Leveraging large language models to enhance urban building energy modeling: A case study

In light of ambitious carbon neutrality targets by 2025, urban building energy modeling (UBEM) has become a promising method for reducing energy consumption and evaluating retrofitting strategies in urban environment. Establishing UBEMs at a large scale, however, faces multiple challenges including limited data sources, specialized building science requirements, and complex calibration processes. These make building modeling labor-intensive, hindering its practical applications. This work presents an innovative method to address these challenges by enhancing UBEM significantly using large language models (LLMs). We explore the potential of LLMs to streamline data acquisition, preprocessing, and preliminary overview of building and energy datasets, while also translating natural language building descriptions into formal UBEM models, ultimately aiding model creation, error detection, calibration, and retrofit scenario analysis, offering a more nuanced understanding of potential energy-saving strategies. The application of LLMs in UBEM was demonstrated through a case study involving 200 low-rise residential buildings in Montreal, Canada. By integrating LLMs into the UBEM workflow, we contribute to the advancement of UBEM methodologies, potentially accelerating the adoption of energy-efficient practices in urban planning. The findings suggest that LLMs can significantly enhance the accessibility, accuracy, and interpretability of UBEMs, ultimately supporting more effective decision-making in urban energy management and carbon reduction efforts.