why we're interested," said Richard. "But then you need a new model. We're working with them to develop a program that will change as conditions change. Smart Signal is privately held and has raised more than $7 million in capital from venture firms that include CID Equity Partners, Perfect World Capital LLC and ARCH Development Corp. A few months ago the grain processing giant ADM not only signed on as a customer, but invested $1 million in backing the small firm. The company has 24 employees and intends to grow to 70 by year's end, Conkright said. "It's the perfect time for our technology," he said. "Devices are getting smarter and wireless sensors are churning out more data. The world's awash in data, and we're the ones who can make sense of it."