Evaluation of Various Turbulence Models and Large Eddy Simulation for Stall Prediction in a Centrifugal Pump

Abstract

Rotational stall is an unstable flow phenomenon that reduces the performance of centrifugal pumps, usually occurring under partial load conditions. It causes instability in the flow resulting in intense vibrations and noise under certain flow conditions. In this study, the one-equation Wray–Agarwal (WA) turbulence model, which was recently developed, is employed to numerically simulate the internal flow field of a centrifugal pump under the deep stall condition. The aim of this study is to examine the prediction accuracy for stall by using the WA turbulence model. The method based on computational fluid dynamics (CFD) has been widely applied for investigation of complex flow patterns in pumps by solving Reynolds-averaged Navier-Stokes (RANS) equations. Particle image velocimetry (PIV) experimental results were compared with simulations predicted using the WA, renormalization group (RNG) k−ε, shear stress transport (SST) k−ω, and realizable k−ε turbulence models and large eddy simulations (LES). The comparisons indicated that the WA turbulence model can accurately predict the flow separation and has a good agreement with the PIV data. The WA model adds a cross-diffusion term and a blending function to the eddy viscosity R equation, so that this model could be expressed as a one-equation k−ω model or one-equation k−ε model as needed by using the switching function. The results show the strong potential of the WA model for accurately computing the stall in rotating fluid machinery. The outcomes of the study are useful in development and optimization of fluid machinery with a low calculation cost and a high accuracy.