Projecting city-scale energy performance under future weather scenarios through urban building energy modelling incorporating cluster archetypes

• Introduce a scalable UBEM method based on morphology-driven cluster archetypes. • Capture urban form and shading to improve heating and cooling demands by 6% and 3%. • Simulate future demand using both typical years and extreme summer weather data. • Reveal a 36.7% drop in heating and 29.8% rise in cooling demand by the 2080s. • Enable policy support by linking UBEM results to local urban planning units. Accurate modelling of urban building energy demand remains a critical challenge for large-scale planning under data constraints. Conventional methods often struggle to balance physical detail with spatial scalability, particularly when applied to complex urban environments. This study proposes a scalable modelling framework that leverages representative building cluster archetypes, capturing the collective morphology and functional attributes of spatially coherent building groupings. The framework is applied to Greater London, simulating heating, cooling, and electricity demand under both current and projected climate conditions. Under present-day scenarios, the model yields total energy demand within 3.4% of officially reported values, supporting its validity at the urban scale. Compared to aggregating individual building simulations, the cluster-based approach produces approximately 6.1% higher heating demand, primarily due to reduced solar access from intra-cluster shading, while cooling estimates remain within 3% of each other. Future climate projections indicate a decline in total energy demand of approximately 15% by 2050 and up to 26% by 2080, primarily driven by reductions in space heating. Taken together, these results indicate accuracy and scalability in the selected case, while generalisation beyond this context requires local data, re-derivation of archetypes and calibration to city-specific conditions. Accordingly, the framework is presented as a potentially transferable tool that could support long-term scenario modelling and integrated urban energy planning when adapted to local datasets and practices.