1. Home ›
  2. Research ›
  3. BUILT2LAST

BUILT2LAST

BUILT2AFFORD – Big Data Enabled Energy-Efficiency and Health Assessments to Provide Affordable Housing

Our research objective is to co-develop, test, and validate BUILT2AFFORD—a framework and tool designed to preserve affordable housing through low-cost, energy-efficient retrofits. Utilizing Machine Learning and Google Street View Images, BUILT2AFFORD will pre-identify passive retrofit strategies, demonstrated through the retrofitting of eight testbeds to showcase its viability.

Five-panel image showing different perspectives of community resilience. Top left: Marmain Apartments. Top right: Two people review paperwork indoors. Bottom left: A person gestures from a rooftop while another stands inside by a window. Bottom right: A Red Cross volunteer speaks with a resident inside, next to a window draped with an American Red Cross blanket. An exterior view of an older brick building is visible in the bottom right corner.

AI-Augmented Environmental Assessment for Sustainable Building Design

This research develops a methodological framework that integrates BIM, LCA, and artificial intelligence to support sustainable building design. The framework automatically extracts material data from Revit models, maps them to life cycle inventory processes in OpenLCA, and performs midpoint and endpoint environmental assessment across user-defined time horizons. An LLM-based chatbot is integrated to help users query results, compare scenarios, and receive plain-language interpretations, making environmental assessment more accessible during early design decision-making.

Machine Learning – Life Cycle Assessment – Environmental Product

We developed an AI-powered system to evaluate the environmental impact of construction materials by analyzing thousands of Environmental Product Declarations (EPDs). Our approach automates data collection, fills in missing information, and helps make sustainability assessments faster and more reliable. This work supports smarter material choices and promotes a more sustainable construction industry.

Machine learning flow step 1: API-based Data Retrieval diagram. Two databases, labeled
Machine learning flow chart step 2: Data Extraction Modules diagram. LLMs feed into three data containers labeled Table, Text, and Image. A light blue arrow points right, indicating data flow.
Machine learning flow chart step 3: Data Preprocessing steps: Normalization, Categorical Encoding, and Outlier Detection, shown in light blue rounded rectangles within a dashed teal border, with an arrow pointing right from Categorical Encoding.
Machine learning flow chart step 4: Data Imputation process. ML Techniques feed into K-Nearest Neighbors and XGBoost, which in turn create a robust database, indicated by a green check mark.