Washington Show n' Tell

Thirty years ago, who would have suspected the Internet? Yet it was back in the '60s that the Department of Defense funded the underpinnings of today's www.world.

Hoping something equally cool (and useful) will result in 2020, a contingent of Penn State faculty joined an exhibition last May meant to persuade Congress to "maintain its investment in fundamental research, even though the payoffs may not be known today," according to University president Graham Spanier, the event's emcee.

The exhibition was held in a caucus room in the Cannon House Office Building, an ample, old-fashioned gallery, its rococo gold-leaf ceiling and grand chandeliers an odd contrast to the 30 universities' futuristic displays. Here were panels on jet lag and sleep deprivation, wavelets, a medical free-electron laser, the non-destructive testing of airplane canopies, wireless video terminals, subsurface ocean waves, x-ray lithography, the psychology of tactical decision-making, terahertz science and technology, the theory of chaos applied to intelligence, protein-based 3-D memories, telemedicine, superconducting digital electronics, the adhesion of thin polymer films to metal, and The Command Center of the Future. Explanations, freely offered, were lengthy, technical, and enthusiastic, garnished by shish kebabs and strawberries courtesy of the American Association of Universities.

Asked to say a few words to the gathering, Senator Strom Thurmond declared, "You scientists, you mathematicians and others, you're doing a great job. This exhibit underscores the need to have a robust budget. We need to allow you people to provide us with technology."

Admiral Paul Gaffney, chief of Naval Research, in his turn noted, "The people in this room are consumed with investing in the future. The people out there are consumed with consumption. You're right, they're wrong. You've got to convince them of that."

Lyle Long, an associate professor of aerospace engineering at Penn State, stood with hands clasped and thumbs discreetly twiddling while the dignitaries spoke, as if eager to get on to the convincing part. Behind him a computer scintillated with a gaudy, almost Pop-Art eyeball design.

"It's jet engine noise," he said. "It runs for three weeks."

Back in his office a few days later he drew a cartoon to explain. "This is a slice through a jet engine," he said. "Combustion—" he pointed to the pod in the middle— "drives the turbine, the turbine drives the compressor—" on the right, from which the exhaust shoots out, "and the fan—" on the left, where the air is sucked into the engine. "If somebody can tell us the pressure right here, in front of the fan, we can compute the noise in the far field, that is, what a person on the ground would hear as a jet with this engine flew over."

"Five years ago you couldn't do this," he added."Even now the calculations to model this "fan noise," as well as the noise coming out the engine's other end, what he simply terms the "jet noise," took about 1,000 hours of computer time at the National Supercomputer Center on Maui, or three weeks each.

Long had also displayed the jet noise image in Washington, this one suggestive, in an ink-blot sort of way, of ocean waves or chest x-rays. "It's just a slice through another 3-D model," he said. He and Philip J. Morris, the Boeing professor of aerospace engineering at Penn State, had worked out the computer algorithms, based on the 100-year-old Navier-Stokes equations describing fluid flow. "With a slice," Long explained, "you can see the flow structures, the turbulent eddies and things."

"These two pieces of the engine-noise puzzle are very, very different," he continued. "The fan noise is sort of a nonlinear propagation problem. The jet noise is more of a turbulence problem. Turbulence is one of the grand challenges left in physics. To get the jet noise right, we have to get turbulence right."

When engineers know what noise looks like, where it comes from and how it magnifies itself, they can design to muffle it. "If you look out the window on an airplane today," Long said, reaching for a long metal brick sitting on a bookcase, and you look at where the air goes into the engine, you'll see metal with holes in it. Behind that metal is this honeycomb structure." He turned the brick to show off an underside made of many boxes. "The sound gets inside all these little metal boxes and just dissipates. We'd like to be able to tell companies like Boeing where exactly to put this honeycomb stuff on the engine."

There are technological fixes for the back end of the engine, too. "You can put shrouds, called ejectors, out there to shield the very noisy part of the jet. We hope to model those."

And then, there's a lot of data on circular jet engines, but now people are interested in rectangular shaped engines. There might be some noise reduction with these rectangular jets because the mixing is completely different. The thinking is, if the fuel is mixed faster there might be less noise."

And less noise would allow engineers to build a bigger engine without overtopping the population's noise tolerance. "The U.S. would really like to build a supersonic transport like the Concorde, but there are a lot of objections from people. Look at the number of places the Concorde has been given permission to land. It's not many. Two or three, I think, in this country, where people are willing to put up with the noise. The folks from NASA are saying if they cannot reduce the jet noise, this project will never get off the ground."

To get a sense of the total noise from a supersonic engine as it zooms overhead, however, Long and his colleagues need to combine the fan noise, the jet noise, and the noise inside the engine itself. "Right now NASA has codes that do this inner part. What we hope to do is take all three codes and combine them."

He's hoping to have Internet II soon to help with that. Sort of an upgrade of the current Internet, Internet II would allow researchers to send more data farther and faster, without worrying about the heavy traffic on the Internet. "Here at Penn State," Long explained, "our biggest parallel computer is an 84 processor SP-2. But one person typically can use only 32 processors at a time. So we do all of our code development here and when we want to run these huge problems, we have to go to a National Supercomputer Center like the one on Maui. Now we go through the Internet. We can log into Maui from a classroom if we want to. We FTP our files out, compile them, run the job. But because it's so slow, we take back as little data as possible and do the visualization here.

"What we'd like to do—what Internet II would let us do—is to run the project in Maui and to watch it develop graphically on our own terminals here. It's called computational steering. If something's wrong, you could see it right away and fix your code. You could shut it down and change some variables, start over with different data. You could also see when it's done—when it's computed enough. You might save 100 hours of computer time.

"What'll it take? Some fiberoptic cables and some infrastructure here: electronics, switches. And the phone company will have to dig a line all the way down the freeway to Pittsburgh." He grinned. "Funding that is the biggest problem."

A week after the exhibition in Washington, the New York Times published word of a National Science Foundation-funded study: "73 percent of the main science papers cited by American industrial patents in two recent years [1987-88 and 1993-94] were based on domestic and foreign research financed by government or nonprofit agencies. . . . Such publicly financed science, the study concluded, has turned into a fundamental pillar of industrial advance."

"Look at the things that are coming out of the research pipeline," the study's lead author, Francis Narin of CHI Research, told the Times. "We'd be fools to close it down."

Lyle N. Long, Ph.D., is associate professor of aerospace engineering in the College of Engineering, 233M Hammond Bldg., University Park, PA 16802; 814-865-1172; lnl@cac.psu.edu; http://cac.psu.edu/~lnl/lnl.html. His work is supported by the Department of Defense and NASA. Philip J. Morris, Ph.D., is Boeing professor of aerospace engineering.

Last Updated September 01, 1997