Improving Predictability of Renewable Generation Through Optimal Battery Sizing.

The number of large-scale photovoltaic (PV) and wind farms is rapidly growing in Australia and all around the world. When these resources participate in the wholesale electricity market, their uncertain nature of generation results in revenue loss due to the penalty incurred by deviating from day-ahead and real-time commitments. In an attempt to avoid financial losses, they typically bid in the market conservatively. This, in turn, might lead to wasting clean energy and lowering overall profit for the producers. To address these issues, various energy storage devices are considered as a potential solution by academic and industrial researchers alike. In this study, an optimal battery sizing methodology is proposed to improve renewable generation predictability using “Seasonal-Trend decomposition based on LOESS” 1 locally weighted regression. decomposition technique, self-similarity estimation, and enhancing it through filtering. The ultimate goal is to determine the optimal battery size that enhances predictability of renewable generation regardless of the prediction technique and time horizon, which necessarily improves the accuracy of predicted values. The goal is achieved by the proposed method through designing a forecasting-technique-agnostic algorithm. For optimal battery sizing, an optimization formulation is proposed including battery degradation through its useful lifetime. Moreover, prediction studies are carried out to prove predictability enhancement using four prediction techniques and three prediction horizons. The simulation results show the effectiveness of the proposed method in improving self-similarity index (i.e., Hurst exponent) in the PV production time series and economic viability of the proposed methodology in a particular application. [ABSTRACT FROM AUTHOR]