abstract
Climate change demands a rapid transition of the energy system towards renewable energy sources. Low levelized costs of electricity and a wide range of integration options combined with high modularity makes photovoltaic energy generation a key technology in this transition. Accurate and at the same time fast and scalable methods for estimating the yield of roof-top and façade PV systems are crucial for identifying the overall PV installation potential and the best locations for PV installations with respect to the energy yield. Within this contribution, we implement a graphics processor accelerated monte-carlo raytracing algorithm for the calculation of the energy yield of both roof-top and façade PV systems with high spatial and temporal resolution. Our algorithm works on geodata supplied in the open CityGML format which are widely available from official sources [1, 2]. Since our algorithm is highly parallelized, it can be applied to large scales within reasonable computation time while maintaining the high resolution. Figure 1 shows the calculated irradiance relative to the direct horizontal irradiance for a sample area ($\approx$ 250 $\times$ 250 m$^2$) in the city of Hanover (Germany). The calculation of figure 1 takes $\approx$ 1 hour on a single graphics card with 108 rays while including up to two reflection iterations. We are confident that the computation time can be reduced down to $\approx$ 20 minutes. Furthermore, by the time of the conference we will have implemented diffuse radiation from the sky. Finally, the PV energy yields are calculated based on the irradiance values using the software package pvlib [3]. Thus, our algorithm is a powerful tool for the evaluation and planning of PV installations on rooftops and façades especially in urban environments.