Owing to the different degrees of bending in rivers in nature, it is difficult to conduct experiments in situ. In this study, the renormalization group (RNG) k-ε turbulence model in ANSYS Fluent was used to analyze the three-dimensional flow structure and turbulence characteristics near a spur dike and to evaluate the variation trend of flow in rivers with different degrees of bending. The results show that in channels with different curvatures, the vortex appears between the spur dikes and is disturbed by the permeable hole, and the backflow area moves downstream. The strength of secondary flow (SSF) fluctuates greatly in the vicinity of the spur dike and the downstream region, and the peak value appears 3.22 m (21.5 times L) away from the inlet of the bend. The SSF increases as the bend curvature increases. The SSF displays similar variation trends in the three kinds of bends. The peak value of normalized turbulent kinetic energy (NTKE) appears 3.14 m away from the entrance of the bend, the NTKE is the largest in the 45° bend and the smallest in the 180° bend, and it decreases only at distances of 3.25–4.19 m away from the entrance of the bend as the bend curvature increases.

Unfavorable flow patterns, such as vortices and backflow in pump station forebays, can deteriorate pump suction conditions and exacerbate sediment deposition. This study investigates the effects of different pump unit operation combinations on the flow patterns in the forward-intake forebay of the Hongsi Pumping Station. A three-dimensional numerical model was established using the RNG k-ε turbulence model combined with the VOF method for free surface tracking, with Enhanced Wall Treatment employed for the near-wall region. The model was validated against field measurements, with mean absolute errors of 0.058 and 0.085 for two characteristic lines, respectively. Four unit operation combination schemes were simulated and compared. Results show that the velocity in the front end of the main flow zone reaches 0.344–0.345 m/s across all schemes, which is 19–24% higher than in the middle section and 26–37% higher than in the rear section. The inter-scheme velocity variation is minimal near the forebay inlet (1.1% at section X1) but increases significantly near the pump inlets (approximately 11% at section X4), indicating that unit operation combinations primarily affect the rear forebay region. Large-scale vortices are observed near the side walls and pump inlets, with vortex cores located at different positions depending on the operation scheme. When the units at both ends are operated symmetrically (Scheme 1: units 1#, 2#, 5#, 6#), the flow pattern is the most favorable, displaying characteristic dual-core vorticity distribution with symmetric pattern and moderate intensity, with the velocity uniformity at the inlet section reaching 81.77%, which is 5.21, 11.35, and 1.91 percentage points higher than Schemes 2, 3, and 4, respectively. For practical operation, Scheme 4 (1#, 2#, 4#, 5#) can serve as an alternative with a velocity uniformity of 79.86%, while central concentration and single-side bias configurations should be avoided due to their larger vortex scales and lower velocity uniformity. The findings provide quantitative guidance for the operation management of forward-intake pump stations.

To address the problems of disordered flow patterns, wall-separated recirculation, and sediment deposition that commonly occur in the forebay of forward-facing pumping stations on sediment-laden rivers, the first-stage Xin Zhuangji Pumping Station in Ningxia was selected as the study case. A three-dimensional numerical simulation was conducted using the Realizable k–ε turbulence model coupled with the Mixture two-phase flow model for water–sediment flow. The regulation effects of splayed guide walls with different guide-wall deflection angles on the velocity distribution, vortex structures, and sediment deposition in the forebay were investigated. The results show that large-scale recirculation zones exist on both sides of the prototype forebay, accompanied by uneven velocity distribution and severe sediment deposition within the recirculation regions. The installation of splayed guide walls can effectively suppress lateral recirculation, expand the mainstream flow region, and reduce the deposition area. However, the regulation effect of the diffusion-type guide wall varied significantly with the guide-wall deflection angle. Since the main objective of this study was to control sediment deposition rather than to maximize a single hydraulic indicator, a multi-criteria screening method oriented toward sediment-reduction control was adopted. Under the (θ = 25°) scheme, the reduction ratio of the overall potential deposition area based on the primary threshold criterion reached the maximum value of 53.67%, the recirculation area on plane Z1 decreased by 23.23%, the global recirculation coefficient increased to 68.69%, and the sediment deposition efficiency decreased to 0.033. The axial velocity uniformities at the suction-pipe sections of pumps B# and C# were 71.23% and 80.61%, respectively. Although the (θ = 30°) scheme produced the highest velocity uniformity for pump B# and showed a slightly better reduction effect on local high-concentration sediment-enrichment regions, the (θ = 25°) scheme exhibited a more balanced improvement in overall sediment-deposition control and flow-pattern regulation. Therefore, under the investigated operating condition, the (θ = 25°) scheme is recommended as the guide-wall deflection angle oriented toward sediment-reduction control.

Velocity distributions in curved open-channel flows exhibit strong three-dimensionality and nonlinear behavior, posing challenges to both accurate prediction and physical interpretation. Using measured velocity data from nine discharge–water-depth combinations in a laboratory 180° open-channel bend, this study developed an integrated eXtreme Gradient Boosting (XGBoost)–SHapley Additive exPlanations (SHAP) framework, with multiple linear regression (MLR), random forest (RF), and a back-propagation neural network (BPNN) used for comparison. Leave-one-condition-out cross-validation was used to evaluate the predictive accuracy and stability of the four models. A stratified sampling strategy was then adopted to construct the training dataset, allowing information from all flow regimes to contribute to robust parameter calibration; the two data-partitioning strategies yielded broadly comparable predictive performance. Using models trained with stratified sampling, multidimensional model evaluation was further conducted using global statistical metrics, segment-wise predictive performance, held-out extreme-condition tests, and measured–predicted agreement, among other criteria, with XGBoost consistently showing the best performance. Multilevel SHAP analyses quantified global feature importance, pairwise interactions, streamwise variations in feature contributions, SHAP–PDP dependence relationships, and condition-specific attribution. The SHAP results indicate a two-level attribution structure in the model: hydraulic variables jointly define the global velocity baseline, and their contribution signs can switch between positive and negative. Spatial variables characterize cross-sectional velocity redistribution. Strong discharge–depth interaction is associated with width-to-depth-ratio-dependent adjustment of the bend flow field. The proposed framework establishes a complete experiment-driven prediction–mechanism interpretation workflow for sharply curved open-channel flow and provides new quantitative insight into model-represented multifactor hydrodynamic interactions in open-channel bends.

To mitigate the severe recirculation and sediment deposition induced by highly sediment-laden flow in the lateral-inflow forebay of the No. 3 Pumping Station in Hong sipu, Ningxia, within the Yellow River Basin, this study investigates the effects of different corner configurations of the left and right sidewalls on the hydraulic characteristics of the forebay flow. A three-dimensional numerical model was established using ANSYS Fluent, with the Realizable κ−ε turbulence model and the Mixture multiphase model for water–sediment two-phase flow. The numerical model was validated against field measurements, and the variations in the flow-field structure and flow velocity in the lateral-inflow forebay with different sidewall contraction angles were obtained. The results indicate that after the flow enters the forebay from the channel, it rapidly changes direction under the influence of suction from the pump units, resulting in clearly differentiated flow regions within the lateral-inflow forebay. A high-velocity flow region forms in the central part of the forebay, while two irregular vortices of different sizes develop on both sides. As the angle α at the left sidewall corner gradually decreases, both the area of the recirculation zone and the vortex scale are significantly reduced. When the sidewall contraction angle α is 0.12 °, the reverse-flow area in the forebay decreases by 36.04%, while the sediment accumulation area decreases by 17.81%, effectively alleviating sedimentation in the forebay of water-diversion pumping stations located on sediment-laden rivers. Furthermore, under the α = 0.12° condition, which exhibited the best hydrodynamic performance among the investigated cases, variations in the angle β of the right sidewall within the range of 50–84° produce no significant changes. The findings can provide guidance and reference for the design, renovation, and upgrading of similar pumping stations.