Operative generative design using non-dominated sorting genetic algorithm II (NSGA-II)

Massing studies during the early stages of architectural design play an essential role in determining the final building’s performance across design objectives. This paper aims to answer the question: How can early-stage architectural design workflows be translated into a generative design process to create valuable massing solutions? In response, a new application of the Non-Dominated Sorting Genetic Algorithm II (NSGA-II) using the Pymoo framework is proposed for the field of Operative Design. Nine experiments are discussed that test the algorithm’s geometry optimization capabilities based on objective functions reflecting common architectural design goals, including Floor Area Ratio (FAR), Non-Passive Zone (NPZ), Roofs and Best Oriented Surfaces (RBOS), and Usable Open Space (UOS). Selected cases are visualized among non-dominated solutions in each experiment demonstrating the trade-offs between different objectives while programmatically generating successful building designs. In the future, the proposed generative design workflow can be implemented to run optimizations independently from other software within immersive environments. •A generative design workflow is proposed based on Operative Design in architecture.•Design operations including carving and expansion are integrated into an NSGA-II algorithm.•Pymoo in Python, a multi-objective optimization framework, is used to apply NSGA-II.•Nine different experiments are proposed using FAR, NPZ, RBOS, and UOS objectives.•The Non-dominated design solutions were hybrids of carving and expansion, and cases with only either carving or expansion.