Complex engineered systems are core to today’s integrated society. Take the 85,000 aircraft flights each day over the United States, for example. This travel comes with a cost – aircraft noise disrupts communities and aviation is responsible for 2 percent of carbon dioxide emissions. Significant reductions in fuel burn and noise level are theoretically possible, but it is very difficult to develop new designs that can achieve these theoretical limits.

Evaluating the performance of any one aircraft subsystem requires a large computer simulation or expensive physical experiment, and overall quality depends on how that subsystem interacts with all the others. Consequently, design today largely focuses on incremental improvements to existing designs. We need to be able to use our scientific tools to enable radical design in order to ensure performance, safety, and reliability, but we cannot blindly add advanced simulations into our current design tools. Our work seeks to bridge this gap by using the many sources of available information to efficiently find good designs using simulations we trust.