Solar Resource Estimation using an Integrated Radiance Model with LiDAR Defined Geometries and Shading

摘要

Microgeneration has already demonstrated substantial potential to become a significant contributor to the UK's energy mix and could form an important part of meeting the UK's 15% renewable energy by 2020 target set by the European Commission. Of the microgeneration technologies available, solar photovoltaics (PV) has shown the most potential to meet energy demand. At the end of June 2013, the UK could boast 1.6GW of installed solar PV capacity from over 447,000 installations. As the market has grown so too has interest in the viability of PV at individual properties which has led to an increasing number of academic and industry studies of solar resource that have utilised the EU JRC PVGIS webtool or the solar radiation toolset within the market-leading GIS software, Esri ArcGIS. Building from these studies, a methodology to predict solar resource is presented that combines an integrated radiance model, a building geometry model and a shading model. Accurate appraisal of solar resource requires detailed knowledge of a property's geometry and positioning. Digital Surface Models (DSM) are a data source that can be used to estimate this information for buildings across an entire city. However, the DSMs that are currently available are of an insufficient resolution to accurately portray the slope and orientation of buildings that are less than approximately 200m~2 in plan area. This is particularly problematic for city-scale roof shape modelling given that over 70% of properties in Sheffield, a typical UK city, fall below this threshold. The neighbouring buildings method developed by Gooding et al. (in press) effectively increases the LiDAR spatial resolution by combining LiDAR data from nearby rooftops that share identical geometry. Therefore, this method is used to determine South-facing area, slope and orientation to then inform the integrated radiance model of Smith et al. (in press) with an adaptation for shading effects. Shading is incorporated by generating hemispherical viewshed models derived from a DSM for each prospective installation. The DSM is searched in 32 directions from the location of a prospective PV installation to determine the maximum angle of sky obstruction, or horizon angle. Horizon angles are then converted into a hemispherical coordinate system, thus representing a three-dimensional hemisphere of obstructions as a two-dimensional image. The hemispherical viewshed pixels are binned in the same manner as the radiance model outputs and are defined as either unobstructed or obstructed. The fraction of unobstructed pixels to the total pixels in that bin is used to calculate a skyview fraction for each bin. This information is used to appraise the direct and diffuse radiation reaching the point of interest. The outputs of the model are validated using data from seven sites across Leeds, UK and show good agreement at 5.83% and 0.79% mean error under 0.8 and 0.75 performance ratios. The method outperformed both the Esri ArcGIS and EU JRC PVGIS methodologies. Esri ArcGIS incurred -12.43% and -17.49% mean errors under the 0.8 and 0.75 performance ratios whilst the results for EU JRC PVGIS were 13.5% and - 6.4% mean error respectively.