Abstract Cavitation in low-head, high-flow-rate axial-flow pumps can create a hidden performance risk in which hydraulic efficiency deteriorates before the conventional breakdown point defined by a 3% head drop. This study investigates a axial-flow pump using a unified framework integrating recirculating-loop experiments, high-speed visualization, scale-resolving computational fluid dynamics (CFD), unsteady-load diagnostics, and entropy-generation-based loss decomposition. An energy-oriented stable boundary, σs, is introduced together with the critical cavitation number, σcrit, to identify an efficiency pre-degradation window (σcrit s ss). Within this window, hydraulic efficiency decreases by approximately 4-6% while the head remains nearly unchanged. As s or the available net positive suction head (NPSHa) decreases, phase-locked casing-pressure signals decorrelate and intermittent broadband bursts emerge near the suction-side tip region. Stripwise spectral analysis localizes the dominant unsteady energy disturbance to the outer tip-dominated radial region, where tip-leakage-vortex cavitation develops. Entropy-generation-based analysis further reveals a rapid loss reallocation: the phase-change-related residual loss fraction increases from 1.81% to 11.30% and further to 25.31% under severe cavitation, while the outer tip region persistently contributes 76.70-79.26% of the liquid-phase entropy-generation-based loss, and the loss hotspots migrate toward the downstream pressure-recovery region. The coupled evolution of normalized total loss, phase-change-related residual loss fraction, and broadband response provides an interpretable indicator set for early warning and tip-focused cavitation mitigation in axial-flow pumps.
Liang et al. (Fri,) studied this question.