Heuristic-guided solution search through a two-tiered design grammar

Lucas Puentes, Jonathan Cagan, Christopher McComb

Research output: Contribution to journalArticlepeer-review

9 Scopus citations


Grammar-based design is typically a gradual process; incremental design changes are performed until a problem statement has been satisfied. While they offer an effective means for searching a design space, standard grammars risk being computationally costly because of the iteration required, and the larger a given grammar the broader the search required. This paper proposes a two-tiered design grammar that enhances the computational design generation with generalized heuristics to provide a way to more efficiently search a design space. Specifically, this two-tiered grammar captures a combination of heuristic-based strategic actions (often observed in human designers) and smaller-scale modifications (common in traditional grammars). Rules in the higher tier are abstract and applicable across multiple design domains. Through associated guiding heuristics, these macrorules are translated down into a sequence of domain-specific, lower-tier microrules. This grammar is evaluated through an implementation within an agent-based simulated annealing team algorithm in which agents iteratively select actions from either the higher tier or the lower tier. This algorithm is used in two applications: truss generation, which is commonly used for testing engineering design methods, and wave energy converter design generation, which is currently a relevant research area in sustainable energy production. Comparisons are made between designs generated using only lower-tier rules and those generated using only higher-tier rules. Further tests demonstrate the efficacy of applying a combination of both lower-tier and higher-tier rules.

Original languageEnglish (US)
Article number011008-1
JournalJournal of Computing and Information Science in Engineering
Issue number1
StatePublished - Feb 2020

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Science Applications
  • Computer Graphics and Computer-Aided Design
  • Industrial and Manufacturing Engineering


Dive into the research topics of 'Heuristic-guided solution search through a two-tiered design grammar'. Together they form a unique fingerprint.

Cite this