Evaluation of Irradiance Transposition Models When Utilized with Single Axis Tracking PV Systems in the Southwestern United States

This work examines the accuracy of commonly-used irradiance decomposition and transposition models when modeling single-axis tracking photovoltaic (PV) arrays in the southwestern United States. Inputs to the models include measured one-minute global horizontal, diffuse horizontal, and direct normal irradiance. The bias and variability associated with decomposition models' predictions are analyzed and compared according to sky clearness and solar zenith angle, with results varying widely depending on the model employed. Separately, transposition model predictions are compared to measured plane-of-array (POA) irradiance at an operating PV plant with single-axis tracking. Good agreement is found between the transposition models' predictions and measured irradiance on clear sky days, with an absolute mean bias difference of <;2% for all anisotropic models. The decomposition and transposition models are then combined, with their predictions compared to measured POA irradiance at a second tracking PV plant on an annual and seasonal basis. More variability and bias are seen when both decomposition and transposition are performed, though good agreement with measured data is still possible with some model combinations.