Analyzing Generative Design Grammars

Abstract

In the last decades, research on computational design synthesis using generative design grammars has achieved great improvements due to advanced search and optimization strategies. Even though meaningful grammars are inevitable to generate valid and optimized designs, only little attention has been paid to improving the development of grammar rules. The research presented in this paper focuses on supporting human designers in developing better grammars. The presented Grammar Rule Analysis Method (GRAM) supports a more systematic development process for grammar rules. It enables the rule designer to gain detailed knowledge of the performance of grammar rules, their relations to objectives, constraints and characteristics, and their interaction. The method’s goal is to have a major impact on the quality of the generated designs by improving the quality of the rules. The case study uses two different grammars for automated gearbox synthesis to validate GRAM and show its potential.