Design Engine

Design Engine develops a custom multi-objective generative design workflow as a framework for speculating on how future building typologies might be impacted by advances in envelope technology, combined with advances in computational design, and in the context of increasing off-site modular construction on one hand and increasingly stringent local codes for energy performance and human health and well-being on the other. It points to a realignment in future design workflows that become more inclusive of different knowledge domains, and ultimately to a more holistic approach in energy and carbon reduction for buildings. 

Generative design is potentially disruptive to early design decision making since an evolutionary solver is able to generate families of solutions with characteristics based on trade-offs across sometimes contradictory fitness parameters. This method represents a realignment in building design wherein form and environmental performance can be assessed simultaneously, early on in the design process. An evolutionary solver is a tool that uses a non-destructive method of design search using genetic algorithms and a Multi-Objective Evolutionary Solver is designed to balance across two or more goals at the same time. The underlying logic of evolutionary algorithms is inspired by processes of biological evolution, where the inheritance of genes and genomes, and the randomness of mutation, drive the evolution of a species toward a form that is best-suited to a particular ecological niche. 

The workflow developed is built in the Rhino/Grasshopper environment. The tool processes massing geometry for quantitative evaluation using the Ladybug and Honeybee plugins for climate and thermal behavior simulation, which connect Grasshopper 3D to EnergyPlus, Radiance, Daysim, and OpenStudio. The Octopus multi-objective solver is used for the evolutionary optimization of massing schemes. Quantitative information about a selected scheme is aggregated and displayed in a visualization dashboard within the Rhino modeling environment using Proving Ground’s Conduit software.

The project used the tool to develop scenarios to forecast possible future building typologies in the context of advanced energy performance and occupant health codes as well as automation, prefabrication, and modularization in the construction industry. The scenario planning method included data on growing construction markets, projected urban growth, and increasingly stringent local energy codes to inform the choice of focus on three US cities that represented different climate zones: Austin, Texas; Minneapolis, Minnesota; and Washington, D.C. Different fitness goals and construction typologies were selected for each city and the tool was used to demonstrate how multi-objective trade-offs would impact building form. One of the scenarios was chosen to be developed in further detail.

This research anticipates a future of design decision-making that will increasingly depend on complex models that can optimize simultaneously across several parameters to shape informed early design decisions. In the process, we discovered that models are only as accurate as the information and data they contain, and currently there are real gaps in data available to designers, especially in terms of embodied carbon and life cycle indicators for high performance material assemblies. Fragmentation and the idiosyncrasy of the construction industry makes costs almost impossible to anticipate. The tool was useful in highlighting the need to improve databases and simulation tools, through collaboration across architects, manufacturers, the construction industry, and software developers.