Optimal Offer Curve Design of a Virtual Power Plant Considering Multi-Stage Uncertainties

Abstract

Virtual power plants (VPPs) participate in electricity markets by aggregating distributed energy resources (DERs). Some resources have time-coupling characteristics, such as the energy limit of storage devices, and can be internally controlled without affecting predetermined offering profiles  Hence, VPP must design its economic bids by holistically incorporating future scenarios and flexible control of DERs to maximize profit. This paper proposes a multistage stochastic optimization model to design VPPs' multi-segment power offerings curve considering day-ahead market and real-time market price evolution, as well as renewable generation uncertainties. Our proposed model includes a virtual stage to determine the profit-maximization multi-segment offering curves and operation stages to minimize the dispatch cost of DERs considering uncertainties realizations for delivering market-clearing results. We propose a two-level algorithm combining stochastic dual dynamic programming (SDDP) and zeroth-order method (ZOM). At the lower level, SDDP estimates the real-time operation costs based on given power offerings, which are iteratively updated by ZOM from the upper level. Simulations conducted using a 15-bus power distribution network and historical electricity prices demonstrate that the multi-segment offering and DER dispatch plans from our MSO model yield a 5.6% higher profit compared to the single-segment power offering, and a 27.7% higher profit compared to the offering from the single stage model.

Keywords: Electricity markets, stochastic optimization, virtual power plant, offering strategy