Validation of Novel Bifacial Photovoltaic Performance Model with 3D shading for Fixed-Tilt and Single-Axis Tracked Systems

Abstract

Existing bifacial photovoltaic (PV) performance models fall primarily into two categories: (1) ray tracing models that capture complex shading but lack the computational efficiency required for optimization applications; and (2) view factor (VF) models that efficiently simulate energy transfer but rely on user-defined losses which neglect temporal variation in phenomena such as shading and electrical mismatch. Hybrid VF / ray tracing models selectively employ ray tracing while balancing computational efficiency. This paper describes the validation of a novel hybrid model, DUET, which combines a 3D VF model with deterministic ray-object intersections. The software provides 2D irradiance profiles and mismatch-inclusive current-voltage curves for each scale of components: from cells to the full array. Validation against open-access data from Denmark shows that DUET predicts bifacial energy yield at 0.76% and 0.65% lower than measured yield for fixed-tilt and horizontal single-axis tracked (HSAT) rows, respectively, over 3370 and 2731 daylight hours. Monthly relative error in bifacial energy ranges from < 1% to ~4.5% for both systems. The mean absolute error (MAE) in hourly bifacial power is 18.1 mW/Wp for fixed-tilt and 18.4 mW/Wp for HSAT. These errors fall below the lowest previously reported MAE for six software at the same field site by ~0.77 mW/Wp for fixed-tilt and ~1.1 mW/Wp for HSAT. Modelled average rear insolation agrees with pyranometer data within +4.4% for fixed-tilt and -0.76% for HSAT. For both configurations, rear irradiance MAE aligns with the lowest error previously reported for other software at the site.