Mechanism design for reliability in demand response with uncertainty

In this paper, we consider a two-stage demand response (DR) program where a DR aggregator calls upon customers to reduce demands at stage II in order to clear a targeted amount of electricity supply deficit. At stage I, customers only possess distributions of their load shedding costs instead of accurate values because of the intrinsic uncertainty of load shedding. We design an outcome-contingent mechanism in which customers report their private types of the cost distributions and the DR aggregator selects customers based on the reported information. The mechanism guarantees incentive compatibility and individual rationality. In addition, the mechanism ensures high reliability of the DR. That is, the deviation between the total reduced loads and the targeted amount is small. We provide both theoretical analysis and numerical studies to demonstrate the high reliability.