The need to increase the use of non-conventional energy sources in order to reduce the energy demand and greenhouse gas emissions is currently one of the world’s main challenges. Buildings need to become much more energy efficient and the energy demand should be primarily (increasingly) satisfied through renewable energy resources (e.g. solar energy).
One way to increase the production of renewable energy is to promote the solar energy deployment in cities particularly through the use of PV technology. To achieve this aim, a comprehensive assessment of solar PV potential is required. Different approaches have been implemented by researchers in order to study the large-scale solar PV potential of rooftops. Among several methodological approaches, the following have been widely used in order to quantify the available potential of rooftop solar photovoltaic panels. The methods include sampling techniques, statistical methods, aerial images and Geographic Information Systems (GIS) together with LiDAR (Light Detection and Ranging) data.
In a recent paper published in Solar Energy, Dan Assouline, Nahid Mohajeri and Jean-Louis Scartezzini from the Solar Energy and Building Physics Laboratory at Ecole Polytechnique Fédérale de Lausanne in Switzerland used a combination of data-driven methods including a machine-learning algorithm and GIS together with LiDAR data to estimate photovoltaic solar energy potential on building roofs. They investigated the rooftop solar photovoltaic potential of about 1901 communes (the smallest administrative division) in Switzerland. They estimated monthly global horizontal solar radiation (diffuse horizontal, global horizontal and extra-terrestrial horizontal radiation) as well as global tilted solar radiation over rooftops.
A support vector regression (a kernel-based machine learning technique) model was developed and its performance evaluated by the root mean square error (RMSE) and the normalized root mean square (RRMSE). The geographical potential which includes the available roof area and shading factors for the installation of solar photovoltaic is estimated. The authors finally provide an estimation of the technical potential of the rooftop solar photovoltaic energy production per month. Assuming 80% performance ratio and 17% efficiency of solar photovoltaic solar panels, the team found an annual photovoltaic power generation of 17.86 TWh which was equal to 28% of Switzerland’s power consumption as at 2015.
They further found that 15% of the investigated communes provided 53% of electricity used in the country. Maximum values of electricity derived from solar photovoltaic were obtained from large cities like Zurich, Bern and Basel. However, the highest photovoltaic electricity production per capita was found in less populated areas. The total available area for PV installation on the rooftops in the urban areas of Switzerland was found to be 328 km2. The total available roof area per capita was also estimated, that is, 41m2/capita. With an annual increase in cell efficiency of the crystalline silicon wafers of 0.3% per year, the PV panel efficiency will increase to 27.2% by 2050. Assuming an increase to 90% of the performance ratio, the rooftop solar PV electricity production for urban areas in Switzerland in 2050 is expected to reach about 32 TWh. This PV electricity production would then provide 37% of the total forecasted (IEA scenario) electricity use in Switzerland in 2050.
Assouline, D., Mohajeri, N., Scartezzini, J.L. Quantifying Rooftop Photovoltaic Solar Energy Potential: A Machine Learning Approach, Solar Energy 141 (2017) 278–296.
Solar Energy and Building Physics Laboratory (LESO-PB), Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland.Go To Solar Energy