Decision-Based Conceptual Design: Modeling and Navigating Heterogeneous Design Spaces

Abstract

Information gathering and refinement are critical activities in conceptual design. A decision-based framework is developed consisting of three main components: a flexible, extensible design space model based on a Gaussian kernel which synthesizes information from design instances; expected value decision-making which focuses the design process on the most promising subspaces within this model; and information value theory which identifies uncertainty in the design evaluation whose reduction could redirect the design process. Together, these components form a normative method for conceptual design around a key process—the co-evolution of a design and the evaluation model used to quantify its value. Formalizing conceptual design toward reducing arbitrary design decisions and focusing attention on the most critical design concerns holds the potential to substantially improve both the process and product of design. The proposed methodology is demonstrated through an example in the domain of electric motor selection.