Using TMY3 data to stochastically predict the performance of a PV/VRB microgrid

Abstract

Accurately characterizing the performance of an off-grid photovoltaic (PV) microgrid system can help ensure that the system is appropriately sized to reduce the reliance on supplemental power generation via diesel-fueled generators. However, deterministic models cannot account for the inherent variability of solar insolation, ambient temperature, initial battery charge, and electrical load. A Monte Carlo model was developed by identifying those four variables as random variables. Typical Meteorological Year 3 (TMY3) data were used to develop the Probability Density Functions (PDFs) for the environmental variables, and the initial charge PDF was developed using engineering judgment while the load PDF was based on observed data. Comparison of the stochastic model results against limited performance data from two PV-based microgrid systems with vanadium redox batteries in Missouri indicated that the stochastic technique has the potential for widespread applicability. This potential is due in part because TMY3 datasets are available throughout the United States, and the basic model may be modified to include energy storage systems other than the subject vanadium redox battery.