STOCHASTIC DYNAMIC PROGRAMMING (SDP) AND SAMPLE STOCHASTIC DYNAMIC PROGRAMMING (SSDP) FOR OPTIMIZATION OF KOREAN HYDROPOWER PLANT

Abstract

As society is increasingly aware of the ecological value of water. As a result, sustainable eco-friendly hydropower reservoir operation is a priority to preserve downstream biodiversity while minimizing the impact on energy production levels. This study develops Stochastic Dynamic Programming (SDP) and Sample Stochastic Dynamic Programming (SSDP) optimization models to address minimum environmental flow constraints on hydropower operations levels and storage targets while reflecting the uncertainty in future inflow forecasts. A case study of the Bosunggang Hydropower system in Korea compares the performance of historical operations with decisions generated by SDP and SSDP models with different hydrologic state variables, state variable discretization, and system turbine capacities. A watershed model, SSARR, was successfully employed to obtain a daily soil moisture series representing the watershed’s wetness. Importantly, simply adopting sophisticated optimization models without careful consideration of system characteristics such as basin hydrology and system objective does not guarantee better optimized system performance.