Due to machining precision, spraying coating, cavitation, and erosion, walls of pump flow passage components have different roughness, which will affect the flow state near the wall and even the main flow. In the present work, four types of roughness Ra = 0, 0.1, 1, and 10 μm are set to investigate the influence of wall roughness on the interaction of leakage flow and main flow in a mixed flow pump with tip clearance. Results show that the interaction area of leakage flow and main flow can be divided into three regions: (a) tip clearance zone, (b) shear flow zone, and (c) primary vortex zone. In the tip clearance zone, as roughness increases, the velocity magnitude and its derivative at the same wall distance gradually decrease, and the velocity angle at blade tip continuously increases in the inner circular boundary layer. As the roughness increases, the leakage flow gradually decreases, especially in the center chord region. In the shear flow zone, the velocity gradient rapidly decreases along the direction of pressure side to suction side, and finally, it becomes stable. As roughness increases, the velocity gradient intensity decreases, and the main reason is that the differences of velocity magnitude and velocity direction between the inner circular boundary layer and potential region decrease. In the primary vortex zone, the tip leakage vortex core indicated by the Liutex vortex method coincides with the low-pressure core location. As the roughness increases, the tip leakage vortex core becomes closer to the next blade. The pressure coefficient and vorticity of the tip leakage vortex core decrease with increasing of roughness, and the slopes remain constant on different cross sections with value of 5.9 × 103 and 2.4 × 103 m−1, respectively.

The axial-flow pump is a widely used piece of general machinery which consumes large amounts of energy. In this study, an axial-flow pump with the specific speed of 536 is firstly designed and experimentally measured; then, the orthogonal method is employed to conduct the energy performance optimization. Five optimization parameters, including hub control point, hub stagger angle, shroud stagger angle, camber angle and centroid position were set with four levels. Sixteen individual pumps were designed according to the orthogonal method; then, a numerical simulation was implemented to obtain the energy performance and flow pattern. Results showed that the shroud stagger angle has the maximum influence on the pump head and efficiency, and the hub stagger angle and camber angle are also very important. At a design point of flow rate 70 kg/s, the efficiency of the optimal pump is 86.29%, which improved by 2.05% in comparison with the baseline pump. The pressure gradient of the optimal pump from blade inlet to outlet becomes more fluent than that of baseline pump, and the low-velocity region of the optimal pump at the blade head shrinks, compared to that of the baseline pump.

Pressure fluctuations induced by the tip leakage vortex (TLV) in mixed flow pumps are the primary source of structural vibration and noise, which affects operation stability. The T shape blade tip is introduced to suppress pressure fluctuation of a mixed flow pump in this work. A closed-loop test rig of mixed flow pump is established to measure the pressure fluctuation on pump shroud wall with high-frequency pressure sensors. The numerical simulation is conducted to capture the detailed flow pattern and pressure fluctuation in pump with high-fidelity stress-blended eddy simulations. Results show that the T shape blade tip significantly reduces the spatially averaged pressure fluctuation intensity in the whole operating region, and the pressure fluctuation intensity reductions are 4.55% at part-load condition, 7.88% at design point condition, and 2.10% at over-load condition. Two mechanisms of pressure fluctuation suppression in the mixed flow pump are revealed. The first suppression mechanism is the improvement of circumferential uniformity in the time-averaged pressure field. The T shape tip effectively weakens the TLV strength and mitigates the low-pressure region on the shroud induced by the vortex, which suppresses the periodic pulses at the blade passing frequency. The second suppression mechanism is the improvement of vortex stability. The T shape tip controls the TLV trajectory to a position closer to the blade suction surface, which disrupts the feedback loop and prevents the periodic breakdown of vortex structure. Consequently, the broadband fluctuations caused by flow instability are suppressed.

Separated flows generate intense unsteady wall pressure fluctuations that are critical sources of aerodynamic noise and structural fatigue. Although reduced-order models (ROMs) have been widely used to represent such dynamics, many existing approaches remain largely data-driven and offer limited physical interpretability of the underlying instability mechanisms. The present study aims to develop a highly compact and interpretable ROM framework for representing wall-pressure spectral dynamics using a sparse set of physical parameters. To generate a reliable high-fidelity database, a modified delayed detached eddy simulation (DDES) model incorporating a Vortex Tilting Measure-based subgrid-scale definition is employed for a backward-facing step flow at ReH = 3.7 × 104. Spectral proper orthogonal decomposition (SPOD) is then applied to disentangle the multiscale dynamics, followed by the introduction of a Gabor wave packet model to analytically parameterize the leading SPOD modes. The modified DDES successfully mitigates the gray-area issue, accurately predicting the rapid shear-layer transition and a reattachment length of Xr/H ≈ 6.2. SPOD analysis demonstrates that the wall-pressure field is dominated by coherent structures associated with the Kelvin–Helmholtz instability, with the leading mode capturing 40%–60% of the local fluctuation energy. The proposed SPOD-Gabor ROM achieves extreme analytical compression, maintaining a high spatial reconstruction accuracy (R2 > 0.9) for the dominant fluctuations using only nine scalar parameters per frequency. Furthermore, the identified parameters exhibit consistent physical trends, explicitly capturing the asymmetric growth-decay behavior of the wave packets and recovering a convection velocity of Uc ≈ 0.55U0, in excellent agreement with classical free shear layer theory.

Cavitating flows are of great interest in the fields of hydraulic machineries, which can significantly affect mechanical performance and safety. Despite various efforts being dedicated to figuring out the interaction between flow and cavitation fields, their correlation has not been clearly addressed. To this end, in this study, a convolutional neural network, U-Net, was adopted to build a model that can predict the vapor volume fraction from velocity fields. Large eddy simulations of cavitating flows around a Clark-Y hydrofoil were conducted, and the simulated snapshots with velocity and vapor volume fraction were adopted as a dataset for training the network. The predicted vapor volume fraction shows good agreement with the referred simulation results, with a L₁ deviation lower than 2 × 10⁻⁴, considering all the snapshots. The comparable L₁ deviation between the training and validation datasets suggests the existence of a strong correlation between velocity and cavitation fields. The cavitation–velocity interaction derived from using U-Net suggests that the location with zero velocity indicates the interior part of attached and cloud cavitations, and the local vortical velocity fields usually suggest the existence of cavitation shedding.