EFFICIENT AEROSTRUCTURAL DESIGN OPTIMIZATION COMBINING GRADIENT-ENHANCED KRIGING WITH COUPLED ADJOINT METHOD

Aerostructural design optimization is crucial for green aviation due to its capability of balancing low weight and high aerodynamic performance. However, aerostructural design optimization suffers from numerous expensive coupled analyses, each requiring multiple iterations of computational fluid dynamics (CFD) and finite element method (FEM) simulations, as well as data transfers and mesh deformations. To tackle the challenges, an efficient global aerostructural design optimization method combining the gradient-enhanced Kriging (GEK) model with coupled adjoint method is developed. Firstly, the coupled adjoint equations are derived and solved by linear block Gauss-Seidel method. Secondly, the total derivatives are computed and used to assist direct GEK model for the surrogate-based optimization. Finally, the GEK-based optimization method is verified against analytical test case and applied to aerostructural design optimizations of 36- dimensional Drag Prediction Workshop W1 (DPW-W1) and 72-dimensional NASA Common Research Model (CRM) wings. It is observed that, GEK method using coupled adjoint gradient information is much more efficient than traditional Kriging method only using response information.