Reinventing the wheel - Jack Bitterly is developing the flywheel automobile engine

Bitterly began by following a dream he had nurtured ever since starting his career in the Lockheed Skunk Works during World War II. That original team had been housed in a top-secret compound; everyone not involved in the jet-fighter project was told to pretend the compound stank and not even think of going near it. Five months later, a new kind of airplane rolled out the door. Today Bitterly advances a concise theory of skunk-works efficiency. He draws it on a napkin: a simple graph that plots productivity versus the number of engineers working in one place on a project. His graph takes the shape of a bell curve, peaking at 15 to 20 persons. To the left of this peak, you have too few people--not enough ideas. To the right, with more people, his curve actually goes negative. "You produce not ideas," he snorts, "but dogma."

The Bitterlys recruited all the differing specialists they would need, including several of Steve's former colleagues in high-energy laser research. "You don't find many people who are trained in flywheels," Bitterly says. "They are not nine-to-five types," he adds. Often, they spent weekends holed up in a glass-and-concrete building on a dead-end street off the Ventura Free way. No one outside the company knew much about it, and Bitterly says he "discouraged" inquiries from the press. A fait accompli, he and Costner agreed--a working system, ready to roll out the door--would bring the best publicity.

In November 1993 the nouveau skunk-works team began designing a totally new flywheel system. The wheel itself, they decided would be simple: one fly; eel on one axle, without embedded magnets--just a clean fiber disk. To cancel gyroscopic forces, they would simply stack flywheel--containing canisters together as pairs. After they settled on a basic wheel design, a few months into the effort, "there was never a major flaw," Bitterly says. "Our problems tended to be things like a loose wire in an instrument." The complete, integrated system design "locked down," he says, early in 1995. By fall it was up and running.

One key to this speed was computer-assisted design; Bitterly installed three workstations. Another was the well-matched skills of father and son. Steve is a whiz at computer modeling; Jack is a hands-on engineer, with an uncanny feel for materials. He knew, from years of experience, that fiber wheels would not behave like a simple disk when spun at high speeds. The overall shape expands like pizza dough, but the individual fibers stretch at different rates, opening up space between them and stressing the epoxy filler. "We didn't realize those problems in days past," Bitterly says. To design a new wheel, he needed to analyze every spot on the wheel, not just some idealized, disklike shape. "There are so many variables," he says, "it's hard to conceive." The Bitterlys' design program divides a simulated wheel into as many as 50,000 "nodes," or elements, while tracking all the stresses on the fiber at each node. In the old days, when the Bitterlys limited their analysis to simple shapes, it took them a week to work through 15 equations. Their new workstations juggle tens of thousands of equations and can finish the analysis in an hour.