The "optimum" point represents the best possible combination of factors that will achieve reduced emissions of both soot and NOx. The "baseline" point represents the best existing technology.
|
Computer models developed at the University of Wisconsin-Madison are using genetic algorithms to simultaneously increase fuel efficiency and reduce pollution. Peter Senecal, a post-doctorate engineer at UW-Madison, created the computer models to help sort through literally billions of factors that determine engine performance, an enormous task for conventional computer simulations.
According to Senecal, the most important advance is in improving pollution emissions without sacrificing fuel efficiency, and vice versa. Normally, engine designers who concentrate on solving one problem end up with major tradeoffs in another. Using a Silicon Graphics supercomputer at UW-Madison's Engine Research Center (ERC), Senecal created a diesel-engine design that reduced NOx emissions threefold and soot emissions by 50% over the best available technology. At the same time, the model reduced fuel consumption by 15%.
Six engine-performance measures were studied, including fuel injection timing, injection pressure, and amount of exhaust gas recirculation (EGR). The simulation was then reproduced experimentally in an actual diesel engine housed at the ERC. "We found that the agreement was excellent between what was measured in the lab engine and what the computer predicted," said Senecal.
Senecal addressed the topic in June at the CEC/SAE Spring Fuel & Lubricants Meeting and Exposition in Paris. His work is of interest to the engine manufacturing industry, which faces major new federal pollution control mandates by the year 2002. Caterpillar Inc., the Peoria, IL-based manufacturer of diesel engines for trucks and heavy equipment, is funding Senecal's post-doctorate work, which will focus on improving the geometry of engines. Genetic algorithms have been developed in recent years for other engineering challenges, such as designing bridges and airplane wings. "I kind of stumbled onto this in the literature, and wasn't sure if it would work for something as complex as engine design," he said.
Senecal begins with five "individuals," which are defined as one distinct set of the six engine parameters. Four of the individuals are randomly selected and the fifth is the baseline, or best known set of parameters.
Next, a computer model is used to weed out the best parameters of the first group. The two fittest "parents" are then allowed to "reproduce" and a new generation is formed, complete with "mutations" that represent marked improvements over the previous generation. The process is continued through successive generations until the computer identifies the most "fit" member of the group.
Senecal says this process narrows the field of potentially one billion calculations on the computer down to 200 to 250 of the best possibilities. The computer can accomplish in weeks what would otherwise take decades to run.
Mechanical Engineering Professor Rolf Reitz, Senecal's Ph.D. thesis advisor, says the computer model is extremely versatile and could be used for all types of engines. While current work focuses on questions such as fuel injection and air intake, studies of engine hardware are just beginning. Reitz says the typical engine piston, for example, has not been fundamentally improved upon for decades. Yet some engineers have no idea whether a different geometry could produce much better engines.
If engine manufacturers want a more powerful engine, or a more durable engine, the genetic model can be programmed to find those traits, too. "If you want your children to be long jumpers, high jumpers, or sprinters, you can specify these attributes with this program," Reitz said.
The diesel engine industry faces a U.S. Environmental Protection Agency mandate to cut NOx emissions in half by 2002. Wisconsin's small engine industry, also facing pollution-control deadlines, has initiated a research program at UW-Madison using the genetic model.
Jean L. Broge
AEI September 2000
For more information, circle 220 & 221
