Probabilistic optimal sizing of stand-alone PV systems with modeling of variable solar radiation and load demand

This paper presents a comprehensive sizing methodology which could contain all key elements necessary to obtain a practical sizing result for a stand-alone photovoltaic (PV) system. First, a stochastic solar radiation model based on limited/incomplete local weather data is formulated to synthesis various chronological solar radiation patterns. This enables us to evaluate a long-term system performance and characterize any extreme weather conditions. Second, a stochastic load simulator is developed to simulate realistic load patterns. Third, two reliability indices, Expected-Energy-Not-Supplied (EENS) and Expected-Excessive-Energy-Supplied (EEES), are incorporated with an Annualized Cost of System (ACS) to form a new objective function called an Annualized Reliability and Cost of System (ARCS) for optimization. We then apply a particle swarm optimization (PSO) algorithm to obtain the optimum system configuration for a given acceptable risk level. An actual case study is conducted to demonstrate the feasibility and applicability of the proposed methodology.