Design Engine
Kathy Velikov, Kazi Najeeb Hasan, Matias del Campo, Ruxin Xie, Lucas Denit, Brent Boyce · ACADIA quarterly · 2021
Generative design offers the possibility to heuristically explore data-driven design iterations during the design process. This enables performance-informed feedback and the possibility for exploring viable options with stakeholders earlier in the design process. Since architectural design is a complex, nonlinear process that requires trade-offs and compromises among multiple requirements, many of which are in conflict with each other, a multi-objective solver provides a spectrum of possible solutions without converging on a single optimized individual. This enables a more informed design possibility space that is open to collaborative decision-making. This paper describes the development of a custom multi-objective generative design workflow to visualize families of possible future building typologies with a focus on the impact of site, form, envelope performance, and glazing. Three future design scenarios are generated for three urban U.S. locations projected to grow and where progressive environmental performance stretch codes have been adopted. Drivers such as plausible site, procurement, financing, value chain, and construction typology inform possibilities for built form, envelope technologies, and performance in relation to local codes, environment, and occupant health, are transformed into design inputs through urban, spatial and environmental simulation tools for a "building design generator," or a multi-objective optimizer tool that produces an array of possible building massing and schematic envelope design options. The paper concludes with pointing out some of the gaps in data of current evaluation tools, the need for interoperability across platforms, and this points to multiple trajectories of future research in this area.