Household photovoltaic systems optimization methodology based on graph theory reliability

Abstract

A novel passive strategy to overcome the inconveniences of possible failure or underperformance of the components of a household scale photovoltaic (PV) system is proposed. The developed methodology is based on network reliability and determines the optimal array topology to be implemented under certain constraints through the maximization of a merit function that considers three different aspects: the total power that a structure produces, the power per panel ratio and the average reliability of the structure. Every factor is weighted through a parameter that controls its relative importance in the merit function. The restriction implemented in this study corresponds to the inherent limits in the voltage and current input of an inverter. The optimization process has been developed for several scenarios considering different decision criteria. Intuitive results are obtained when the optimization relies on the power per unit module and also when reliability is the main contributor in the merit function. When power production is preferred, structures with large number of panels prove to be the best even when they are asymmetrically connected and thus some panels are underperforming. A case study of a particular inverter has been developed to gain insight in an example, computing the optimal structure in terms of the three parameters.