For this research project, two airfoils have been optimized using a Direct Search optimization algorithm and a cost function determined from the results of a far-field drag decomposition method. The latter is a powerful tool allowing to breakdown the drag into wave, viscous, induced and spurious drags. The latter type of drag is caused by numerical and truncation errors, as well as by the addition of artificial viscosity by most solvers to smooth strong gradients. Furthermore, the spurious drag is dependent on the configuration: a blunt body will produce more spurious drag than a slender body. Thus, if an optimization process is based on the total drag it will tend to find a configuration that reduces among others, the spurious drag which can limit its efficiency. The optimization process in this research used the net drag only, excluding the spurious drag. First, the NACA0012 airfoil in an Euler flow at Ma = 0.85 was optimized. The final configuration had a flat nose shape and an almost constant thickness along the chord. The computed net drag was 74 d.c, an improvement of 393 d.c. An additional control optimization was done, but on the total drag. The optimized configuration was more rounded, which is a direct consequence of including the spurious drag in the objective function. This shows that the spurious drag has a large influence on the optimum airfoil. Second, the RAE2822 airfoil in viscous flow with a constant lift coefficient of 0.824 and a Mach number of 0.734 was optimized. The final configuration was thinner than the original airfoil on the first 50% of the chord length, then got thicker for the rest of the chord length. The configuration also showed a cambered trailing edge typical of supercritical airfoils. The computed net drag value was 104.3 d.c., an improvement of 83 d.c. Most of the improvement had been achieved by the wave drag reduction.
展开▼
