Calibration of CFD grid for simulating the impeller clearance flow and axial hydraulic force of centrifugal pump

Abstract

The axial hydraulic force of centrifugal pump is an important parameter affecting pump performance. The force mainly includes the force inside impeller and in clearance. Due to the special structural characteristics of the clearance, the influence of grid discretization method on the calculation of axial force in the clearance is not fully understood. Therefore, based on the Reynolds-averaged method with shear stress transport turbulence model, an orthogonal experiment was designed to compare the correlation coefficient of velocity and pressure distribution between shear stress transport model and large eddy simulation models. A more suitable grid discretization strategy was found by artificial neural network for grid calibration. When the strategy is applied to the entire centrifugal pump, the prediction of axial force has high accuracy. The range analysis shows that the grid node number in the wall-wall direction has the greatest impact on velocity distribution. When the mesh parameters are in a certain range, it can compromise between the simulation accuracy and computational resource. The Reynolds-averaged model based simulation is proved accurate in capturing the complex velocity and pressure field inside clearance. The entire pump model is also used for the verification after the calibration of grid. The typical axial force law can be found under different flow rate conditions. This study provides a significant guidance in determining the grid scheme for accurate prediction of centrifugal pump’s axial force. It makes the computational fluid dynamics simulation feasible in the initial design of centrifugal pump which specifically considers the axial force problem.