Abstract: By previous calculations, a significant imbalance has occurred for decades long in the upstream and downstream flows of the Three Gorges reservoir and Gezhouba reservoir, severely affecting the practices of flow forecasting, reservoir operation, and power production. After examining the inflow and outflow on different scales, we find that this imbalance is caused by the errors in the previous calculations of power generation flows, and the underlying cause is the poor calculation method of head loss around the Gezhouba dam. This paper develops a novel head loss model for achieving the water balance of the two reservoirs. We adopt the latest unit characteristic curves and the Shuffled Complex Evolution Algorithm in our calibration of the head loss coefficients for each generating unit, and apply the corrected loss coefficients to the calculations of power generation flows, achieving satisfactory water balance results. For the whole year of 2024, or the verification period, our calculations reduce the water volume gap from 11.52 billion m3 to 0.16 billion m3, and correct the previous average head loss of 28.31 cm to a new smaller value of 6.27 cm. Thus, this study significantly reduces water imbalance in cascade reservoirs caused by power generation flow calculation errors.

Key words: hydropower station, unit characteristics, head loss, water balance, evolutionary algorithms