Box-culvert axial-flow pumps are widely used in flood control, drainage, and water diversion projects; however, when operating under ultra-low or even negative head conditions, they are prone to severe flow separation, backflow, and intensified pressure fluctuations, which threaten operational safety and energy efficiency. To uncover the intrinsic instability mechanisms under these extreme conditions, this study investigates an engineering-scale box-culvert axial-flow pump using a combination of numerical simulations and physical model tests. Three representative operating conditions—design head (1.00Qbep), near-zero head (1.38Qbep), and negative head (1.44Qbep)—are analyzed to examine the internal flow structures, pressure pulsation characteristics, and energy dissipation behavior. By applying the Liutex–Omega vortex identification method, the evolution of flow structures is captured, revealing a continuous transition from locally stable small-scale vortices to large-scale reverse-flow vortices occupying the entire passage. Time-frequency analysis indicates that the pressure pulsations evolve from being dominated by the fourth harmonic of the rotational frequency at the design condition (4.0 fn), to a vortex-shedding-related characteristic frequency at near-zero head (0.75 fn), and finally to a pronounced low-frequency response around 0.25 fn under negative head conditions, reflecting a stepwise amplification of flow instability. Entropy-production analysis demonstrates that the total entropy production under negative head reaches 1.91 times that of the design condition, with turbulent dissipation accounting for more than 77% of the total. The major loss regions expand from the impeller to the guide vanes and outlet conduit, coinciding closely with zones of strong backflow and separation vortices. This study reveals a coupled mechanism characterized by vortex deterioration, pressure pulsation intensification, and enhanced energy dissipation under extreme low-head conditions. The findings provide theoretical guidance for operational optimization, structural improvement, and early warning of instability in pump stations subjected to extreme water-level variations.
Zhao et al. (Sun,) studied this question.