Gaining System Design Knowledge by Systematic Design Space Exploration with Graph Based Design Languages

摘要

The conceptual design phase in the design of complex systems such as satellite propulsion systems heavily relies on an exploration of the feasible design space. This exploration requires both: topological changes in the potential system architecture and consistent parametrical changes in the dimensioning of the existing system components. Since advanced engineering design techniques nowadays advocate a model-based systems engineering (MBSE) approach, graph-based design languages which embed a superset of MBSE-features are consequently used in this work to systematically explore the feasible design space. Design languages allow the design knowledge to be represented, modeled and executed using model-based transformations and combine this among other features with constraint processing techniques. The execution of the design language shown for the satellite propulsion systems in this work yields topologically varied designs (i.e. the selection of a monergol, a diergol or a coldgas system) with consistent parameters. Based on an a posteriori performance analysis of the automatically generated system designs, novel system knowledge (most notably in form of so-called "topology change points ") can be gained and extracted from the original point cloud of numerical results.