The Value of Real-Time Energy Arbitrage with Energy Storage Systems

Abstract

The growing penetration of renewable generation has increased the volatility of energy prices, especially in the real-time market. Energy storage owners collect revenues from this price variation by performing energy arbitrage. This paper develops a framework to determine the value of energy arbitrage in the real-time and day-ahead markets. A statistical analysis on the historical energy market data of Pennsylvania-New Jersey-Maryland (PJM) reveals higher variations in the real-time prices than the day-ahead making the former more favorable for energy arbitrage. First, the maximum potential revenue from each market is calculated using a linear optimization program assuming a perfect foresight of future prices. In this simulation, the real-time market revenues are twice as high as the day-ahead market. A more realistic assumption is also modeled to incorporate the uncertainties and forecast errors in both markets. Two techniques are proposed to handle the uncertainties: back-casting and normal errors. A dynamic shrinking horizon optimization algorithm is developed to better capture the maximum revenue of energy arbitrage in the real-time market based on the actual participation model of energy storage. Even under large forecast errors in the real-time market, higher arbitrage revenues are obtained using the proposed strategy compared to the day-ahead market with the perfect foresight. The higher arbitrage value of the real-time market is guaranteed under a wide range of forecast errors and storage parameters.