Improvement of internal flow characteristics, pressure pulsations, and axial forces in the hump region of pump-turbines using a non-dominated sorting genetic algorithm combined with multi-resolution dynamic mode decomposition

This study addresses the hump characteristics observed in mixed-flow pump-turbines under partial load conditions. A multi-objective optimization design for the movable guide vane profile was achieved using the response surface method combined with the non-dominated sorting genetic algorithm (NSGA-III). Through numerical simulations of the pump-turbine, the study systematically analyzed the vortex characteristics in the impeller, movable guide vane, volute, and fixed vane regions before and after optimization, along with the energy dissipation mechanisms within the volute and fixed vanes. Frequency-domain analysis examined pressure pulsations in the vaneless space, movable guide vane region, and fixed vane region. Multi-resolution Dynamic Mode Decomposition (MRDMD) analyzed axial force variations in the impeller. Results demonstrate that the optimized movable guide vane profile significantly mitigates head fluctuations in the hump region. It effectively reduces flow incidence angles, suppresses recirculation and vortex phenomena, and produces smoother, more orderly streamline distributions. The optimized high-entropy production area in the volute and fixed vane regions was reduced by over 60%. Multi-frequency peaks in the movable guide vane region and non-dominant frequency peaks in the fixed vane region were effectively suppressed. Axial force vibrations in the pump-turbine primarily exhibited mid-to-low frequency characteristics dominated by rotational frequency and its harmonics. This study significantly improved the operational stability and efficiency of the pump-turbine in the hump zone by optimizing the movable guide vane profile. It provides an effective design methodology and theoretical basis for enhancing the low-load performance of mixed-flow pump-turbines.