City-District Rooftop Solar Potential Digital Twin Using OSM and ML

How much electricity could the roofs of a dense Indian district produce once the neighbouring buildings cast their shadows - and which roofs and street blocks should a rooftop programme start with? This project builds a 2.5-D digital twin of T. Nagar, Chennai (1.5 x 1.5 km, 2,867 buildings) from open data: OpenStreetMap footprints, building heights from OSM where they exist and otherwise from a Random Forest (compared honestly with the GHSL 100 m height layer), and the PVGIS typical year. Every roof pixel is ray-cast against the neighbours for 225 sun-position bins, the yield follows from a pvlib PV chain and is compared with the classical planning rule (unshaded PVGIS yield x roof area). A gradient-boosted morphology surrogate replaces ray casting for quick what-if screening of new towers. The planner gets a building ranking, street-block potentials, the share of an assumed demand and scenarios - in a web application on five databases. A teaching prototype, not a bankable yield study.

Category: School of Architecture and Planning

Architecture & Planning Project AP07 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.