A Simulation-Driven Parametric Building Envelope Façade Design Using Generative AI and Virtual Reality in the Early Design Phase

Recent advances in generative AI have significantly impacted architectural design, enabling the exploration of complex tasks such as site layout planning, interior design, and exterior design. While architects recognize the potential of generative AI, personal barriers often hinder its widespread adoption in the field. Understanding the principles and applications of generative AI is crucial to overcoming these challenges and improving both the sustainable and functional aspects of design outcomes. This study aims to enhance design productivity by digitizing early-stage processes, such as Bubble Diagrams, into graph-structured datasets and developing a Generative Design algorithm. The algorithm automates the creation of space layouts by quantifying design elements like spatial adjacency and environmental factors, defining building mass outlines, and generating floor plans complete with walls and doors. Additionally, the paper integrates AI-generated floor plans with parametric design tools to optimize building performance during the early design phase. This process involves generating 3D geometry from spatial layouts, organizing them into multi-level floor plans, and using Rhino and Grasshopper to create building envelope facades. The facades are then simulated with Honeybee to evaluate performance, producing multiple design iterations. These iterations are visualized in a virtual reality (VR) environment, where a variant manager enables real-time interaction and analysis of various parametric building envelope designs. By leveraging these tools, architects can explore complex forms, resulting in more sustainable, contextually relevant designs that enhance user experiences.