Comparative research on various turbulence closure models for unsteady cavitation flows in the ultra-low specific speed centrifugal pump

29th IAHR Symposium on Hydraulic Machinery and Systems

Wang, C.^1^2 ; Zhang, Y.X.^1^2 ; Zhang, J.Y.^1^2 ; Li, Z.W.^1^2 ; Xu, A.^1^2

College of Mechanical and Transportation Engineering, China University of Petroleum, Beijing

102249, China^1

Beijing Key Laboratory of Process Fluid Filtration and Separation, China University of Petroleum, Beijing

102249, China^2

关键词: Experimental visualization;    Global sensitivity analysis;    Low specific speed centrifugal pump;    Predictive capabilities;    Shear-stress transport;    Turbulence closure models;    Unsteady cavitating flows;    Unsteady characteristics;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/240/6/062038/pdf
DOI  :  10.1088/1755-1315/240/6/062038

来源: IOP

【 摘 要 】

This paper is aimed to develop a numerical framework for simulating unsteady cavitating flows in the ultra-low specific speed centrifugal pump. Various turbulence models, including RNG (Re-Normalization Group) k- model, SST (Shear Stress Transport) k-ω model and FBM (Filter-Based model), were modified by density correction and applied to predict explicit performances, cavitating and pressure fluctuation characteristics of an ultra-low specific speed centrifugal pump. Afterwards, comparisons were made with experimental visualization and measurement results and the numerical results among those models. It can be concluded that the modified turbulence viscosity can fit well with the hydraulic performances measured by the experiment, but the unsteady characteristics of the cavity shapes and pressure fluctuation were predicted well only by filter-based density corrected model (FBDCM), which take the local meshing resolution into consideration. Lastly, a global sensitivity analysis was conducted to assess the limitation of the filter size λ. The results indicate that when λ is less than 3 times the largest grid size, the numerical framework has a reasonable overall predictive capability, and it is expected to choose large value of λ for the computational efficiency.