Scalable model predictive control of demand for ancillary services

In this paper, we develop an integrated decision making framework for the planning and real-time control decisions made by a Load Serving Entity (LSE) providing ancillary services to the wholesale market. Due to the multi-settlement structure of the energy market, planning decisions by the LSE are naturally made at multiple temporal stages. The tight interdependence among decisions demands an integrated approach to minimize the overall costs of operation. In order to model the dynamics of the load at large-scales when making these decisions, we propose a classification-based model that captures the effect of scheduling decisions made for individual appliances at aggregate levels, with reasonable effort. To provide a tangible example of how this load aggregation technique can be applied, we study the case of Electric Vehicle (EV) charging in detail.