Performance of a Small Network of Grid Interactive, Residential Solar Photovoltaic Systems

Abstract

The performance of a network of five recently installed, grid interactive residential solar photovoltaic (PV) systems in Palm Beach County, FL is analyzed, and a probabilistic model for estimating the performance the network is developed. To first order, integrated network performance — whatever the combination of individual PV arrays — can be estimated using generalized tilt factors. These take into account basic geometrical information such as array size/orientation and solar position; as well as atmospheric effects, and module efficiencies. They are computed using a model termed the solar simulator that integrates the instantaneous solar irradiation striking a given PV system over a day. The resulting estimates for mean network performance are within ∼ 6% of the observed values. At present, work on estimators of higher moments of the energy production distribution is incomplete, but local meteorological factors that may influence their values as well as data (Pearson correlations and distribution skewness) useful for future developments are discussed.