Ever since computers have been used to support human designers, a variety of representations have been used to encapsulate engineering knowledge. Computational design synthesis approaches utilize this knowledge to generate design candidates for a specified task. However, new approaches are required to enable systematic solution space exploration. This paper presents an approach that combines a graph-based, object-oriented knowledge representation with first-order logic and Boolean satisfiability. This combination is used as the foundation for a generic, automated approach for requirement-driven computational design synthesis. Available design building blocks and a design task defined through a set of requirements are modeled in a graph-based environment and then automatically transferred into a Boolean satisfiability problem and solved, considering a given solution size. The solution is then automatically transferred back to the graph-based domain. The method is validated through the synthesis of automotive powertrains. The contribution of the paper is a new method that is both able to determine that an engineering task is solvable or not given a set of design building blocks and able to systematically explore the solution space.

This content is only available via PDF.
You do not currently have access to this content.