Authors: Mr. Ankit Sethi, Abhishek
Abstract: The construction industry contributes approximately 39% of global CO₂ emissions, with embodied carbon—emissions from material extraction, manufacturing, and transportation—accounting for 11%. Traditional life cycle assessment (LCA) tools for estimating embodied carbon are often disconnected from Building Information Modeling (BIM) environments and require manual input, limiting their usability during early design stages. This study presents an AI-integrated BIM framework that enables real-time embodied carbon estimation directly within Autodesk Revit. Using Python-based machine learning models—Random Forest, Gradient Boosting, and Support Vector Regression—trained on structural data extracted via Dynamo, the system predicts carbon values and visualizes results through heatmaps in the Revit model. The Random Forest model achieved the highest accuracy (MAE: 5.4 kg CO₂, R²: 0.93) and outperformed traditional tools like One Click LCA in both speed and precision. The framework enhances decision-making during the design phase and demonstrates strong potential for scalable, automated, and sustainable design practices in the built environment
