A Stochastic Dual Dynamic Programming model for medium-term hydrothermal scheduling in Greece

This paper describes the development and application of the Stochastic Dual Dynamic Programming (SDDP) algorithm for solving the optimal medium-term hydrothermal scheduling in the Greek power system. The method aims at providing a more elaborate optimization tool compared to the traditional approaches used for medium-term hydrothermal scheduling in Greece; the latter have been developed on deterministic assumptions, while the proposed tool recognizes the impact of inflow uncertainty and provides well-hedged management strategies for each reservoir. The described method is developed using GAMS/CPLEX and tested for a power system similar to the Greek Interconnected Power System. The method proposed can be a stepping stone towards the calculation of the variable cost of each hydroplant on a monthly basis, a goal that has been officialy set forth by the Greek Regulatory Authority for Energy.