Prediction of horizontal diffuse solar radiation using clearness index based empirical models; A case study

Diffuse solar radiation is an important factor for design and simulation processes of the photovoltaic (PV) panels as well as other solar systems. Nowadays, hydrogen generation can be accomplished by harnessing solar energy via PV systems to electrolyze the water in a clean and environmentally friendly manner. In this study, 13 existing functional forms of the clearness index based diffuse radiation models, reported so far in the literature, have been firstly categorized on the basis of how they correlate: (1) the diffuse fraction and (2) diffuse coefficient with clearness index as the sole input variable. Then in a case study, by utilizing the long term global and diffuse solar data, in overall 26 new models have been established for both daily and monthly mean daily predictions for the city of Kerman, Iran. The performances of the new established models have been evaluated using different well-known statistical indicators. For the daily based prediction, the diffuse fraction model with linear form is the most suitable one with root mean square error (RMSE) and correlation coefficient (R) of 1.3081 MJ/m2 and 0.8767, while for the monthly mean daily prediction, the diffuse coefficient model again with linear form provides the utmost accuracy, for which RMSE and R are 0.5391 MJ/m2 and 0.9258, respectively. The employment of these only clearness index based models for any solar system application, in particular solar hydrogen production, is convenient; given that the knowledge of clearness index only requires the global solar radiation data.