Putting Fluid Mechanics On More Solid Ground
Doing their science on a supercomputer, researchers led by a team at Cornell University are challenging accepted wisdom about how to predict fluid turbulence. If their new theory holds up, it could affect the design of aircraft, racing yachts, oil and water pipelines, and perhaps even cars.
Traditionally, engineers use eigenvalues--frequencies of oscillations--to predict when a stable system will become turbulent. This technique works fine 90% of the time, explaining such things as how much a building will shake in an earthquake. But for the other 10%, eigenvalues don't do the job, asserts Lloyd N. "Nick" Trefethen, associate professor of computer science at Cornell. For so-called fluid-mechanics problems, a new set of equations, dubbed "pseudospectrums," seem to predict more accurately when small perturbations will disrupt a previously smooth flow of liquid or gas. Trefethen believes this will point the way to piping systems that can handle bigger flows and airplane wings that produce more lift.