Abstract
Computers are playing an increasingly important role in design and engineering. However, we currently lack a process for developing these computer tools, instead relying on expert experience and ad-hoc experimentation. In this thesis, we develop a design process for design automation, based on a study of the behavior of expert design automation practitioners. In addition to this process, the study finds several strategies used by those design automation practitioners, as well as the most common challenges faced in developing design automation tools. To validate the study, we apply the process to several engineering and design efforts, across the design process; we develop a tool to generate computer-aided mind maps for ideation, we develop a simulator and design exploration tool for rocket engine fuel grains, and we reduce the amount of physical testing necessary to develop novel wheg designs for a climbing microrobot. Finally, to address one of the major challenges facing design automation found in our study, we develop a tool to convert unstructured engineering component data into a structured, searchable form. Through all this, we show the potential for our design process for design automation to significantly, positively impact the design process.