Making jet engines fail, life jackets float, and cars crash by computer
CAR-CRASH tests are impressive looking things. The bumper crumples into the wall. The dummies' heads move forward. Talk about visual impact. Especially in slow motion.
But each crash can cost anywhere from $50,000 to $750,000. And when you're Robert Whirley, technical vice president of a small electric-car manufacturer, you can't send hundreds of vehicles screaming into concrete walls.
So Mr. Whirley of US Electricar relies on a cheaper method. It's called virtual prototyping - a process that came out of the US military's weapons-testing program in the mid-1970s.
On his computer, Whirley displays the original design for a large tray to hold the batteries for his company's electric car. Traditionally, the company would have physically tested the tray, made modifications, tested again, and so on until the part met safety specifications. But using software called DYNA3D the engineers instead crash tested the part on a computer. They went through five designs until they finally met the predicted requirements. Then they tested it in a real crash test.
``The simulations aren't 100 percent accurate,'' Whirley says. But they're pretty close. Instead of six months and hundreds of thousands of dollars to develop the battery tray, US Electricar designed it in two weeks for about $25,000. The simulations also predict what the structure goes through during the split-seconds of a crash, giving engineers a better idea of what goes on than in slow-motion physical tests.
The irony of DYNA3D is lost on no one. The same technology now used to save lives had a much different origin. In the mid-1970s, America's generals needed nuclear weapons that didn't explode on impact. If a pilot launched a nuclear missile, he needed time to fly away to safety.
The answer was DYNA3D, developed by Lawrence Livermore National Laboratory. Although the weapons simulations were classified, the software code was not. So Jerry Goudreau, who heads the lab's methods development group in mechanical engineering, lined up industrial collaborators to test the code.
The result was military-civilian collaboration long before US presidents began pushing the idea. The military got better code, and industry began to realize that DYNA3D had commercial applications. Besides US Electricar, General Motors, Chrysler, Saab, Porsche, Mercedes-Benz, and Volkswagen all use the code or some offshoot of it. Boeing uses it to study what happens to a jet's airframe if debris starts spewing out from a failed engine.
As the costs of computing have come down, engineers have found even more applications - from making air-bag sensors more accurate to improving automatic teller machines' security.
After a highly publicized British train wreck in 1991, railroad officials asked a British mechanical engineering firm to simulate what might have happened during the crash. More recently, the consulting firm has used its own version of DYNA3D to test whether life jackets would keep a person's head out of water.
The company is now looking at modeling the launch of a lifeboat from a platform. If, say, workers need a quick exit from an oil-drilling rig, they could jump into a boat, go down a short ramp, and then free-fall about 20 meters (62 ft.) into the ocean. The software has to address not only the lifeboat's survival but also the impact's effect on the people inside. The computer can test the results in rough seas too dangerous for a live test.
Had the military not pushed the technology forward, Mr. Goudreau says, DYNA3D probably would have taken an additional 10 to 20 years to develop.
Send your comments on this column to CompuServe (70541,3654) or Prodigy (BXGN44A).