Hybrid Mesh Deformation for Aerodynamic-Structural Coupled Adjoint Optimization

Abstract

This paper demonstrates that a discrete coupled-adjoint aeroelastic shape optimization can be made more efficient with the use of different mesh-deformation algorithms for the fluid-structure interaction (FSI) simulations and the coupled-adjoint calculations within the optimization loop. Mesh deformation using radial basis functions (RBF) with only a subset of surface points is popular due to the efficiency and mesh quality produced by the technique. However, this technique can reduce the rate of convergence of the coupled-adjoint and even prevent the coupled-adjoint equations from converging. This paper proposes a hybrid mesh-deformation strategy to improve the efficiency of coupled-adjoint optimizations: use the RBF method with a data-reduction algorithm when deforming the mesh within FSI simulations but use the Delaunay graph mapping (DGM) method in the coupled-adjoint procedure. The DGM method increases the rate of convergence of the coupled-adjoint matrix, relative to the RBF approach with a data-reduction algorithm, with the additional benefit of being a faster method. Using this hybrid approach, an optimization in which lift and pitching-moment constraints are satisfied within the FSI simulation is performed. The results of the optimization, and the effects of the hybrid approach, are presented.