Multistage Day-Ahead Scheduling of Energy and Reserves

This paper presents a new economic dispatch model for the day-ahead scheduling of energy and reserves under uncertainty. The proposed model considers a multistage stochastic framework whereby scheduling decisions are dynamically updated according to observed new information about uncertain parameters such as nodal net injections. As a result, the proposed model provides a one-hour-ahead implementable energy and reserve schedule. Moreover, a particularization of this multistage model is presented, namely a two-stage model, in which a single generation and reserve schedule is provided for each hour of the day. A recently developed regularized linear decision rules framework is used to reduce the computational complexity of the multistage stochastic linear problem at hand and to prevent the in-sample overfitting issue and the threat of poor out-of-sample performance. Numerical simulations based on the IEEE 300-bus system demonstrate the effectiveness of the proposed approach, as well as its economic and operational advantages over the widely used two-stage model.