This is like creating a very detailed weather-and-energy map for every building in a city, but instead of treating all buildings the same, it looks carefully at where each one is, what surrounds it, and how that place behaves. That “sense of place” is then used to better predict how much energy buildings will use.
Traditional urban building energy models often over- or under-estimate real energy use because they rely on generic assumptions and ignore local context (urban form, microclimate, building types by neighborhood). A place-based approach improves prediction accuracy for city-scale planning, retrofits, and energy policy design.
Domain-specific datasets and methodologies that link geospatial/urban form characteristics with building energy behavior; methodology can become a defensible asset when combined with proprietary city data and calibration against real utility readings.
Classical-ML (Scikit/XGBoost)
Time-Series DB
High (Custom Models/Infra)
Data availability and quality for both building stock (geometry, use, envelope) and measured energy consumption across many buildings and years; plus computational cost of simulating or learning from large urban-scale datasets.
Early Adopters
Focus on a ‘place-based’ perspective—explicitly encoding neighborhood/urban context into energy models—rather than only modeling individual buildings or using one-size-fits-all archetypes, which can materially improve accuracy for planning and policy use in specific cities.