Aerodynamic Optimisation of a Morphing Leading Edge Airfoil

A morphing leading edge airfoil is optimized for aerodynamic performance with different objectives. A constant arc length parameterization employing the Class/Shape Transformation technique is built to limit the axial deformation introduced by morphing. The optimization is performed with a standard methodology based on genetic algorithms, comparing the results for three different aerodynamic models: a potential flow solver with boundary layer calculation (XFOIL), a fully turbulent RANS model (Spalart-Allmaras) and a transitional RANS model (gamma-theta). Whereas the solutions obtained with the third model are standard droop nose shapes, those found via transitional models show an uncommon deformation with an upward leading edge deflection. Development of optimization strategy is also performed by building an hybrid procedure based on a metamodel-assisted approach. Several nonlinear regression methods are investigated to compare the accuracy in fitness approximation and an Artificial Neural Network (ANN) was finally selected. Application of the improved algorithm to a probelm previously solved with a standard approach shows that the use of a surrogate model, combined with a gradient based method for local individual improvement, is able to provide a reduction of the convergence effort when approximating the highly nonlinear relationship between the constant arc length parameterization and the aerodynamic behavior predicted with gamma-theta model.