One of the most important decisions in the product development process is the selection of a promising design concept because of the large influence it has on the final product. A thorough search for the best design is a significant challenge to designers, who are trying to balance the objective and subjective performance of the designs they create. In this thesis, a computationally-assisted design methodology is developed and used in the early stages of design to more thoroughly search for designs that perform well according to objective physics-based models and subjective designer-specific preference-based models. The method presented herein uses an initial pool of user-created designs that is parameterized and used in a numerical search that recombines design features to form new designs in a semi-automated way. Designs are then evaluated quantitatively by objective performance calculations and evaluated qualitatively by human designers. Designer preference is interactively gathered when visual representations of new computer-created designs are presented to the designer for subjective evaluation. A mathematical model is then formed using statistical probability methods to approximate the designer's preference and incrementally updated after the designer subjectively evaluates a new set of designs at each iteration of the automated search process. The methodology uses a multiobjective approach to search for optimally performing designs, treating both the physics-based models and the preference-based models as objectives. The methodology couples the speed of computational searches with the ability of human designers to subjectively evaluate unmodeled objectives. The method is demonstrated with two product examples to find optimal designs that designers may not have otherwise discovered among the vast number of possible combinations of features. The proposed methodology brings the ability to search for and find numerous, optimal solutions across a wide solution space, in an efficient and human-centered way, and does so in the early stages of design.



College and Department

Ira A. Fulton College of Engineering and Technology; Mechanical Engineering



Date Submitted


Document Type





conceptual design, concept generation automation, multiobjective optimization, preference capture, Garrett Barnum