# UFR 4-05 Best Practice Advice

# Curved passage flow

Underlying Flow Regime 4-05 © copyright ERCOFTAC 2004

# Best Practice Advice

## Best Practice Advice for the UFR

Key Physics

The physics of the underlying flow regime of a **curved passage flow** is governed by the interaction of the pressure gradient linked to the curvature of the main flow with the non-uniform flow. The flow, moving slower in the boundary layers, is pushed from the pressure side to the suction side, leading to an overturning in the end-wall regions.

The best practice advices for this underlying flow regime are based on the analysis of 29 simulations performed on the DLR turbine stator and coming from two main studies.

Numerical Issues

Grids and grid resolution

- An adequate resolution in the radial direction is required. This is particularly true outside of the near-wall region where the refinement of the mesh is often neglected.

- At least 81 grid points in the spanwise direction is advised.

- In order of importance the grid should be refined in the spanwise, pitchwise, and streamwise directions.

- Nearly 500,000 adequately distributed grid points is required to get a good simulation of annular turbine cascade.

Computational domain and Boundary conditions

Boundary conditions

- Use radial equilibrium boundary condition at the outlet boundary.

- A more complex outlet boundary condition is required in order to perfectly match the downstream pressure field.

Physical modeling

Turbulence modeling

- In this accelerating flow the effect of the turbulence model is reduced. Therefore, the Baldwin-Lomax model is appropriate to reasonably well predict this type of flow regime.

- In order to better capture gradients of total pressure ratio, the Low Reynolds Chien should be preferred to other
*k-ε*models.

- If a value of y
^{+}greater than unity is used at the first grid point near solid wall, the Low Reynolds Chien model should be preferred to the Low Reynolds Yang-Shih model.

Near wall model

- A value of y
^{+}of 1 or less at the first grid point near solid walls is required.

Application uncertainties

Because the velocity profile was not measured at the entrance of the stator, uncertainties have been identified on the inlet flow angle and velocity distributions derived from the total pressure profile and the extrapolated pressure.

Recommendation for future work

As already noticed computations performed with the Spalart-Allmaras model in the scope of the AGARD study were performed at a time when this turbulent model was not mature enough. Therefore, it is here recommended to performed new computations with this one-equation turbulence model together with computations using more recent SST and V2F models.

© copyright ERCOFTAC 2004

Contributors: Nouredine Hakimi - NUMECA International