Bi-Level Optimization Model for Daily Operation with Heterogeneous Hydropower Units in Multiple Reservoirs with Application to the Three Gorges-Gezhouba Cascade Power Stations

摘要

The complexity of the operation of the Three Gorges-Gezhouba cascade hydropower stations lies in not only the enormous unit installed capacity, but also the range of different characteristics from different types of hydropower generating units. The system has two reservoirs and the generation system consists of 47 power generating units. These units are of 10 different types that have different functions describing the relationships among power output, water head and turbine water discharge. A bi-level optimization framework is presented to determine the best hourly allocation of water to each of these types of power generating units to minimize the amount of water used while meeting load demand. The optimization has an outer global optimization loop that determines the amount of water allocation each time period. The inner loop uses linear programming to optimize the allocation of that water between reservoirs and among power generating units. The optimization methodology uses water heads and other seasonal data as input so that hourly allocation to power generating units can be updated daily. Dynamically Dimensioned Search (DDS) and Simulated Annealing (SA) are tested for global optimization in the outer loop, and DDS performs better than SA.