Best Practice Advice AC1-08

From KBwiki
Revision as of 15:31, 5 August 2009 by Tonyh (talk | contribs)
Jump to navigation Jump to search

Front Page

Description

Test Data

CFD Simulations

Evaluation

Best Practice Advice

L1T2 3 element airfoil

Application Challenge 1-08 © copyright ERCOFTAC 2004


Best Practice Advice for the AC

Key Fluid Physics

The main flow physics associated with this application challenge are strong interactions between turbulent boundary layers and wakes, wake-wake mixing and boundary layers subject to adverse pressure gradients. The relevant UFRs are UFR3-01 (Boundary Layer - Wake Interaction NLR 7301), UFR3-03 (2D Boundary Layers with pressure gradients) and UFR3-04 (Laminar-turbulent boundary layer transition).

Application Uncertainties

2.1 Transition Location

Transition location is not specified and therefore not modelled on the slat and flap. Experience suggests that the location of transition, particularly on the slat, has a significant impact on the overall forces, especially at higher incidence. In addition, there is evidence to suggest that, at high lift, transition on the upper surface of the main element is forward of the applied transition fix.

2.2 Wake Resolution

The location of the wakes that develop downstream of the slat, wing and flap trailing edges is not known in advance. Furthermore, the location of these wakes will depend on the onset flow conditions. This makes it difficult to construct a mesh that will adequately resolve both the flow structure in the wakes and in the interaction regions between wakes and boundary layers.

There is no evidence to suggest that the prediction of the pressures on the lifting surfaces is particularly sensitive to the mesh resolution across the wakes and boundary layer/wake interaction regions. However, this may be because the mesh used in the application challenge was of sufficient resolution to capture the essential flow features of the developing wakes, wake/wake and wake/boundary-layer interactions. It is not clear what the limits of sensitivity are.

The prediction of the boundary layer and wake profiles for the L1T2 test case differ in a number of respects from the experiment. There is a tendency to over predict the wake deficit levels and the wake centre locations differ from experiment. Also, the interaction between wakes is under predicted by the CFD. These discrepancies in total pressure deficit will affect the accuracy of the drag predicted by the CFD.

It is not possible to give clear best practice advice for the effect of grid resolution on the structure of the wakes and wake/boundary layer interaction regions. This advice is not given in UFR3-01 (Boundary Layer – Wake Interaction NLR 7301). In addition, the sensitivity of the flow structure in the wake-wake interaction region on grid resolution is unknown as no UFR is currently available for wake-wake interactions.


Computational Domain and Boundary Conditions

• If a far-field circulation correction method is used at the far-field boundary, ensure that the far-field boundary is at least 15 chord lengths away. If not, ensure that the far-field boundary is at least 50 chord lengths from the body.

• Use no-slip and adiabatic boundary conditions on solid surfaces.

• The test case may be considered to be wholly two-dimensional. The test case may also be considered to be in free-air as the experimental data has been corrected for solid blockage and the effects of wall constraint. Thus the uncertainty in free stream boundary conditions is very small.

Discretisation and Grid Resolution

• Use a scheme with as little numerical dissipation as possible (At least second order accurate in space)

• Use a mesh with y+ values about 1.0 near solid walls, and 20 to 30 grid points placed up to y+ = 100.


Physical Modelling

• Use the full, compressible, Reynolds Averaged Navier-Stokes formulation.

• If an accurate prediction of the pressures on the lifting surfaces (and so lift coefficient) is required, USE the k-ω turbulence models [1],[2].

• It is not possible to give best practice advice on which turbulence model to use in order to give an accurate prediction of the profiles of boundary layers and wakes. The advice given in UFR3-01 for accurate prediction of boundary layer/wake interactions is to use the Spalart-Allmaras or k-ω turbulence models. This advice appears to be inconsistent with the evidence in this Application Challenge. This inconsistency may be a result of one or more of the following:

  • Inadequate mesh resolution of the wakes, wake-wake and wake-boundary layer interactions for the L1T2 test case (See Application Uncertainty).
  • UFR3-01 does not feature wake-wake interactions
  • UFR3-01 has less demanding physics than the L1T2 AC (i.e. less severe adverse pressure gradients)


Recommendations for Future Work

• A grid sensitivity study should be undertaken for the L1T2 test case. In particular, this study should focus on the sensitivity to grid resolution of the profiles of the wakes, wake-wake and wake-boundary layer interaction regions. Guidance on the required resolution should be obtained from the UFR3-01 (Boundary Layer – Wake interaction).

• The UFR3-01 pressure gradients are much less severe than for the L1T2 test case. A UFR with pressure gradients that are more representative of practical applications such as the L1T2 application challenge should be identified and studied. Furthermore, a UFR should be identified with wake-wake interactions from which best practice advice can be derived.

• The L1T2 test case should be repeated using the Spalart-Allmaras turbulence model to ensure consistent advice with the UFR3-01.

• The UFRs should contain best practice advice on the required mesh resolution around each of the lifting surfaces. In particular, the mesh spacing at critical locations such as leading and trailing-edges should be specified, and number of cells on upper and lower surfaces.


References

1. D.C. Wilcox, “Reassessment of the scale determining equation for advanced turbulence models”, AIAA J. 1988, Vol 26, pp1299-1310.

2. F.R. Menter, “Two-equation eddy viscosity turbulence models for engineering applications”, AIAA Journal, Vol 32, pp1598-1605, 1994.


1 It is not possible to give best practice advice on which turbulence model to use in order to give an accurate prediction of the profiles of boundary layers and wakes. The advice given in UFR3-01 for accurate prediction of boundary layer/wake interactions is to use the Spalart-Allmaras or k-ω turbulence models. This advice appears to be inconsistent with the evidence in the Application Challenge. This inconsistency may be a result of one or more of the following:

Þ Inadequate mesh resolution of the wakes, wake-wake and wake-boundary layer interactions for the L1T2 test case (See Application Uncertainty).

Þ UFR3-01 does not feature wake-wake interactions

Þ UFR3-01 has less demanding physics than the L1T2 AC (i.e. less severe adverse pressure gradients)


© copyright ERCOFTAC 2004



Contributors: Antony Hutton; Jan Vos - QinetiQ; CFS Engineering SA


Front Page

Description

Test Data

CFD Simulations

Evaluation

Best Practice Advice