Towards efficient and accurate energy yield modelling of bifacial PV systems

There is an increasing interest towards bifacial PV modules because of their potential to reduce the Levelized Cost of Electricity (LCOE) thanks to the additional energy yield obtained from rear side light collection. It is understood that bifacial module energy yield gain over monofacial modules depends on the complex interplay of several installation parameters e.g. ground reflectance, module tilt, installation height, row spacing, climate, system size, etc... Different combinations of these parameters may produce bifacial gains ranging from 5 to 50%. Unfortunately, the deep understanding, as well as design tools are – to date – not established for accurately assessing bifacial gain, causing uncertainty in field performance and difficulties in financing. The complex phenomenon of rear side light collection can only be addressed using physics-based approaches, among which, ray tracing has the highest potential of accounting to all of the important phenomena. Currently, ray tracing is computationally very intensive, which hinders its application for designing and optimizing bifacial PV systems. The present work proposes to use the Daylight Coefficient method (available in the open source Radiance 5 software package), which – next to preserving the detailed scene description and physical representation of light transport from the sky to the PV module surface – promises to reduce computational times by a factor of 1000 approximately. Daily simulations indicate that the time needed for annual irradiance simulations at 1 minute resolution can be reduced from 6 months (with ray tracing) to only 2 hours (with Daylight Coefficients). Analysis at specific days shows that Ray tracing and Daylight Coefficient-calculated front side irradiance values are virtually identical and rear side irradiance values differ only by 1-2%.