The aerodynamic design optimization method is accepted as a very usueful tool by the aerospace and mechanical industries. However, it has not been widely used in the industry since the computing cast is too expensive for practical 3-dimensional problems.
In this work, an efficient aerodynamic design optimization method is developed for practical large-scale problems. The gradient-based numerical optimization method is used to find optimum design and the flow fields are modeled by the compressible Navier-Stokes equations. The three ingredients of design optimization algorithm-flow solver, sensitivity method and optimization method-are modified or newly developed to improve the over-all efficiency of design optimization. The flow solutions are computed by a new multigrid DADI algorithm which is designed for fast convergence of steady-state solutions. A semi-discrete adjoint formula is derived from the discrete form of the flow eqation to compute more accurate sensitivities. The design equation, derived from the optimality conditions, is introduced to make a simulation design optimization algorithm. The flow and adjoint solver are tightly coupled with the optimization procedure. The flow and adjoint solvers are parallelized with the decomposition method.
The efficiency and accuracy of the present method are assessed by the drag minimization problems of wing and airfoil. The result shows that an optimum shape of 3-dimensional wing under transonic turbulent flow conditions can be found in less than 8 times of flow analysis cost.
