Making Sense Of Swirling Airflows

TURBULENT AIR CURRENTS around a car or jet are very complex. They can strain even supercomputers using traditional equations that break the smooth movement of air into billions of tiny steps. Errors become magnified as equations are repeated. So Exa Corp. in Cambridge, Mass., is trying a different approach--one that traces its roots to the work of John von Neumann in the 1940s. Its "lattice gas" method assumes that globs of molecules are restricted to moving along the lines of a three-dimensional lattice, bouncing off one anther only at the intersections. Exa has found a way to erase the traces left by this simplification.

If it works, Exa's approach could let designers solve aerodynamic problems while seated at workstations instead of supercomputers. The company, founded in 1992 by Kim Molvig, a theoretical physicist at Massachusetts Institute of Technology, is still working out some kinks. Last year, a glitch developed in software. But some big potential customers are willing to wait. Says Mark Gleason, a supervisor at Chrysler Corp.'s aerodynamics and fluid dynamics department in Auburn Hills, Mich.: "Some of their results on benchmark problems have been excellent."

Before it's here, it's on the Bloomberg Terminal.