Shape Grammar Composition and Salient Feature Identification for a Product Family
Open Access
- Author:
- Culbertson, Timothy David
- Graduate Program:
- Industrial Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- None
- Committee Members:
- Timothy William Simpson, Thesis Advisor/Co-Advisor
- Keywords:
- product family
style
similarity
shape
aesthetics
feature identification
freeform surface
engineering design
industrial design
design automation - Abstract:
- In modern product design, product form and aesthetics play a major role in consumer preference and product differentiation. Because product form design traditionally requires designers with a level of aesthetic skill and intuition, common practice is to entrust product form considerations to industrial design personnel. The industrial design process, however, is highly subjective and does not translate well across mechanical design disciplines nor is it necessarily consistent across industrial designers. In addition, due to economic considerations, companies are increasingly delegating industrial design activities to engineers with little formal industrial design training. An automated or semi-automated approach for producing new designs with a consistent style would be desirable to address these issues and reduce a design’s sensitivity to the background of the designer. This thesis presents the development of such an approach. Shape grammars have recently arisen as methods for producing designs with a coherent style with the ability to control the variation of the output shapes. A shape grammar for a family of medical ultrasound transducers is first developed, producing a family of shapes with a common architecture in a commercial CAD package and applied to freeform surfaces without maximal line representation. The question of shape parameter impact on product style is also examined, specifically for cases where the number of design artifacts on which to base a style is few and no formally-trained industrial design personnel are available. A survey-based approach is proposed wherein the impact of a shape parameter on product style is evaluated by comparing design variants to a baseline design and having respondents rate the style similarities on a Likert-like scale. Candidate parameters are screened for aesthetic significance using fractional experiment designs, identifying four significant variables. This screening is validated using a full factorial experiment with ultrasound transducer designers and engineers. The relative weights of the four variables are evaluated in a series of factorial experiments. Consistent main effect and interaction effects are observed within product classes; comparisons across product class showed no statistically significant common factors. Respondent demographics were also investigated. Minor correlation improvements were found among some demographics such as subjects with artistic backgrounds, but survey sub-populations were not large enough in size to warrant statistically-backed conclusions. Subject familiarity with the product line was found to significantly improve response consistency. Follow-on experiments validated the relative weights of the four variables investigated; one variable is found to dominate style. A second follow-on validation experiment confirms observations that non-linearity within the resulting model is significant and that model application is only valid within the variable ranges studied.