Operation and Valuation of Multi-Function Battery Energy Storage under Uncertainty

Electrical energy storage resources (ESRs) oer a promising solution to many of the issues facing the electric grid. In order for this promise to be fully realized,
new intelligent decision-making technologies are required. This dissertation studies the operation and valuation of ESRs in an uncertain electric grid environment.
ESRs can include both stationary battery energy storage systems (BESSs) and distributed deferrable loads such as plug-in electric vehicles (EVs). An ESR can be operated to provide multiple services simultaneously, maximizing
its value. An EV can provide transportation services as well as participate in electric grid frequency regulation. A BESS can also provide frequency regulation while providing load peak shifting. In this thesis, we propose new and
innovative solutions that enable optimal operation and accurate valuation of multi-function ESRs under uncertainty. New Markov decision problems (MDPs) for smart charging of EVs are developed for cases of price, ancillary services, and
driver behavior uncertainty. In order to compare the proposed MDP approaches with deterministic optimization approaches, a Dynamic Monitoring and Decision
Systems (DYMONDS) energy market simulation is developed. We also propose an innite horizon MDP approach to estimating the net present value of a BESS
that degrades over time. In order to optimize the economic scheduling of an ESR that provides frequency regulation service, one needs a predictive model of the
automatic generation control (AGC) signal. We investigate timeseries and other statistical models for the prediction of an AGC signal and its cumulative effect on the state of charge of an ESR.