Bottom-up mapping of urban building stock high-resolution energy profiles: A case study of the City of Melbourne, Australia

• A citywide demand simulation model is generated and verified. • 79 archetypes are developed to capture Australia's diverse building stock. • Simulations offer detailed temporal, spatial, and end-use granularity across scales. • Modelled citywide demand is 7.9 % higher than the reported aggregate consumption. • Methodology supports nationwide replication using publicly available data. This study presents a high-resolution, bottom-up urban building energy model for the Australian context, underpinned by a comprehensive archetype library and demonstrated using the City of Melbourne. Addressing the absence of spatially explicit and temporally granular simulations in Australia, the model estimates city-wide building energy demand using 79 archetypes spanning typologies, construction eras, and height classes. An automated workflow was established to extract GIS building footprints and heights, integrate non-geometric attributes, and run simulations with EnergyPlus as the core engine. The total annual energy demand is estimated at 4,876 GWh. Average annual energy use intensities (kWh·m⁻²) are highest for hospitals/clinics (376.2), followed by retails (213.0), commercial accommodations (210.5), houses/townhouses (143.0), offices (136.6), educational/research buildings (86.0), and residential apartments (73.8). Compared with reported references from yearly energy records, official reports, and public databases, the simulated city-scale total is 7.9% higher than the reference value. At the typology level, deviations generally fall within ±20.0%, except for retails (30.5%). The results reveal typological and spatial disparities: offices and residential apartments jointly account for over half of city-wide demand, while Melbourne CBD, Docklands, Parkville, and Southbank are major energy hotspots, collectively contributing around 70.0% of the city-wide total. At finer temporal resolutions, daily demand fluctuates between 7.2 and 18.5 GWh and hourly demand ranges from 0.2 to 1.5 GWh, showing patterns characteristic of a temperate climate and a pronounced diurnal cycle aligned with occupancy and operational schedules. A Monte Carlo based uncertainty analysis shows that input variability yields well-bounded uncertainty ranges, remaining within 25.2% across typologies and 20.8% across spatial scales, indicating stable model performance. By generating fine-grained, end-use-specific energy profiles, the model provides a robust analytical foundation for building energy planning and management, with potential to support Australia’s net-zero transition, and offers an extensible platform for future methodological advancements and scenario-based analyses.