Title:Optimal control of end-user energy storage

Abstract:An increasing number of retail energy markets exhibit price fluctuations and provide customers such as data centers and residential users the opportunity to buy energy at lower than average prices. Such cost savings however are hard to obtain in practice because they require users to observe the price fluctuations and shift their consumption to low price periods. We propose to use energy storage to allow users to satisfy demand at times when prices are high with energy previously bought during low-price periods. We investigate how to control the battery to minimize energy costs, subject to fluctuating prices and user demand. We formulate this problem as a Markov Decision Process and show that the optimal policy has a threshold structure. We then numerically compute the thresholds using real-world traces from existing energy markets. We show that energy storage using this policy leads to significant cost savings, and study the impact of the battery size on the optimal policy and the potential savings.