The effect of model generated solar radiation data usage in hybrid (wind–PV) sizing studies

Abstract

Wind speed and solar radiation characteristics belonging to past years of a region are the main input parameters in wind–photovoltaic hybrid system (WPHS) sizing studies. Classically, these data are fed to several scenarios with different solar panel, wind turbine, and storage battery number combinations. The solutions with minimal cost which also satisfy the desired maximum loss of energy probability are selected. Since the utilized data have random fluctuations because of atmospheric phenomenon, past years’ data are unlikely to appear in a similar manner in future years. Hence, using a robust model that characterizes the general behavior of the data instead of directly using past data should yield more accurate sizing solutions. In order to compare the sizing accuracy obtained by directly using the data to the accuracy obtained by indirect modeling from data, an analytical solar radiation model is first explained. Using this model, 3-year solar radiation data of three geographical sites are analyzed. It was observed that the differences between sample-by-sample hourly recordings corresponding to different years are significantly larger than the difference between these recordings and the data model obtained from an arbitrary year. This provides a hint that a sizing approach carried out using the data of a previous year would not be accurate in producing the same Loss of Load Probability (LLP) for a future year. On the contrary, the accuracy would improve if a generic analytical model of the solar radiation is used in the sizing process. This foresight is tested by comparing the LLPs obtained in the two ways mentioned above. Results obtained using available data are in accordance with the aforementioned propositions.