Enabling Scalable and Efficient Building Energy Simulation through Scan-to-BIM

Buildings account for approximately 40% of primary energy use and one-third of carbon emissions in the United States, underscoring the urgent need for energy-efficient design and operation of buildings. Building energy simulation is a key tool for optimizing performance at various scales, from individual buildings to city-wide planning. However, these simulations require detailed building information, including geometry, materials, components, orientation - which traditionally involves labor-intensive, error-prone manual data entry. This challenge is particularly significant for existing buildings where data acquisition demands time-consuming surveys, data collection, and manual data entry. Scan-to-BIM (Building Information Modeling) has emerged as a promising method to generate precise digital models from point cloud scanning data, yet conventional workflows remain resource-intensive and difficult to scale for large applications. This study introduces a novel deep-learning based Scan-to-BIM framework that automates the transition from point cloud data scanning to building energy simulation modeling. The framework employs advanced algorithms to extract architectural features and generates accurate building models with detailed geometry, significantly improving efficiency and scalability. The proposed method was validated through a case study, achieving an average vertex error of 0.0956 meters and a 3D Intersect ion over Union (IoU) of 91.81%. These results demonstrate the potential of the framework to deliver efficient, scalable, and accurate building energy analysis while reducing manual effort. Beyond energy simulation, the streamlined Scan-to-BIM process has broad applications in decommissioning, structural analysis, facility management, energy auditing, urban planning, and smart grid modeling, offering substantial advancements for the building sector.