Sensitivity analysis of turbulence models for internal flow and vibration characteristics in centrifugal pumps

Abstract

To address the complex internal flow characteristics and diverse unsteady excitation sources in centrifugal pumps, this study systematically investigates the predictive capabilities of different turbulence models on internal flow features and vibration responses. Taking an IS65-50-125 centrifugal pump as the research object, numerical simulations covering a wide flow rate range (0.2Q–1.2Q) were conducted using four turbulence models: Standard k–ε, Renormalization Group k–ε, Baseline k–ω, and Shear Stress Transport (SST) k–ω. Comprehensive validation was performed through external characteristic experiments and vibration measurements. Results demonstrate that the SST k–ω model achieves the most accurate predictions regarding static pressure distribution, velocity field structures, and vortex system evolution. Its simulations of impeller radial forces and pressure pulses in critical regions show high consistency with experimental vibration trends. In terms of hydraulic performance, this model achieves the smallest prediction errors of head and efficiency under all operating conditions. The head error ranges from 0.57% to 3.24%, and the efficiency error ranges from 0.47% to 2.64%. Its superiority is especially remarkable under low-flow conditions. In terms of vibration response, the average prediction error of vibration displacement at the three measuring points calculated by this model is <8.33%. The study reveals that the SST k–ω model combines near-wall accuracy with robustness in core flow regions, achieving optimal comprehensive performance in multiphysics coupling simulations. These findings provide a scientific basis for computational fluid dynamics model selection and low-vibration design in centrifugal pumps.