A scalable and efficient framework for city-scale building energy modeling with microclimate considerations

Abstract

Urban building energy modeling (UBEM) is essential for supporting urban energy efficiency assessment and low-carbon policy making. This study proposes a scalable and efficient UBEM framework and applies it to Nanjing, China. Multi-source urban data were integrated to construct a unified urban building geodatabase. Building archetypes were developed based on building type and construction year, and modified weather files for each block were generated using the Urban Weather Generator (UWG). Through an automated workflow combining Python scripting and parallel computing, annual building energy simulations were conducted for 49,793 buildings across 1693 urban blocks. The comparative analysis reveals that neglecting microclimate effects results in an 11.4% underestimation of cooling demand and a 10.5% overestimation of heating demand, with the largest deviations occurring in high-density districts. The results also indicate notable variations in average energy use intensity (EUI) across building types, with healthcare and commercial buildings exhibiting the highest demand. Spatially, high-rise clusters in newly developed areas and large public facilities form major energy hotspots, whereas older low-rise residential areas show lower overall demand. A comparison with measured data showed that the simulated EUIs of public buildings were within ±20%, confirming the reliability of the framework. The proposed approach completed city-scale simulations within approximately 36 hours on standard hardware, highlighting its scalability and computational efficiency. Overall, this study provides a practical UBEM framework for identifying spatial patterns and energy hotspots of building energy use at the city scale, supporting energy-efficient urban planning and targeted mitigation strategies.