World Solar Resources for Vertical Building-Integrated Photovoltaic Applications and Beyond |
Yuan Gao, Jacob Jonsson, Charlie Curcija Lawrence Berkeley National Laboratory, Berkeley, CA, United States |
This work introduces a novel approach to predicting the global vertical irradiance (GVI) at various surface azimuths using machine learning models, addressing a critical gap in the photovoltaic (PV) industry. With the growing importance of building-integrated photovoltaics (BIPVs), understanding GVI distribution worldwide is essential for optimizing vertical PV installations in urban environments. Our study utilizes solar radiation data from over 2000 weather stations to train and test multi-layer perceptron regression models. These models generate world maps of daily total GVI, providing valuable insights for BIPV design, particularly in high-rise urban settings. Additionally, we present a new open-source Python tool for generating these GVI maps, which is crucial for future solar energy resource assessments in the building sector. |