Strong formulations for multistage stochastic self-scheduling unit commitment

Kai Pan, Yongpei Guan

Research output: Journal article publicationJournal articleAcademic researchpeer-review

4 Citations (Scopus)

Abstract

With the increasing penetration of renewable energy into the power grid system, the volatility of real-time electricity prices increases significantly. This brings challenges for independent power producers to provide optimal bidding strategies. The traditional approaches of only attending the day-ahead market might not be profitable enough without taking advantage of real-time price volatility. In this paper, we study the optimal bidding strategies for the independent power producers utilizing self-scheduling strategies to participate in the real-time market considering real-time electricity price volatility, with the objective of maximizing the total expected profit. Considering the correlations of renewable energy generation outputs among different time periods, the correlations of real-time prices are captured in our modeling framework, in which we explore a multistage stochastic scenario tree to formulate the price uncertainties. Accordingly, the derived multistage stochastic self-scheduling unit commitment problem is transformed as a deterministic equivalent mixed-integer linear programming formulation. To overcome the curse of dimensionality, we develop strong valid inequalities for the derived stochastic unit commitment polytope to speed up the algorithms to solve the problem. In particular, we derive strong valid inequalities that can provide the convex hull descriptions for the two-period case and a special class of the three-period cases with rigorous proofs provided. Furthermore, strong valid inequalities, including facet-defining proofs, for multistage cases are proposed to further strengthen the model. Finally, numerical experiments verify the effectiveness of our derived strong valid inequalities by incorporating them in a branch-and-cut framework.
Original languageEnglish
Pages (from-to)1482-1498
Number of pages17
JournalOperations Research
Volume64
Issue number6
DOIs
Publication statusPublished - 1 Nov 2016

Keywords

  • Cutting planes
  • Self-scheduling unit commitment
  • Stochastic integer programming

ASJC Scopus subject areas

  • Computer Science Applications
  • Management Science and Operations Research

Cite this