Technology Bulletin

Artificial intelligence analyzes engine failures in tanks

The U.S. Army may soon have a new mechanic working on its tanks. TEDANN, or Turbine Engine Diagnostics using Artificial Neural Networks, monitors engine operations in the M1 Abrams main battle tank's turbine engine. The technology, developed by the Dept. of Energy's Pacific Northwest National Laboratory (Richland, WA), uses diagnostic engineering, artificial neural networks, and model-based decision algorithms to predict failures and abnormal operations in the engines. Presently, sensors aboard a tank can only indicate if a problem does or doesn't exist. TEDANN monitors engine conditions continuously and tracks potential deviations from normal operations, identifying problems as they develop. Thirty-two existing sensors on the gas turbine engine and 16 new sensors, added using a wiring harness, collect data. These relay the information back to a computer processor for analysis. The analysis is done using artificial neural networks or ANNs. These process information much like a human brain, learning by example. For TEDANN, these neural networks have been configured to model the behavior of normal engine performance and to recognize deviations. Either an onboard display alerts the tank driver if problems occur, or a mechanic via a laptop computer. A TEDANN prototype was installed on a Washington National Guard tank in September 1998. Over the next year, seven more M1 tanks will be equipped with the device. "By using TEDANN, the Army could expect to extend the life of the M1 Abrams tank fleet, lengthen the time between costly overhauls by 25%, and avoid hundreds of millions of dollars in maintenance costs over 30 years," says Frank Greitzer, program manager for Pacific Northwest. E-mail staci.west@pnl.gov.

Computerized program helps train athletes

Football fans watched the Buffalo Bills closely this year partly because of the exciting play of quarterback Doug Flutie. In addition to Flutie, part of the reason for the team's success could be a method of training developed by University of Buffalo (UB) physiologist David Pendergast and University of Buffalo men's swim coach Budd Termin. "It's the only program where performance in training is actually quantified," says Pendergas. "It's also the only program that provides precise individualized training and direct feedback to swimmers." Initially, Pendergast and Albert Craig, Jr. of Rochester developed a swim meter, which measures the velocity achieved at different stroke frequencies. Training is enhanced by using a computer-programmed, underwater, light-pacing system developed by Pendergast, Termin, and UB equipment designers John and Michael Zaharkin. A patent on the system is pending. Using the swim meter in a special ring-shaped pool, Pendergast and Termin clock each swimmer's speed at increasing strokes per minute, and measure oxygen consumption and lactate production. Those data are plugged into formulas and transferred to graphs, forming a baseline stroke-velocity curve and a metabolic-velocity curve, which monitor energy consumption at various speeds. Once done, new curves are plotted, representing each swimmer's immediate goal. The athletes are trained to match this higher curve by increasing the distance the body covers with each stroke, thus swimming faster. The light-pacing system helps the swimmer know if he reaches this goal. If the swimmer passes over the light as it flashes, he is on the mark. Pendergast and Termin have been working on the system for eight years. Using the system, UB swimmers improved their stroke frequency-velocity ratio by as much as 30% during a college career, Pendergast says. Competitive runners, the Buffalo Bills, and the Buffalo Sabres have all used the basic method, without the lights. FAX (716) 645-3765.

Images of blood vessels made clear with prototype camera

Brain stents, thin wires placed inside blood vessels in the brain, can prevent stroke by shoring up weak spots in arteries or blocking off aneurysms. However, they are difficult to see. At the annual meeting of the Radiological Society of North America on December 2, 1998, Stephen Rudin, Ph.D., University of Buffalo professor of radiology and physics, reported that a prototype camera developed by researchers at the University at Buffalo's Toshiba Stroke Research Center provides images that may allow the viewing of even the tiniest blood vessels in the brain. These vessels, called perforators, are only 50-200 microns in size and can't be seen with conventional imaging equipment. "Even with the most advanced imaging equipment available at present, we weren't seeing features we knew existed," Rudin says. "We expect this detector prototype to help us locate the stent optimally in the vessel, visualize its integrity in place, and reposition it if necessary. If you can't see exactly what condition the stent is in when it's deployed, it's not possible to change the deployment." The detector technology, a high-resolution region-of-interest microangiographic digital detector, is similar to that being introduced now in mammography, Rudin says. E-mail ljbaker@buffalo.edu.

Yes, there is such a thing as thermal contraction

Everyone knows that solids must expand when heated, right? Well, don't tell that to zirconium tungstate--a material that actually shrinks instead. Reporting in the Nov. 12 issue of the journal Nature, researchers at Lucent Technologies (Murray Hill, NJ) and the Johns Hopkins University (Baltimore, MD) say they found that zirconium tungstate atoms vibrate at low frequencies, causing the material to fold in on itself when heated. "Naturally, everyone has heard about thermal expansion," says Collin Broholm, an associate professor at Johns Hopkins. "But who has ever heard of thermal contraction before? Really, it's the ideal engineering material. It's quite astonishing." Although several materials actually exhibit the same phenomenon, zirconium tungstate is the only one that both shows the behavior around room temperature and acts peculiarly at a constant rate, exhibiting an overall shrinkage or expansion rate of 0.0005% for each degree from -375 to 700F. Because zirconium tungstate can counteract unwanted shrinkage or expansion effects in other materials, it may have industrial applications. One possible application: forming composite-based components in next-generation fiber optic technology for optical networking. Currently, Lucent is evaluating a zirconium tungstate composite material, developed by Bell Labs Ceramic Engineer Debra Fleming, as a potential packaging material for a filter, or grating, used in glass optical fiber. The material's unique shrinkage properties would compensate exactly for variations in the glass fiber as temperatures change, researchers say. Otherwise, multiple wavelengths, or channels, of light transmitted through a fiber would become a scrambled mess. FAX (410) 516-5251.

NIST proposes software to regain machine-tool controller market

The US share of the machine-tool controller market is 13%, says the National Institute of Standards and Technology (NIST, Gaithersburg, MD). Competition in this market is currently based on keeping costs low, and edging out competition through proprietary barriers that make it difficult for products from competing vendors to coexist, NIST says. In an attempt to combat this trend, NIST's Manufacturing Engineering Laboratory's Intelligent Systems Div. is developing open system interface standards that will reduce the life-cycle cost of controllers for machine tools through a project called the Enhanced Machine Controller (EMC). Several industrial sites are presently testing the EMC software. Jim Tuthill, production manager for Flat Plate Inc., an EMC test site in York, PA, recently demonstrated the company's use of the EMC to drill a series of small holes in copper tubing for heat exchangers. When he was asked for a scrap part that could be used during EMC presentations at NIST, Tuthill replied that they don't scrap parts with the new controller, emphasizing the potential benefits of the EMC and numerical control in general to job shops. Call: Fred Proctor at (301) 975-3422.

Skipping: the only way to fly

People who want a faster way to travel may get their wish. Preston Carter, of the U.S. Dept. of Energy's Lawrence Livermore National Laboratory, has derived a concept for a hypersonic aircraft that can fly between any two points on the globe in less than two hours. Dubbed the HyperSoar, the airplane would fly at about 6,700 mph (Mach 10), while carrying roughly twice the payload of subsonic aircraft of the same takeoff weight. To travel faster, the HyperSoar would "skip" along the edge of Earth's atmosphere--much like a rock skips across water, says LLNL researchers. A HyperSoar aircraft would ascend to approximately 130,000 ft, outside the bulk of Earth's atmosphere, turn off its engines, and coast back to the surface of the atmosphere. There, it would again fire air-breathing engines and skip back into space. The craft would repeat this process until it reached its destination. Researchers estimate that a commercial flight from Midwestern United States to Japan would take 11/2 hours and require about 25 skips. The aircraft's angles of descent and ascent during the skips are predicted to be about 5 degrees. LLNL scientists say that passengers would feel 1.5 times the force of gravity at the bottom of each skip and weightlessness while in space. They compare 1.5 Gs to the effect felt on a child's swing, though HyperSoar's motion would be 100 times slower. All previous hypersonic aircraft concepts have suffered from heat buildup on the surface of the aircraft and in various components due to friction with the atmosphere. Researchers say a HyperSoar plane would experience less heating because it would spend much of its flight out of the Earth's atmosphere. Also, any heat the craft picked up while "skipping" down into the atmosphere could be at least partially dissipated during the aircraft's time in the cold of space. E-mail yano1@llnl.gov.

A silicon ruler for shrinking microchips

Smaller. Faster. Cheaper. That's what the computer industry has been demanding for years as semiconductor manufacturers continue to do just that. However, with the tiny chips comes a problem--circuits now have features that are too small to be measured reliably with existing metrology systems. To remedy the situation, Sandia National Laboratories (Albuquerque, NM) and the National Institute of Standards and Technology, with support from International SEMATECH and the U.S. Dept. of Energy, are creating a reference material for microscopes designed to measure features one-tenth of a micrometer in size, or about 500 times thinner than a human hair. Researchers make the material from lines etched in silicon. Call: Michael Cresswell at (301) 975-2072.