Experimental and computational evaluation of strut losses in a diverging annular diffuser.

摘要

An experimental and computational study was performed to evaluate the performance of several strut configurations spanning an annular diffuser under a wide range of inlet swirl conditions. Four strut configurations were investigated to determine the influence of the number of struts and the strut profile shape. Experimental tests were conducted on a cold flow windtunnel over a range of Reynolds numbers. Computational studies were conducted using the commercial Computational Fluid Dynamics (CFD) program Fluent.; The experimental work focused on evaluating the ability of the various strut configurations to assist in pressure recovery from the swirling component of flow through the mechanism of flow straightening. Oil flow visualizations were conducted to obtain information regarding the size and structure of the horseshoe vortices present at the strut endwall intersections under no-swirl conditions. Visualizations also provided information about the complex aspects of the flow field associated with flow separation from the struts under high inlet swirl conditions. The influence of fillets at the strut endwall intersections was also briefly investigated.; The CFD work evaluated the ability of various turbulence models to predict experimental flow field; with focus on the ability to predict the horseshoe vortices under no-swirl conditions, and to predict the downstream redistribution of the wake flow present under high inlet swirl conditions. The influence of near wall treatment selection was evaluated as an integral part of this investigation. The work contains extensive grid development studies. The influence of inlet condition specification, including velocity profile and turbulence intensity, has also been investigated.; The CFD had a tendency to over-predict the pressure recovery and under-predict the total pressure loss; the ideal pressure recovery was generally well predicted. The choice of turbulence model and near wall treatment was shown to influence the predicted magnitudes of pressure recovery and total pressure loss coefficients. The standard k-epsilon turbulence model used in conjunction with the enhanced wall treatment was found to provide the best agreement with the experimental performance parameter magnitudes.