Photovoltaic module model accuracy at varying light levels and its effect on predicted annual energy output

Models to predict photovoltaic (PV) module power output are constantly improving, yet many are tuned to maximize accuracy at high irradiance, which has the potential to introduce errors when predicting energy output with other operating conditions. Correct power estimation under lower light conditions is very important, particularly in cloudy climates or when an array is non-optimally oriented, as is often the case with building-mounted or building-integrated PV systems. Newer CEC and IEC standards require manufacturers to provide data on module performance under a variety of operating conditions, and the purpose of this study is to explore the potential of these new data to improve model accuracy for different PV technologies. This study examines the accuracy of the commonly used, CEC-Wisconsin single diode five parameter PV model, comparing its predicted power output to measured and empirical model datasets over a range of incident irradiance levels and temperatures. The same comparisons are also performed for a modified, seven parameter version of this model, which includes the use of low irradiance data to assign values to the extra parameters. Results of this work further understanding of model misprediction at low irradiance and its impact on estimation of annual energy production. At medium-low irradiance (<;500 W/m²), the original model is found to overpredict power output by as much as 15% for crystalline modules when compared with the datasets, and even more for thin film modules. The modified model brings the power overprediction down to ~5% or less for most cases. Annual simulations are performed comparing the predicted energy output of the original and modified models to that of the datasets. The modified model shows a significant improvement in accuracy of annual energy prediction, especially for cloudy climates and situations where the array has a non-optimal tilt or orientation.