UFR 4-08 Best Practice Advice: Difference between revisions
(No difference)
|
Revision as of 14:19, 7 April 2009
Orifice/deflector flow
Underlying Flow Regime 4-08 © copyright ERCOFTAC 2004
Best Practice Advice
Best Practice Advice for the UFR
The best practice advice is based on the only known and available numerical computation performed by Wang (1991).
Key Physics
The flow is characterised by an axi-symmetric separation at a sharp-edged circular orifice, leading to a highly turbulent shear layer which is subject to streamline curvature. The shear layer re-attaches some distance downstream from the orifice, producing a region of flow recirculation. The length of the recirculation region is a good overall indicator of predictive performance. This recirculation region has its greatest radial extent a short distance downstream from the orifice, at the 'vena contracta'. Overall there is a large loss in static pressure, due to substantial turbulence generation and subsequent energy dissipation.
Numerical Modelling Issues
To capture the main flow features, a two-dimensional axi-symmetric computation can be undertaken.
To obtain a mesh-independent prediction of the re-attachment length, a grid of at least 128 x 28 cells, refined near the orifice, is required - at least when combined with a flux-blended scheme for convection discretisation (50% upwind, 50% central difference).
Physical Modelling
The standard k-ε turbulence model with wall functions can be used to provide a very good prediction of the reattachment length, mean velocity field and wall friction coefficient. Whilst the turbulent kinetic energy field will be less well-predicted, it will still be found to match measurements quite well at most locations.
Application Uncertainties
The mean velocity and turbulence field is likely to be quite sensitive to the specification of the dissipation rate at the inlet cross-section. This can be specified according to fully-developed pipe flow.
Recommendations for Future Work
The application of a full Reynolds stress turbulence model should, in principle, provide more meaningful predictions of the turbulence field - in particular turbulence anisotropy.
© copyright ERCOFTAC 2004
Contributors: Martin Sommerfeld - Martin-Luther-Universitat Halle-Wittenberg