Benchmarking irradiation models for photovoltaic applications: A comparative analysis of radiance-based tools
Erlend Hustad Honningdalsnes, Erik Stensrud Marstein, Magnus Moe Nygård, Marie Syre Wiig, Heine Nygard Riise
Abstract
• Radiance-based ray tracing tools are compared for bifacial photovoltaic applications. • Annual modeled front + rear irradiance within 1.3 % of measured for all models. • Ray tracing model architecture choice can boost speed while maintaining high accuracy. • Ray tracing stochasticity varies among tools and should be accounted for. This study compared the accuracy, efficiency, and capabilities of three ray tracing tools—bifacial_radiance, ClimateStudio, and Honeybee Radiance—in modeling irradiance for photovoltaic (PV) systems. Despite similar performance in modeling hourly irradiance and monthly and annual cumulative irradiation levels, the tools exhibited differences in functionalities and computational efficiency. All models estimated cumulative annual front irradiation within 1.3% of measured values, with bifacial_radiance showing the largest error. For rear side irradiance, bifacial_radiance, ClimateStudio, and Honeybee Radiance underestimated irradiance by 2.82%, 5.40% and 8.74%, respectively. bifacial_radiance showed the highest model stochasticity in cumulative simulations due to its reliance on a single cumulative sky model. In contrast, ClimateStudio and Honeybee Radiance employ matrix-based approaches with hourly resolution within the period, averaging out stochasticity and making them more dependable for precise cumulative irradiation values. These matrix-based methods also significantly enhanced time-series analysis efficiency by modeling yearly irradiance with hourly resolution faster than bifacial_radiance analyzes a single point-in-time, while maintaining similar hourly error and bias levels. However, bifacial_radiance exhibited lower model stochasticity in hourly simulations. Enabled by the efficient simulation capabilities of ClimateStudio and Honeybee Radiance, a method for improving rear side irradiance modeling was tested. By combining simulations with different ground albedo values, this approach significantly reduced error and bias in rear side irradiance modeling. Ultimately, the choice of tool depends on study requirements, with bifacial_radiance being advantageous for hourly and sub-hourly resolution in short-term analyzes, while ClimateStudio and Honeybee Radiance are better suited for cumulative and extended time-series analyzes.