Four turbulence models are applied to the numerical prediction of the turbulent impinging jets discharged from a circular pipe measured by Cooper et al. [Int. J. Heat Mass Transfer 36, 2675-2684 (1993)], Baughn and Shimizu [ASME J. Heat Transfer 11 1, 1096-1098 (1986)] and Baughn et al. [ASME Winter Annual Meeting, November 1992]. They comprise one k-epsilon eddy viscosity model and three second-moment closures. In the test cases selected, the jet discharge was two and six diameters above a plane surface orthogonal to the jet's axis. The Reynolds numbers were 2.3 x 10(4) and 7 x 10(4), the flow being fully developed at the discharge plane. The numerical predictions, obtained with an extended version of the finite-volume TEAM code, indicate that the k-epsilon model and one of the Reynolds stress models lead to far too large levels of turbulence near the stagnation point. This excessive energy in turn induces much too high heat transfer coefficients and turbulent mixing with the ambient fluid. The other two second-moment closures, adopting new schemes for accounting for the wall's effect on pressure fluctuations, do much better though one of them is clearly superior in accounting for the effects of the height of the jet discharge above the plate. None of the schemes is entirely successful in predicting the effects of Reynolds number. It is our view, however, that the main cause of this failure is the two-equation eddy viscosity scheme adopted in all cases to span the near-wall sublayer rather than the outer layer models on which the present study has focused.
