A New Methodology to Generate Long Time Series of Solar Radiation Based on Stochastic Analysis

Abstract

A detailed knowledge of the solar resource is essential for the planning and operation of solar energy systems. For characterizing the solar resource in a specific location over a longer period of time, such as 30 years, typical meteorological year (TMY) data sets are commonly used. The TMY comprises hourly values of solar radiation and meteorological elements for a 1-year period, chosen to represent ‘typical’ conditions in ten or more years of data. Consequently, TMY are not suited for designing systems to meet both the variability and the extreme conditions occurring at a location. In this work, probability distribution functions that contain the natural variability of the solar resource are analyzed. These distributions facilitate the generation of large data sets of monthly series of solar radiation; each of these yearly series can be seen as a feasible year and, therefore, possible years are used instead a median year that doesn’t contain information about variability or extreme conditions.