Improved Solar Cell Temperature Modeling for Weather to Power Forecasting
“Prediction of solar cell and module operating temperatures are a critical component of solar power forecasting. Elevated operating temperatures reduce overall power conversion efficiencies on the order of .4% per degrees C. This leads to power conversion errors of ~3% from errors of only 5 degrees C in cell temperature prediction. Models for calculating cell temperature from ambient conditions (irradiance, ambient temperature and wind speed) vary from first principles physical models to simple empirical relationships. First principles thermodynamic models are complex, often having prohibitive calculation times for timely forecasting, and require additional empirical fitting for individual installations because of the complexities that the geometry of the installation impart on the model. Simplified empirical relationships, on the other hand, often ignore critical physical phenomenon such as transients based on the thermal mass and wind direction.
We present an improved temperature model and methodology, specific for weather to power forecasting. Test solar field data with module backplane temperature recordings are used to show where current models fail, leading to the inclusion of relevant physical effects. Specifically, we explore the influence of wind direction, thermal mass, seasonal parameter changes, and techniques for handling the discretization of the calculations. The model is empirical, with a set of coefficients specific to each installation, but it is structured with appropriate physical relationships and constants as to aid in the understanding of coefficient influence. We show the evolution of parameters through an entire year of data and use the model to then predict module temperature for the following year. Results are analyzed through overall root mean square error and maximum absolute error and are compared with currently used models.
Support for this work is provided by the DoE SunShot Initiative and IBM.”