Executive summaries of Action Groups of Group of Responsables in Aerodynamics

• AD(AG36) ‘Three-dimensional high lift computations’

This Action Group started in 1999, with the participation of FOI (Chair), ONERA, Airbus-D, QinetiQ and NLR.

AD(AG36) has investigated an area of great importance to the aircraft industry. In the past, the design of three-dimensional high lift systems has been relying on data sheet methods and two-dimensional flow computations (see AD(AG25)). This project has explored to what extent 2.5D and 3D flow computations can contribute to more accurate predictions of lift.

The aim of AD(AG36) has been to demonstrate the ability to compute three-dimensional high lift flows for a realistic transport aircraft wing. The wing chosen for the evaluation was the AFV wing (Fig. 1): a three-element wing featuring constant chord and sweep. Computations on this wing have been made assuming infinite span for the 2.5D case and limited span for 3D. The experimental results for this wing obtained from a test campaign in 1986 have been made available by ONERA. A few mandatory angles of attack were selected at constant Mach number and Reynolds number. Similar to AD(AG25), a database with all experimental and numerical results has been established.

Figure 1 ONERA-AFV model in the F1 wind tunnel

The main objectives in the group have been to:

· Investigate the validity of the 2.5D approach;

· Gain knowledge and experience in computing the incidence for 2.5D;

· Investigate how well stall can be captured numerically;

· Investigate the influence of turbulence models and transition location;

· Identify differences due to the application ofstructured and unstructured grids.

One common structured grid has been generated and a few unstructured grids. At an early stage, ONERA carried out stability calculations to determine the location of laminar/turbulent transition. These locations have been used as onset positions for the use of turbulence models. An investigation in the 2.5D case on the specification of laminar regions and comparing this with the application of fully turbulent flow revealed only small differences in the lift coefficients. All 3D computations were therefore carried out with fully turbulent flow.

The unstructured grids consisted of prismatic elements close to walls and of tetrahedrons elsewhere. Generation of multi-block grids around such geometries implies several man-months of work. It was therefore especially interesting that some partners choose to apply unstructured grids, which will be the natural choice in the future.

The 2.5D calculations require experimental input to determine the angle of attack. The numerical results depend on this angle and also on the numerical settings. These calculations also required a large number of iterations at higher angles of attack. At the highest computed angle of attack where 3D effects start to occur, only one of the partners was able to produce high quality results. In the linear range good correspondence of experimental and 3D numerical results have been obtained in general. The integrated lift on the wing and its elements obtained from the 2.5D calculations is illustrated in Figure 2 compared to the associated experimental data.

Figure 2 Lift coefficients from 2.5D calculations and experiments for all wing components

The three-dimensional computations for the highest angles of attack revealed strong three-dimensional effects with clear separation on the outer part of the wing. For angles of attack up to 29.7^o, all partners predicted lift coefficients within 3% from the experimental ones. At the highest angle of attack 37.8^o, results from some partners did and others did not predict stall, which implied a larger dispersion of the results. It would have been interesting if the computational points for the high angles of attack were sufficiently dense to enable estimates of the maximum lift for each partner. The fact that RANS computations in general do not to simulate flow in massively separated regions accurately puts a question mark to their applicability to predict the maximum lift. The computational results achieved by the AD(AG36) partners are, however, promising.

At lower angles of attack, the 3D and 2.5D results correspond well with each other. The three-dimensional numerical results show reasonable correspondence with experiments at maximum lift but the stall location is dependent on both the grid as well as the turbulence model. The stall is captured, though, by all partners starting at the tip and spreading inboard. The upper surface skin friction lines derived from numerical results by one partner at stall condition is shown in Figure 3. A similar quality of the numerical results is obtained with structured and unstructured grids at the expense of additional grid points when using unstructured grids.

Figure 3 Upper skin friction lines obtained from NLR computations at a=37.81°

The work of the Action Group has been carried out in a systematic way. The influence of important parameters has been thoroughly investigated. The computational results are quite impressive, demonstrating the applicability of three-dimensional RANS computations for accurate predictions of the lift coefficient for high lift systems.

Finally, AD(AG36) has been a valuable complement to the EU project EUROLIFT. It is clear that AD(AG36) has been a successful initiative from the GoR Aerodynamics Group.

.