High-fidelity Aerodynamic Shape Optimization Method Of Aircraft Based On Computational Fluid Dynamics

With the development of computer technology, Computational Fluid Dynamics (CFD) is playing a more and more important role in the design of aircraft. In the meantime, the elaborate design of modern aircraft gives a higher request to the accuracy of aerodynamic prediction and the design techniques. The combination of high accuracy CFD method and the efficient optimization method will be an effective tool for improvement of aerodynamic performance. Numerical simulation and aerodynamic shape optimizations are performed for supersonic and hypersonic flows in this thesis by utilizing the CFD and optimization methods. The main research contents are as follows:1. A parallelized CFD solver is developed and then applied to the simulation of turbulent and transitional flows. Based on the fully coupled RANS method, the effects of several factors (such as the flux schemes) on the convergence of computations are analyzed. Performance evaluation of the k-co type turbulence models are also performed. A new simplified transition model named y-SST is proposed based on the γ-Reθt model. Validations show that the model can still predict the phenomenon of boundary layer transition accurately. At the meantime, the new model is more concise and computational effective. Besides, compressibility corrections for the γ-Reθt transition model is studied.2. An optimization method and the corresponding procedure, which possesses the abilities of single objective and multi-objective optimizations, local and global optimizations is developed. Techniques of parametric modeling and grid generation are firstly developed for aircraft optimization. Attention is then focused on the particle swarm optimization (PSO) and the quantum-behaved PSO (QPSO) algorithms. Improvements are made for their better convergence speed and accuracy. The radial basis function (RBF) surrogate model is improved and a surrogate-based flowchart is developed for higher efficiency of optimization.3. Typical aerodynamic shape optimizations are carried out by the optimization method developed. Optimization results show that the gradient-based optimization method possesses the merit of excellent performance of convergence and rapid searching of local optimum. The high-fidelity surrogate-based optimization method can reduce the computational cost substantially and hence is very efficient.In general, the aerodynamic optimization method developed in this thesis is demonstrated to be effective and practical for engineering problems.4. Uncertainty quantification, global sensitivity analysis and robust shape optimization are performed. The effect of uncertainty on the aerodynamic performance is quantified by the polynomial chaos expansion method. A robust optimization method is developed by combination of the response surface surrogate model and the uncertainty analysis method. Robust optimization of a supercritical airfoil is performed and results show that the drag of the airfoil is significantly reduced by both the deterministic optimization and the robust optimization. Meanwhile, the shape obtained by robust optimization has the best characteristic of drag divergence.