Abstract
Electric utilities have been investigating methods to reduce peak power demand. Demand response (DR) is one such method which intends to reduce peak electricity demand. DR programs typically have limits on the number and timing of events that may be triggered for a selected group of customers. This paper presents a methodology for optimizing the scheduling of DR events for various DR programs. The proposed optimization mechanism establishes a policy that triggers DR events according to the criteria that govern the cost to the utility and based on probability distributions of exogenous information that is accessible to utilities a priori, for decision making. The policy determines a dynamic threshold for triggering events that optimizes the expected savings over the planning horizon. Case studies using real utility data show that our solutions are better than current industrial practices, and close to ex-post optimality.