ACCOUNTING FOR VARIATIONS IN THE SPECTRAL IRRADIANCE DISTRIBUTION IN PV PANEL PERFORMANCE MODEL

Abstract

Photovoltaic (PV) panel performance is strongly influenced by the available incident irradiance and the module temperature. Since these conditions vary on different timescales, the actual outdoor energy yield of PV modules is hard to anticipate. Diode models exist, correcting for total irradiance and temperature variations, which are able to accurately predict the outdoor performance of crystallinesilicon (c-Si) modules. The performance of amorphous-silicon (a-Si) photovoltaic modules, however, is mainly affected by the spectral distribution of the incident light, which is not included in most PV performance models. With the increasing market share of amorphous silicon, the need to include spectral information is growing. Accommodating variations in the spectral irradiance to improve PV performance models on clear days in the Netherlands is investigated in this report. For this, three spectral characterization methods are examined: the total irradiance, the useful irradiance and the average photon energy of the spectrum. Daily variations in the spectral irradiance are revealed by the useful irradiance and average photon energy; whereas the spectral effects on PV performance stress that the total irradiance is not sufficient for characterization of the incident irradiance for PV performance modeling. Additionally, it is shown that the spectral distribution has only a small effect on crystalline-silicon panel performance modeling. Of the three spectral parameterization methods investigated, the useful irradiance shows best results in accounting for the spectral irradiance on clear days and the modeling accuracy for amorphous-silicon modules can be improved significantly. However, low modeling accuracy is obtained at low incident irradiance, which is caused by a blue shift of the spectrum, not captured by the useful irradiance. The average photon reveals these changes; nevertheless on clear days, a simple relation between module performance and APE to correct for these spectral changes was not found. Improving PV performance modeling under low irradiance remains a challenge and requires more comprehensive characterizations of the spectral irradiance than the average photon energy and useful irradiance.