Multistage Stochastic Power Generation Scheduling Co-Optimizing Energy and Ancillary Services

摘要

With the increasing penetration of intermittent renewable energy and fluctuating electricity loads, power system operators are facing significant challenges in maintaining system load balance and reliability. In addition to traditional energy markets that are designed to balance power generation and load, ancillary service markets have been recently introduced to help manage the considerable uncertainty by reserving certain generation capacities against unexpected events. In this paper, we develop a multistage stochastic optimization model for system operators to efficiently schedule power-generation assets to co-optimize power generation and regulation reserve service (a critical ancillary service product) under uncertainty. In addition, to improve the computational efficiency of the proposed multistage stochastic integer program, we explore its polyhedral structure by investigating physical characteristics of individual generators, the system-wide requirements that couple all of the generators, and the scenario tree structure for our proposed multistage model. We start with the single-generator polytope and provide convex hull descriptions for the two-period case under different parameter settings. We then provide several families of multiperiod strong valid inequalities linking different scenarios and covering decision variables that represent both power generation and regulation reserve amounts. We further extend our study by exploring the multigenerator polytope and derive strong valid inequalities linking different generators and covering multiple periods. To enhance computational performance, polynomial-time separation algorithms are developed for the exponential number of inequalities. Finally, we verify the effectiveness of our proposed strong valid inequalities by applying them as user cuts under the branch-and-cut scheme to solve multistage stochastic network-constrained power generation scheduling problems.