Jockel Finck, Associated Press An EA-6B Prowler aircraft, a reconnaissance plane used by the Air Force, Navy and Marines, is the subject of extensive research to predict where cracks might develop in the aircraft. Prowlers are often based on aircraft carriers and almost all have had their wings replaced at least once. Dr. Anthony Rollett's research aims to take the guesswork out of the process.

By Mark Roth, Pittsburgh Post-Gazette

There's an old joke in the U.S. Navy that landing a plane on an aircraft carrier is a controlled crash.

Coming in slantwise at more than 100 mph, a carrier jet pilot has to hit a moving landing zone that is only 750 feet long, at which point a hook on the plane will catch an arresting wire that jerks it to a stop.

When a carrier jet takes off, the pilot throttles the engines to full power and the plane is then catapulted into the air by a mechanical slingshot.

In such conditions, it's little wonder that the Navy wants to develop better ways of predicting when its planes will develop threatening cracks and other problems that will take the aircraft out of service.

That's where Anthony Rollett comes into the picture.

Dr. Rollett is a materials science professor at Carnegie Mellon University, and his team is part of the Prognosis program sponsored by the Defense Advanced Research Projects Agency.

The aim of Prognosis is to use the latest scientific research and advanced technologies to figure out when a plane or other metallic vehicle may break down.

Dr. Rollett's area of expertise is analyzing the microscopic structure of metals to predict where cracks might develop.

"What we are doing at CMU," he said, "is constructing computer models that actually represent this material, down at the level of the individual crystals inside the alloy. Once you've got this three-dimensional picture of a structure inside a metal along with the impurities, you can then run an engineering model where you apply a cyclic stress and figure out where in this microstructure the stress or strain will concentrate."

The plane that the Prognosis program is testing is the EA-6B Prowler, a Northrop Grumman reconnaissance plane used by the Air Force, Navy and Marines that can jam enemy radar signals and fire missiles to wipe out radar transmitters.

Not only is the Prowler often based on aircraft carriers, said a Northrop Grumman spokesman, but many of the craft are 15 to 25 years old, and almost all of them have had their wings replaced at least once.

	Dr. Anthony Rollett analyzes metals to predict where they might break down as part of the Prognosis program of the Defense Advanced Research Projects Agency.

Right now, the military errs on the side of caution and takes the planes out of service to check for repairs before they think any problems may have developed.

The goal of Prognosis, said DARPA Program Manager Leo Christodoulou, is to use sophisticated sensors and research like Carnegie Mellon's to determine when a particular plane may need to be fixed, rather than having to pull all planes out of service after a certain number of flight hours.

For every 1,000 planes inspected under the present system, Dr. Christodoulou said, only one has a problem.

"So we've taken 999 planes out for no good reason. Wouldn't it be best to know enough to take the bad guy out and service him and leave the good guys alone?

"In a sense, we want to walk up to an aircraft and say, 'How do you feel today?' "

It turns out that the aluminum alloy used in the Prowler is the same one that Dr. Rollett had worked on in a previous research project with Alcoa.

He believes his current team, which includes scientists from Cornell University and Rensselaer Polytechnic Institute, now has a computer model that can predict where cracks are most likely to develop in the alloy.

The model is based on images of the individual crystals of the alloy captured by an electron scanning microscope.

All aluminum has impurities of iron and silicon, Dr. Rollett said, and his team found that the size of those particles and the orientation of the crystals they were in were the key determinants of whether a crack was likely to develop when the material was put under repeated stresses.

The model is now being tested in the lab on pieces of the aluminum alloy. The initial findings support the modeling, but the university is probably a year away from being able to publish the results, Dr. Rollett said.

John Papazian, who coordinates the Prognosis program for Northrop Grumman Corp., noted that Carnegie Mellon, Cornell and Rensselaer are just some of the top-flight technical universities involved in the research.

Other key players are the University of Mississippi, Georgia Tech, Lehigh University and the University of Pennsylvania.

Some of the groups are working on sensors that can be used to detect where defects are developing in the wings or other parts of a plane, Dr. Christodoulou said.

Ultrasonic detectors, for instance, could send sound waves through the metal that would reflect off any cracks, he said, while temperature sensors could detect hot spots where defects exist.

By synthesizing information from several sensors, he said, repair crews could detect problems in the same way a doctor makes a diagnosis after taking someone's temperature, measuring his blood pressure and getting blood tests.

Once they know where the stresses are occurring, crews could use information from the Carnegie Mellon computer model to help determine how likely it is that a particular plane part will fail, Dr. Rollett said.

The model might also help manufacturers figure out whether another type of alloy might be more durable.

Like much other military research, Prognosis may end up having widespread civilian benefits.

The same techniques might one day be used to test metal in civilian aircraft, other vehicles, bridges and buildings, Dr. Christodoulou said.

Dr. Rollett, who enjoys skiing, speculated that his computer model might even be able to predict how long artificial hips and knees might last.

"The methodology," he said, "should surely work and be applicable to any material that is subjected to cyclic loading."