Comparison of view factor and ray tracing models

This study presents a comprehensive comparative analysis of View-Factor (VF) and Ray Tracing (RT) models to evaluate their effectiveness in predicting bifacial irradiance in bifacial photovoltaic (PV) systems. The comparison was conducted using an experimental dataset from NREL (CO, USA), which included front and rear plane-of-array (POA) irradiance values measured by silicon solar cell pyranometers. This initial step established a baseline understanding of each model's predictive capabilities in a real-world context. The comparison with POA measurements indicates that for front illumination, both models are closely aligned with relative Root Mean Square Error (rRMSE) of 7.2% (VF) and 7.6% (RT). For rear illumination, the rRMSE increases to 20.9% and 22.9% for the VF and RT models, respectively. Additionally, both models show a similar relative Mean Bias Error (rMBE) for front illumination of −1.8% and − 2.0% for the VF and RT methodologies, respectively. However, for rear illumination, the RT model predicts with only −0.1% bias, while the VF model underestimates POA irradiance by 8.4%.

The study then focuses on directly comparing the VF and RT models in their simulation of cell-level irradiance. Starting from a VF-compatible problem, the research systematically introduces additional features and components in RT simulations to assess the impact of increased complexity on irradiance predictions. In contrast to underestimating of the rear POA sensor values, the VF model overestimates rear module irradiance by 6.7% due to rear module shading by the frame and structure, which do not equally affect the POA sensors. These findings emphasize the importance of detailed geometric modelling in achieving accurate predictions. The overestimation of the module rear illumination translates to a 0.8% overestimation of energy yield by the VF model, considering cell-level irradiance values. Conclusively, the study highlights the significance of accurately incorporating structural elements when modelling bifacial PV systems.