Designing-in performance through parameterisation, automation, and evolutionary algorithms H.D.S. BEAGLE 1.0

Abstract

Design is both a goal oriented and decision making activity. It is ill-defined by nature as designing includes weighing and understanding trade-offs amongst soft and hard objectives or in other words vague or imprecise and computationally definable criteria and goals. In this regard designers in most contemporary practices face a crisis of sorts. How do we achieve performance or sustainability under these large degrees of uncertainty or with limited design cycle times? Fundamentally design collaborations, teams of domain experts, are not typically given enough time to design-explore, generate design alternatives in order to find or evolve solution quality through expansive design search spaces. Given these limitations of time and the ever more complex criteria for 'designing-in' performance our research approach provides a computational strategy to expand the solution space as well as pre-sort and qualify candidate designs. The research presents a novel methodology and technology framework and an initial implementation that was developed to enhance the human activity of design exploration, domain integration, and further evolve design process for performance goals. The research does so through generating and optimising a highly correlated solution space in conjunction with a near simultaneous evaluation of design alternative fitness. © 2012, Association for Computer-Aided Architectural Design Research in Asia (CAADRIA), Hong Kong.