Current approaches in Computational Design Synthesis enable the human designer to explore large solution spaces for engineering design problems. To extend this to support designers in embodiment and detail design, not only the generation of solutions spaces is needed, but also the automated evaluation of engineering performance. Here, simulation methods can be used effectively to predict the behavior of a product.
This paper presents a generic approach to automatically generate solution spaces for energy- and signal-based engineering design tasks using first-order logic and Boolean satisfiability. These solution spaces not only include the graph-based product concept topologies but also corresponding bond-graph based simulation models. To do this, guidelines to create partial simulation models for the available building blocks for the synthesis are presented, to assure a valid causality in the final simulation model. Considering the connections in the graph-based product concepts, the simulation models are automatically generated and, simulated. The simulation results are then stored to enable a more informed decision as to which concepts to pursue in detail design. The method is validated using automotive powertrains as a case study. 162 different powertrain concepts are generated and evaluated, showing the advantages of electric powertrains in respect to CO2 emissions and the importance of intelligent control strategies for hybrid ones. This research enables the generation, exploration, and evaluation of solution spaces for energy- and signal-based product concept. Guidelines to define compatible bond-graph based partial simulation models that map to building blocks from an object-oriented graph-based knowledge representation are introduced. Additionally, a generic translation between the graph-based product concepts and simulation models is presented.