2D Boundary layers with pressure gradients (A)

Underlying Flow Regime 3-03 © copyright ERCOFTAC 2004

Best Practice Advice

Best Practice Advice for the UFR

This CS0 diffuser testcase is relatively simple to compute concerning the set-up of the flow and the boundary conditions. A grid of 61 x 121 nodes is recommended with equal spacing in the streamwise direction and a stretching in wall normal direction, which is consistent with the wall formulation of the CFD method. The use of a low-Reynolds number grid is recommended if allowed by the code and the wall treatment. The set-up in terms of the boundary conditions is described in chapter 6. The inlet profiles have to be taken from the experimental data, both for the velocity and the turbulence. Bulk profiles are not acceptable, as they alter the boundary layer development and the displacement effects. The outer boundary is to be treated as a slip wall using the contour obtained from the integration of the experimental velocity profiles.

The main issue for the accurate prediction of this case is the selection of the turbulence model. The best solutions, known to the author are from the SST model, the Spalart-Allmaras model and from ω-equation based Reynolds Stress models. Comparable results to the SST model have been obtained with ω-equation based Explicit Algebraic Reynolds Stress Models (Hellsten and Laine 2000).

Turbulence modeling is an ever-developing area and it is likely that more and more models will prove in the future to be able to handle this testcase. This will contribute significantly to the reliability of CFD methods for flows with pressure-induced separation.

The CS0 case does not allow judging the ability of turbulence models to properly predict re-attachment of the flow and flow-recovery downstream of a separation zone. It is known from other studies that all RANS turbulence models seem to have significant problems in that area. Particularly models, which predict the correct separation onset, are prone to give larger separation zones and slower flow recovery than indicated by the data. Note that this does not mean that the other models are superior in that respect. It is only due to a cancellation of errors that models, which fail to predict the separation onset, have little problems in the downstream region. The user is advised to keep this deficiency in mind when judging CFD simulations of diffuser flows.

Recommendations for Future Work

Develop model enhancements or new models, which can predict both, flow separation and flow recovery downstream of reattachment. Note that a highly accurate separation prediction capability is essential to make the second step. The improvement cannot be judged with the CS0 case, but the Buice and Eaton (1997) experiments can serve as a guideline.

Contributors: Florian Menter - AEA Technology