Improved Model for Forecasting Global Solar Irradiance During Sunny and Cloudy Days

Abstract A precise estimation of the electrical energy produced daily by photovoltaic (PV) systems is important both for PV owners and for electrical grid operators. It can be achieved if the received solar irradiance can be accurately estimated during any type of daily solar profile (clear, cloudy, mixed sky), not only average solar profile for larger periods of time, e.g. one month or season, as used in PV system design. The paper firstly describes an existing mathematical model, based on the Meliss approach, which uses mean monthly coefficients for estimating average direct and diffuse solar irradiance. This model is satisfactory for monthly/annual intervals but is not useful for daily estimations. Therefore in the second part of the paper an algorithm which allows to generate daily variations of the model’s coefficients for clear and cloudy sky conditions is proposed. The improved model with variable coefficients was tested during several representative days and can be used for estimating the effect which different meteorological conditions as fog/dew/frost have on the quantity and quality of the solar irradiance received by a PV convertor.