Event-Based Dynamic Programming for Pumped-Storage Hydropower Scheduling
A novel DP framework using event networks beats traditional time-indexed methods...
A team of researchers has introduced an event-based dynamic programming (DP) approach to tackle the complex scheduling of pumped-storage hydropower (PSH) plants. The method, detailed in a new arXiv paper, fundamentally rethinks the classic time-indexed mixed-integer formulation by representing operating schedules as sequences of mode-specific events. Within each event, dispatch decisions—such as generation levels and pumping rates—are determined by linear programs, making the overall problem a deterministic DP on an event network.
This modular framework allows operators to easily add new operating modes (e.g., hydraulic short-circuit operation) without overhauling the network structure. To achieve tractability, the authors develop a finite-grid approximation that leads to a linear programming formulation, and propose an event-based branch-and-bound algorithm for continuous-state problems. Numerical results confirm that the event-based approach provides a computationally efficient alternative to conventional methods while offering substantial modeling flexibility for real-world PSH scheduling.
- Event-based DP replaces time-indexed models with mode-specific events, each solved via linear programs.
- Finite-grid approximation yields a tractable LP formulation; branch-and-bound handles continuous states.
- Modular design allows adding new operating modes (e.g., hydraulic short-circuit) without major changes.
Why It Matters
More efficient hydropower scheduling means lower energy costs and better grid stability for renewable integration.