Reynolds-stress model prediction of 3-D duct flows

The paper examines the impact of different modelling choices in second-moment closures by assessing model performance in predicting 3-D duct flows. The test-cases (developing flow in a square duct [Gessner F.B., Emery A.F.: {\em ASME J. Fluids Eng.} {\bf 103} (1981) 445--455], circular-to-rectangular transition-duct [Davis D.O., Gessner F.B.: {\em AIAA J.} {\bf 30} (1992) 367--375], and \tsn{S}-duct with large separation [Wellborn S.R., Reichert B.A., Okiishi T.H.: {\em J. Prop. Power} {\bf 10} (1994) 668--675]) include progressively more complex strains. Comparison of experimental data with selected 7-equation models (6 Reynolds-stress-transport and 1 scale-determining equations), which differ in the closure of the velocity/pressure-gradient tensor $\Pi_{ij}$, suggests that rapid redistribution controls separation and secondary-flow prediction, whereas, inclusion of pressure-diffusion modelling improves reattachment and relaxation behaviour.