Complex Aerodynamic Optimization And Robust Design Method Based On Computational Fluid Dynamics

Aerodynamic optimization design is a new research area under the integration of modernCFD technologies and optimization algorithms, occupying an important part of aircraft design.With the development of aeronautic technologies, the performance requirement posited on theaircraft increases not only in their quantity, but also in their complexity. Therefore, aircraftoptimization design is faced with more and more complicated design requests, which lead to amany-objective optimization problem. At the same time, the number of design parameters in-creases as a result of enhanced aerodynamic configuration complexity and specified designdetails, which lead to a multivariable optimization problem. The robust design method shouldbe introduced to decrease the effect of uncertain factor to the performance of aircraft for en-gineering robustness concerns. For above problems, this article starts with the research ofaerodynamic simulation methodologies with the establishment of a set of CFD programs ofhigh reliabilities and efficiencies, placing the cornerstone for aerodynamic optimizations.Following is the optimization platform based on the loosely surrogate-model managementcoupling the research works on the aerodynamic parameterizations, optimization algorithmsand surrogate models. The investigations of aerodynamic robust design based on uncertaintyanalysis and complex aerodynamics optimizations problems are carried out utilizing theaforementioned optimization platform.The main research issues and achievements in this paper are as follows:1A set of programs solving RANS equations was established for aerodynamic configu-ration estimations for aerodynamic shape optimization. The precision of the programs wasverified though classical aerodynamic simulation examples. The Patch grid technique was re-alized based on multi-block grids to improve the computation efficiency, with the investiga-tions on its applying strategies. For laminar design, researches of local variable transitionmodel were carried out and introduced to the CFD program. Classical boundary layer transi-tion flows were simulated to validate the method, which laid a foundation for application oflaminar flow technology.2To solve the complex aerodynamic configuration optimization problem, three keytechnologies of optimization system, i.e. the aerodynamic parameterizations, surrogate mod-els and optimization algorithms were investigated. First, parameterization methods for airfoils,wings and other configuration were investigated, including CST method and FFD method,making parameterizations for shapes from airfoils to full aircraft configurations available. For detailed design of complex shape, the FFD parametric method based on NURBS basis func-tions was developed based on original regular Bezier’s FFD (Free Form Deformation) method.The Kriging surrogate model was introduced to replace the CFD calculations to quicklyachieve aerodynamic characteristics of aircraft. The influence of the dimension of the designvariables to the precision of the surrogate model was studied. It is shown that as the dimen-sion of the design variables increases, the precision of the surrogate model decreases sharplyand the number of samples constructing the model quickly increases, referred as the curse ofdimension. The multi-variable optimization problem was carried out based on traditionalLoose Surrogate Management Framework optimization method, and a phenomenon of “preci-sion frozen” of surrogate model was detected during the optimization. The surrogate modelconstruction method based on space-decomposition was proposed to solve the difficulties ofthe surrogate model in high dimensional space. The collaboration optimization managementframe based on space-decomposition surrogate model was proposed. The refined aerodynamicshape design of BWB (Blended-Wing-Body) transports was carried out using the method, tovalidate the reliability and efficiency of physical hierarchical optimization methods.3Many-objective optimization was investigated. In order to overcome the defects of tra-ditional evolutionary multi-objective optimization algorithms in many-objective optimization,such as poor optimization results, difficulties in convergence and difficult to further deci-sion-making and so on. The PCA (Principal Component Analysis) analysis method was usedto extract the primary objective of a many-objective problem. To solve the problems of PCAmethod, a Multi-layer Hierarchical Constraint algorithm was proposed based on modifiedε-constraint method and PCA method, the rotor airfoil aerodynamic design optimization wassystematically studied using this method. The optimization results show that the Multi-layerHierarchical Constraint solves the many-objective problems of rotor airfoil design well. Anadvanced rotor airfoil family was eventually gained.4The robust aerodynamic design optimization was studied. In order to overcome the de-fects of the traditional uncertainty analysis method (Monte Carlo sampling methods), such asin the computation cost and efficiency, the PCE (Polynomial Chaos Expansion) method wasintroduced. The analysis and calculation efficiency is greatly improved because the uncer-tainty response model can be constructed based on a decreased number of quadrature nodesby PCE method. A robust design model was established considering the uncertainty of aero-dynamic flight condition by coupling PCE uncertainty analysis and design optimization sys-tem. The method was validated and compared with robust design methods based on tradition-al based MCS (Monte Carlo Simulation) simulation by robust design of supercritical airfoil. The results show that this method is efficient and reliable. The robust optimization of the nat-ural laminar flow supercritical airfoil was carried out base on the system...