Multistage stochastic hydrothermal scheduling

Abstract

The paper addresses the stochastic long-term hydrothermal scheduling problem. This problem is usually modeled as a multistage stochastic program and solved using algorithms that are based on dynamic programming. One such algorithm is the stochastic dual dynamic programming (SDDP) algorithm, which is implemented for the hydrothermal power system of the Pacific Northwest in the U.S. The importance of the first period is emphasized. It is shown that the impact of the first period diminishes after four periods. This result can be exploited to make the stochastic dual dynamic programming (SDDP) algorithm converge very fast.