An Enhanced Network-Constrained UC Model for Leveraging System Operation Cost and Financial Profitability of Incentive-Based DR Loads

Demand response (DR) programs provide consumers opportunities to play an active role in electricity market operation by encouraging them to reduce and/or shift their electricity usage during peak periods and/or when system security is jeopardized. There are primarily two types of DR programs, including incentive-based DR and price-based DR. This paper focuses on the incentive-based DR. For this type of DR program, although shifting peak loads to off-peak periods could reduce system operation cost, it may raise off-peak electricity prices and the incentives may not be enough to cover the increased payments of DR loads. This would diminish consumers' enthusiasm to the continued participation in DR programs. Federal Energy Regulatory Commission Order 745 states that when dispatching a DR resource as an alternative to a generation resource is cost-effective, which is determined by the Independent System Operator's (ISO) net benefits test, the DR must be compensated at full locational marginal price (LMP) value. However, the order focuses more on load curtailment, but not load shifting. In addition, the net benefits test mainly focuses on consumers' savings, while ignoring operation costs and the value lost by consumers, which may cause system inefficiency. This paper proposes a systematic model for leveraging system operation cost and consumer payment concerns in electricity market operation. The proposed model minimizes the system operation cost, while incorporating explicit LMP formulations and consumer payment requirements into the network-constrained unit commitment formulation. Thus, it determines the proper amount of DR loads to be shifted from peak hours to off-peaks under the ISO direct load control for reducing the operation cost and ensuring that DR load payments will not deteriorate significantly after load shifting. The proposed model is solved in its original mixed-integer nonlinear programming formulation and the mixed-integer linear programming reformulation via piece-wise linear approximation. Numerical case studies illustrate the effectiveness of the proposed model and compare the computational performance of the two formulations.