Evaluation of the Spalart–Allmaras and k–ω GEKO Turbulence Models Relative to k–ω Shear Stress Transport for Centrifugal Pump Simulations With Experimental Validation

Abstract

Abstract. This study examines the efficacy of the Spalart–Allmaras and k–ω Generalized k-ω (GEKO) turbulence models in modeling centrifugal pumps in comparison to the commonly utilized k–ω shear stress transport (SST) model. To guarantee precision and reliability, numerical simulations were conducted utilizing computational fluid dynamics (CFD), and the results were juxtaposed with experimental data. The research is segmented into two phases. The initial phase involves numerical experimentation utilizing the pump's computer-aided design models, succeeded by simulations of various flow circumstances employing all three turbulence models. The subsequent step involves laboratory validation, wherein prototypes are tested under identical settings to assess the correlation between numerical predictions and experimental outcomes. The assessment emphasizes essential performance indicators, including velocity and pressure distributions, pump efficiency, and computational duration. The results indicate that the k–ω GEKO and k–ω SST models yield similar conclusions, albeit with differing levels of accuracy depending on the flowrate. At high flow rates, the Spalart–Allmaras model demonstrates the lowest error of 2.71%, succeeded by k–ω SST at 14.74%, and k–ω GEKO at 15.64%. At low flow rates, the k–ω SST model has the maximum accuracy with a 4.08% error, closely succeeded by the Spalart–Allmaras model at 4.57%, but the k–ω GEKO model displays a higher inaccuracy of 6.20%. Furthermore, the Spalart–Allmaras model functions with almost twice the computing efficiency of both GEKO and SST. This study underscores the importance of selecting an appropriate turbulence model, offering valuable insights for engineers and academics engaged in pump design and optimization. These findings highlight the significant impact of turbulence model selection on the accuracy and computational cost of centrifugal pump simulations.