Targeted Demand Response: Formulation, LMP Implications, and Fast Algorithms

Demand response (DR) is regarded as a solution to the issue of high electricity prices in the wholesale market, as the flexibility of the demand can be harnessed to lower the demand level for price reductions. As an across-the-board DR in a system is impractical due to the enrollment budget for instance, it is necessary to select a small group of nodes for DR implementing. Current studies resort to intuitive yet naive approaches for DR targeting, as price is implicitly associated with demand, though optimality cannot be ensured. In this paper, we derive such a relationship in the security-constrained economic dispatch via the multi-parametric programming theory, based on which the DR targeting problem is rigorously formulated as a mixed-integer quadratic programming problem aiming at reducing the averaged price to a reference level by efficiently reducing targeted nodes' demand. A solution strategy is proposed to accelerate the computation. Numerical studies demonstrate compared with the benchmarking strategy, the proposed approach can reduce the price to the reference point with less efforts in demand reduction. Besides, we empirically show that the proposed approach is immune to inaccurate system parameters, and can be generalized to variants of DR targeting tasks.