You can easily take any engine and make it a very low emission engine, but if you reduce one kind of pollutant, you increase another kind of pollutant.
As concerns about global warming mount, restrictions on carbon dioxide emissions seem inevitable. While regulation will eventually set the standards, engineers like Assistant Professor of Mechanical Engineering Indranil Brahma will have to design the trucks, ships and even wind turbines to meet those standards.
Although his work has so far been directed at engines, Brahma doesn't necessarily design better engines. Instead, he designs better methods for designing better engines.
"What I realized in industry is the major challenge is to optimize the system," says Brahma, who worked for four years for worldwide engine manufacturer Cummins. "You can easily take any engine and make it a very low emission engine, but if you reduce one kind of pollutant, you increase another kind of pollutant," he says.
In the case of diesel engines, particulate matter, nitrogen oxides, unburnt hydrocarbons and carbon monoxide also need to be kept in check, even as carbon dioxide emissions are reduced. "To try to decrease everything will be a major challenge," he says. To meet that challenge, engineers use computer models rather than physical engines, which would be too expensive and time-consuming to build and tweak for every experiment.
While at Cummins, Brahma wrote software used in all Cummins engines around the globe to reduce emissions and optimize performance. At Bucknell, Brahma is working on incorporating the fields of artificial intelligence and statistics to create models that are robust, accurate and practical for optimization problems. Currently, system design is based on models that take either a traditional approach based on computational fluid dynamics, which requires far too many calculations to be feasible for optimization, or empirical methods, which are faster, but can't be applied to conditions outside the range of the experiment.
"My research is to find a middle path to find ways of modeling the systems which are fast and accurate like the empirical methods, but also have some physics in them so that they are robust and can extrapolate outside the area where we collect the data," Brahma says.
Brahma's methods can be applied to engineering systems generally, and he looks forward to branching out from engines to questions within the alternative energy field, including solar and wind applications.
Posted Sept. 22, 2009