Flow in a curved rectangular duct - non rotating

Underlying Flow Regime 4-04 © copyright ERCOFTAC 2004

Best Practice Advice

Best Practice Advice for the UFR

Based on the study test case and a review of related studies it is possible to make the following recommendations for best practice modeling of turbulent flow in a rectangular-section curved duct.

Key Physics

Turbulence, and the effect of wall curvature on turbulence

Boundary layer flow and separation

Generality of Advice Given

The best practice advice is based on data collected from numerous computations of the study test case.

Five computer groups submitted altogether 7 results for this test case as part of an ERCOFTAC workshop entitled “Data Bases and Testing of Calculation Methods for Turbulent Flows” in Karlsruhe from April 3 to 7, 1995.

Ten groups (all different from those who submitted results for the Karlsruhe Workshop) submitted altogether 25 different results for this case as part of a joint ERCOFTAC/IAHR Workshop, which took place in Chatou (near Paris) during April 25-26, 1996.

Numerical Issues

Discretisation — use a higher order scheme (second order or above) for momentum equations.

Grid and grid resolution — Use 200,000 to 300,000 grid points, with typically 50 - 60 grid points in the cross-flow directions. Even this resolution may not be high enough when low-Re versions of turbulence models are employed.

Boundary Conditions and Computational Domain

The results are sensitive to the distributions of inflow variables: therefore (if possible) use the correct (experimentally derived) distributions of inlet axial and swirl velocity.

Physical Modelling

Turbulence — Altogether, the results for this test case are not entirely conclusive. Certain stress equation models reproduce some of the details better than eddy-viscosity models but overall a clear-cut superiority is difficult to discern. However, when significant streamwise curvature effects on turbulence are present, as in the case of curved channel flow, RSM and ASM should yield superior results, and better predict the details of the secondary flow.

Near Wall Modelling — If wall functions are used, the near wall mesh should only be refined to the limit of their validity based on the y+ values.

Recommendations for Further Work

The effect of spatial resolution should be investigated, but this would require considerably finer meshes with more than 1 million grid points.

With the increased computer processing power now available, it should be feasible to analyse this problem using a more advanced turbulence modelling technique, such as Large-Eddy Simulation (LES).

Contributors: Lewis Davenport - Rolls-Royce Marine Power, Engineering & Technology Division