Composition A design for an engineered object is usually in terms of its parts, their attributes, and the means by which they are linked together. We generate a large number of design candidates by combining parts in different ways, and by setting their attributes (size, etc.) differently. Our simulator is capable of taking descriptions of the parts and their links, in terms of equations that may include differential calculus and discrete mode changes, and calculating the performance of the resulting composed device with respect to some environment in which it is deployed. If simulated environments correspond to a diverse set of mission profiles, then design choice can be driven by the operational requirements.

Selection Our interactive visualization environment allows the designer to see the performance tradeoffs between the different designs. For example, Aetion applied this technology to the design of hybrid electric Army trucks, where performance criteria include fuel efficiency and range on different surfaces (highway, mud, etc.), top speed, time to accelerate to various speeds, how various cargo loads affect dynamics, etc. As well as showing the designer the range of performance that can be attained by considering different designs, our environment also allows engineers to investigate the relationship between design choices and the resulting performance. For example, designs that cluster by some aspect of performance may all have some design option in common, and some design options may never be included in the acceptable or the best designs because they compromise performance too greatly.

Possibilities A wide range of engineering design needs is amenable to our approach of automatically composing a simulatable design based on varying the simpler parts, their attributes and their arrangement. For instance, the control strategy for hybrid electric vehicles is a matter of design: how it should choose when to draw from which energy source, or which to replenish. Typical sources under consideration include internal combustion engines, turbines, supercapacitors, fuel cells, and various types of battery. Thus, in investigating control strategies, our system is actually allowing the user to design a piece of software. In other projects, this system has been applied to simulation and tradeoffs for gear trains, and for antennas. The Defense Advanced Research Projects Agency funded the original invention of our patented approach to engineering design.