A stochastic policy algorithm for seasonal hydropower planning

Abstract Hydropower producers need to plan several months or years ahead estimate the opportunity value of water stored in their reservoirs. The resulting large-scale optimization problem is computationally intensive, and model simplifications are often needed allow for efficient solving. Alternatively, one can look near-optimal policies using heuristics that tackle non-convexities production function a wide range modelling approaches price- inflow dynamics. We undertake an extensive numerical comparison between state-of-the-art algorithm stochastic dual dynamic programming (SDDP) rolling forecast-based algorithms, including novel we develop this paper. name it Scenario-based Two-stage ReOptimization abbreviated as STRO. experiments based on convex programs with discretized exogenous state space, which makes SDDP applicable comparisons. demonstrate our handle risk better than traditional by reducing optimality gap from 2.5 1.3% compared bound.