ASSESSMENT OF FREE SURFACE/SUBMERGED VORTICAL FLOW STRUCTURES AND GAS ENTRAINMENT AT PUMP INTAKES USING RANS CFD

The assessment of liquid flow patterns forming due to a submerged pump intake, and its associated gas entrainment phenomena, need to be reliable and accurate for a number of design applications. The use of conservative correlations can lead to gross over-design or, inaccurate predictions in certain complex suction arrangements. Scale model testing often has scale effects which leads engineers to rely on other analysis methods. Simplified analyses can fail to correctly predict the vortical features near the suction intakes and associated gas entrainment, which may lead many applications to use Computational Fluid Dynamics (CFD) as a method for intake flow predictions. However, reliable assessment of gas entraining into a suction intake with high Reynolds number internal flow, involves several physical modeling and numerical challenges. An accurate assessment requires, (1) a transient model, (2) proper resolution in the wall boundary layers near the intakes, (3) resolution of the swirl in the vortical structures, and (4) modeling of the laminar-to-turbulent transitional effects in the vicinity of the intake. This paper addresses and evaluates the performance of several turbulence model options, and modifications within a CFD code for the assessment of intake flow. The results presented in this paper, are compared against published experimental data on vortical flow patterns at intakes, to identify physical and numerical modeling needs for such assessments. The results show that while, adequately resolved meshes, and low Reynolds number RANS (Reynolds Averaged Navier Stokes) turbulence models can capture the location and relative strength of vortices, the results for flow details inside the vortical structures identified need to be interpreted carefully, considering the relevant flow physics and modeling limitations. The turbulence model modifications proposed involve curvature correction and scale adaptiveness.