|
Executive summaries of Action Groups of Group of Responsables in Aerodynamics
TThis GARTEUR Action Group was established in September 1999, with the participation of BAe, DERA/ARA now QinetiQ, DLR, INTA, NLR, ONERA, AIRBUS France, Dassault Aviation. FOI/FFA left the group during the first year. The objectives of the group were threefold: validation of Navier-Stokes calculations with respect to the prediction of laminar boundary layer parameters; implementation of transition criteria in the NS codes; validation of the computational strategies used by the partners. The work done by the group is given in details in the final report . Only some typical results and the main conclusions will be given here. To validate the NS computations with respect to the prediction of the laminar boundary layer parameters, the AIRBUS "B-profile" was considered with an imposed onset transition located as in experiment: x/c=0.42 and 0.3 on the suction and pressure sides, respectively. Due to the particular shape of the airfoil, the pressure coefficient remains nearly constant over the first 40% of the chord and a transition bubble exists in the experiment (RC=2 106, M0=0.15, a=7o). As can be seen in figure 1, all the solvers give very similar pressure distributions with the same over prediction of the pressure coefficient over the rear part of the upper side. To illustrate the key problem for transition prediction, figure 2 compares the computed shape parameter from the ONERA NS solver with the corresponding value from a boundary layer solver using the same pressure distribution. The shape parameter is clearly underestimated although the small value of the artificial dissipation coefficient. As transition criteria are very sensitive to the shape of the velocity profile, a few 10-3 uncertainty on Hi is significant. The conclusion of the first step was that present transition criteria cannot be directly used in NS solvers. In view of these difficulties, two strategies have been chosen by the partners: DLR, INTA and QinetiQ decided to couple their NS solvers with a boundary layer code; NLR used a NS solver which already included transition criteria adapted to 2D industrial high lift configurations; ONERA also chose to introduce transition criteria, with minor changes when necessary, in his NS solver. Both strategies proved to be successful. To illustrate this, only a few typical results obtained during the first validation step will be presented here but validations have also been done for the CAST10 transonic airfoil and a high lift multi-element configuration. With the coupling strategy, the pressure distribution given by the NS solver is periodically used as input of the boundary layer code which, in return, gives the transition location with a high precision level using the best up to date criterion (eN criterion with direct stability computation or data based methods). The details of the various implementations are given in the report. Figures 3 and 4 show the establishment of the solution obtained by DLR for the Somers airfoil (RC=4 106, M0=0.1, a=-6.2o). The eN method was used and the N-value at transition was set to 11. The initial transition location was imposed at 75% of the chord on both sides. All the transition locations determined directly by the data base were under relaxed before the RANS solution continued, except for the last call. This ensures that the eN criterion is applied to laminar boundary layer profiles. Results obtained by INTA with a similar strategy for the "B-profile" are given in figure 5. The initial and final skin friction distribution are given and compared to the results of the Euler/boundary layer MSES solver with the same transition criterion. The initial transition location was fixed to x/c=0.6 on the suction side. On the pressure side the transition location was fixed at x/c=0.5 as in the experiment. Results obtained with the second strategy, consisting in the direct use of transition criteria in NS solvers, are illustrated by figure 6. The Arnal "AHD" criterion is based on the linear stability theory. It uses the Mack relationship to take into account the external turbulence level. To use it in the NS solver, the shape parameter coming from the direct integration of the velocity profile is no longer used but replaced by the local pressure parameter through the Falkner-Skan self similarity solutions. This reduces the sensitivity of the criterion to the poor quality of the velocity profile prediction in the NS solver. By doing so, it is even possible to apply the criterion to a turbulent velocity profile. This simplifies the iteration process by suppressing the need of under relaxation process. In figure 6, the skin friction coefficient distribution obtained with the Arnal and the Abu Ghannam & Shaw criteria are compared to the results of the boundary layer solver (3C3D) for Tu=0.1%. The final solution is the same either starting from a transition location at the rear part of the airfoil or using the solution for Tu=1%. In the BL solver, an intermittency function is used contrary to what is done in the NS solver. The results obtained during the AG35 activity show that transition prediction can be used in Navier-Stokes computations with two different strategies. However, the AG35 was restricted to 2D flows. The extension to 3D flows must now be considered, at least for relatively simple configurations. Moreover, other important phenomena could also be considered, such as transition in separated laminar bubbles and the intermittency region. A new GARTEUR Exploratory group has been proposed and accepted (EG52) to extend the AG35 work.
|
