Optimal Hour-Ahead Bidding in the Real-Time Electricity Market with Battery Storage using Approximate Dynamic Programming

摘要

There is growing interest in the use of grid-level storage to smoothvariations in supply that are likely to arise with increased use of wind andsolar energy. Energy arbitrage, the process of buying, storing, and sellingelectricity to exploit variations in electricity spot prices, is becoming animportant way of paying for expensive investments into grid-level storage.Independent system operators such as the NYISO (New York Independent SystemOperator) require that battery storage operators place bids into an hour-aheadmarket (although settlements may occur in increments as small as 5 minutes,which is considered near "real-time"). The operator has to place these bidswithout knowing the energy level in the battery at the beginning of the hour,while simultaneously accounting for the value of leftover energy at the end ofthe hour. The problem is formulated as a dynamic program. We describe andemploy a convergent approximate dynamic programming (ADP) algorithm thatexploits monotonicity of the value function to find a revenue-generatingbidding policy; using optimal benchmarks, we empirically show the computationalbenefits of the algorithm. Furthermore, we propose a distribution-free variantof the ADP algorithm that does not require any knowledge of the distribution ofthe price process (and makes no assumptions regarding a specific real-timeprice model). We demonstrate that a policy trained on historical real-timeprice data from the NYISO using this distribution-free approach is indeedeffective.