Improved urban irradiation simulations using the Pyrano 2-phase model with semantic obstruction segmentation

Urban areas present challenges for accurate and scalable solar irradiation simulations due to the presence of a wide variety of obstructions. To address this issue, we propose a novel urban irradiation modeling method based on the Pyrano 2-phase simulation engine, which introduces a Dynamic Graph CNN (DGCNN) semantic segmentation model to identify solar obstructions. Our demonstration study shows that the novel method outperforms the widely used baseline method in predicting solar irradiation in complex urban environments while separately quantifying the shading impacts of different obstructions. This method has significant potential for effective photovoltaic (PV) shading risk management and decision-making in PV system configuration and sustainable urban design.