Human-Involved Optimization-Aided Design Exploration: A user-configurable direct-constraint strategy for computational design optimization

This study explores the use of direct-constraint strategies in computational design optimization to improve early-stage architectural design exploration and increase designer involvement in the design process. Specifically, it presents a user-configurable direct-constraint strategy that incorporates two editable direct constraints within a pre-built parametric model for generating residential building layouts. The constraints control the building plot subdivision and building height limitations and are designed to be flexible and configurable externally by users, allowing designers to easily customize the parametric model to incorporate design intents related to spatial arrangements, landscape, and other subjective objectives. By facilitating the customization of the parametric model through the user-configurable direct constraints, the designer can be more engaged in the design exploration by iterating various design strategies or concepts. A case study based on a real-world design project is presented to demonstrate how designers can use these direct constraints to explore different alternative design directions, as well as how the optimization result can provide designers with feedback pertinent to the performance trade-offs characterizing the design problem. Finally, the paper discusses the relevance of the study to architecture research, practice, and tool developments while indicating future research directions.