Page 82 - 2025年第56卷第7期

（5）：128-138.
［ 17］ LIU X，JONATHAN F. Adjoint-based machine learning for active flow control［J］ Physical Review Fluids，2024，
.
9（1）：013901.
［ 18］ 高昌，李正洲，黄江涛，等 . 基于连续伴随方法的高超声速飞行器高精度气动优化［J］ 航空学报，2021，42
.
（7）：171-182.
［ 19］ 吴航空，王丁喜，黄秀全 . 基于高效高精度离散伴随方法的叶轮机叶片气动优化设计［J］ 工程热物理学报，
.
2024，45（1）：46-57.
［ 20］ 罗佳奇，杨婧 . 基于伴随方法的单级低速压气机气动设计优化［J］ 航空学报，2020，41（5）：139-148.
.
［ 21］ WANG H D，XU D，DING C F，et al. Numerical and experimental study on water-sediment flow in a lateral pump⁃
.
ing station forebay［J］ Physics of Fluids，2024，36 （9）：095162.

［ 22］ ERWAN B，THIEN T，KHOA L，et al. A smoothed particle hydrodynamics （SPH）formulation of a two-phase mix⁃
ture model and its application to turbulent sediment transport［J］ Physics of Fluids，2019，31（10）：103303.
.
［ 23］ 廖俊鹏，田野，杨子荣，等 . 基于边界逆向优化算法的任意分光比耦合器设计［J］ 光学学报，2023，43（1）：
.
148-157.
［ 24］ MADER C A， KENWAY G K W， YILDIRIM A， et al. ADflow： An open-source computational fluid dynamics
solver for aerodynamic and multidisciplinary optimization［J］ Journal of Aerospace Information Systems，2020，17
.
（9）：508-527.
［ 25］ ZHOU L， HUANG J T， GAO Z H， et al. Three-Dimensional aerodynamic/stealth optimization based on adjoint
.
sensitivity analysis for scattering problem［J］ AIAA Journal，2020，58（6）：2702-2715.
［ 26］ SHEN S L，ZHANG X K，ZHENG K Q，et al. Wall-function method used to simplify the solid oxide fuel cell model
.
［J］ Journal of Power Sources，2021，510：230396.
［ 27］ 刘超群 . Liutex-涡定义和第三代涡识别方法［J］ 空气动力学报，2020，38（3）：413-431，478．
.

Exploration of flow pattern optimization in the forebay of a sediment-laden pump
station based on the adjoint method

1，2 1，2 1，2 1，2 1，2

WANG Haidong ，XU Dong ，RAN Qihua ，YUAN Saiyu ，TANG Hongwu

（1. College of Water Conservancy& Hydropower Engineering，Hohai University，Nanjing 210098，China；
2. Key Laboratory of Hydrologic-Cycle and Hydrodynamic-System of Ministry of

Water Resources，Hohai University，Nanjing 210098，China）

Abstract：In the forebay of a forward intake pumping station in a sediment-laden river，the sudden cross-sectional
expansion readily leads to the formation of vortices，which cause significant sediment accumulation，which interfere

with the normal operation of the pumping station. To address this issue，measures such as the installation of diversion

piers are commonly employed to optimize the flow conditions. However，traditional optimization methods face signifi⁃

cant bottlenecks，including numerous combinations of operating conditions，strong subjectivity，and reliance on

empirical approaches，which hinder the achievement of optimal design solutions. This study integrates adjoint optimi⁃

zation algorithms from aerodynamics with the theory of two-phase water-sediment flow. By solving the sensitivity func⁃
tion，both the flow field equation and the adjoint equation are iteratively solved simultaneously，achieving optimiza⁃

tion of the water-sediment flow state in the pumping station forebay. The results indicated that for the traditional
octagonal diversion pier，the adjoint optimization method was applied to fine-tune its geometric shape，effectively

eliminating large-scale vortices. The uniformity of characteristic cross-sectional flow velocity reached 90.23%，while

the vortex structure volume decreased by 85.47% compared to no rectification measures，and by 75.32% compared to
the traditional diversion pier scheme. Additionally，sediment deposition was reduced by 65.22%. The optimization

outcomes were highly significant. The proposed optimization algorithm provides a novel approach for the deep optimi⁃
zation and efficient operation of pumping station forebays.

Keywords：sediment-laden river；adjoint optimization algorithm；two-phase water-sediment flow；sediment deposi⁃

tion；vortex structure

（责任编辑：鲁 婧 韩 昆）
— 908 —