A Benchmark and Validation of Bifacial PV Irradiance Models
Amir Asgharzadeh1,3, Marc Abou Anoma1, Adam Hoffman1, Chetan Chaudhari1, Sushrut Bapat1, Richard Perkins1, Dan Cohen1, Daniel Riley2, Fatima Toor3, Ben Bourne1
1SunPower Corporation, San Jose, CA, United States
/2Sandia National Laboratories, Albuquerque, NM, United States
/3Electrical and Computer Engineering Department, The University of Iowa, Iowa City, IA, United States

In this study, four bifacial irradiance models with two different methods: 3D ray-tracing (bifacial_radiance), and 2D view-factor (bifacialvf, pvfactors, and PVsyst) are used to model a fixed 35° tilted south-facing PV system installed at Sandia National Laboratories, Albuquerque, NM and a single-axis tracker PV system installed at the SunPower R&D Ranch, Davis, CA. The modeled front and backside irradiances are compared to the measured irradiance from the reference cells installed at the sites. Modeling results show that the RMS error of bifacial_radiance in modeling backside irradiance is lower than the view-factor models. However, the higher accuracy of the ray-tracing model comes at the cost of higher computational time needed to run the simulations. The modeling results also suggest that the racking system’s impact on the backside irradiance reduction is significant for the fixed-tilt system but is not impactful for the single-axis tracker system. It is also shown that bifacialvf and PVsyst have higher accuracy in modeling the frontside irradiance probably due to the angle of incidence (AOI) correction methods used in these models. In general, this study demonstrates that open source and commercially-available bifacial models are sufficiently accurate to rapidly run energy models for fixed-tilt and tracking bifacial PV systems.

Area: Sub-Area 8.3: Models for Energy Prediction