Conceptual design of mechanical systems: A representation and computational model

This dissertation presents a computational model for the conceptual stage of mechanical design. Mechanical design is the process whereby abstract specifications for a product, such as a transmission, are iteratively refined until a set of physical plants, often in the form of blueprints, have been developed. The conceptual stage of design, the first step in this process, consists of transforming the required functionality of an artifact into a preliminary arrangement of artifacts to meet that functionality. Conceptual design is a critical step in the product development cycle which is still poorly understood. By the end of the conceptual design most of the project costs have been committed. The success or failure of a product is often determined by its conceptual design: no amount of optimization or manufacturing finesse can overcome an inadequate or poorly developed concept. Existing engineering tools, such as CAD and CAE software, aid only the detailed geometric design tasks. The practical need to develop good conceptual design methods and tools is therefore clear. The aim of this research is to explore a solution method for a limited class of conceptual design problems and provide a theoretical basis for future engineering tools. Function describes the external requirement of a system, what it must do, while behavior describes how those functions are met. Behavior provides an intermediate step between function and form allowing a variety of solution principles to be explored before considering particular components. The hypothesis underlying this dissertation is that behavioral reasoning is an important process in solving conceptual design problems. This allows for the decomposition of conceptual design into a two steps: (1) developing behavior specifications to meet the functionality based on a variety of physical principles and phenomena; and (2) developing configurations of components to meet the required behavior. The purpose of this research is to develop a computational model of conceptual design, and the required representations, based on this two step process. The hypothesis is validated by the model\u27s ability to generate solutions to a class of conceptual design problems