Multi-Component Adaptive Aerodynamic Shape Optimization Methods For Civil Transport Aircrafts

In recent years,with the development of computational fluid dynamics(CFD)theory and computing resources,the aerodynamic design process of aircraft has become more and more elaborate.To pursue the optimal performance,aerodynamic shape optimization methods have drawn plenty of attention and become an important design tool in practical engineering problems.As a typical application of optimization methods,transonic civil aircraft requires high fidelity,complex configuration,and high-dimensional design capability.However,the existing optimization methods still have some weaknesses in these cases,for example the problem of how to extract the drag reduction mechanism,the multimodality issue for optimization algorithm,the inadequate capability for designing different kinds of components and the issue of insufficient robustness and efficiency in the optimization process.To tackle with these issues,Reynolds-Averaged Navier-Stokes(RANS)equations are used for the following research.The main works of this study are as follows:1、Hybrid axial velocity defect(AVD)formulas are proposed to tackle with the flaw of existing formulas in far-field drag decomposition(FDD)method.A vorticity based sensor is proposed for both viscous and induced drag calculation.Using small perturbation assumption and Gauss’ s theorem,the definitions of all drag components are derived.All existing formulas are integrated into the same formula framework to use the hybrid formulas.Since wind tunnel experiment results for each drag component are unavailable,a family of test cases including two-dimensional,three-dimensional and viscous,inviscid cases are carried out to show that FDD method is able to capture the varying trend of drag with respect to flow physics.2、As for optimization algorithm,a hypothesis on the non-dimensional sizes of the neighborhood of local minima is proposed to quantify the probability of being trapped in a local minimum using probability theory.Based on this theory,a stop criterion is proposed for random multi-start gradient-based optimization methods to ensure the probability of being trapped in a local minimum to be lower than a threshold value.Considering typical aerodynamic design problems,the multimodality feature of transonic viscous RAE2822 case from ADODG(Aerodynamic Design Optimization Discussion Group)case 2,the inviscid subsonic straight wing planform optimization case from ADODG case 6,and the transonic CRM(Common Research Model)wing shape optimization case from ADODG case 4 are analysed.FDD method is used to reveal the changes in each drag component and the effects of different types of design variables on drag and multimodality feature.The application scenario and usage strategy of gradient-based optimization algorithm are introduced.3 、 Coordinate transformation and virtual frame method are proposed to get the generalized coordinate free-form deformation(FFD)method,which is improved to be applicable to more kinds of shape.Taking cylindrical coordinate system as example,coordinate transformation is used to reformalize the FFD formula and to capture the deformation characteristics of cylindrical components such as fuselage and nacelles,which leads to better design capability.The virtual frame method is proposed to preserve first order and second order derivative continuity of object at the boundary of FFD frame.The proposed method is tested and verified in a cylinder deformation case,a three-dimensional nacelle shape fitting example,and a blunt fuselage head shape optimization problem.4、Dynamic FFD parametrization and optimization method based on B-spline knot insertion are proposed to achieve higher robustness and efficiency.Optimality and slope of objective history are both used as adaption trigger.An adaption metric function is constructed using gradients and constraint Jacobian for selecting the insertion interval for new knot values.The flow solver,adjoint based gradient solver,optimization algorithm,and mesh deformation method are coupled to establish a conventional gradient-based optimization system and a dynamic gradient-based optimization system.In the cylinder case,the conventional system fails to converge while the dynamic system converged with given tolerance.In the transonic CRM wing optimization case,the dynamic method is shown to save 5% iterations compared with the conventional method.5、Using the optimization system,transonic civil aircraft with conventional layout are chosen as the research object.In the wing shape optimization of CRM wingbody configuration,shock-free result is achieved under the non-decrease thickness constraint with0.00077 drag reduction.The FDD result shows that 24.7% of the total reduction comes from induced drag reduction,which means it cannot be ignored.With random multi-start optimizations,only one optimum is found.Wingtip is then optimized and winglet shape is obtained.Coupled optimization of wing shape and winglet provided 0.00122 larger drag reduction than combined drag reduction from separate wing and winglet optimizations.The result shows that adaptive parametrization gave smooth shape and better performance.Besides,the coupled optimization of wing and winglet provides more drag reduction potential than separate optimizations.As for CRM wing-body-nacelle-pylon configuration,the nacelle shape is optimized first to demonstrate the drag reduction potential and the wing shape is then optimized.Shock-free results is obtained with 0.00115 drag reduction,which is larger than the reduction in CRM wingbody configuration,and the induced drag reduction contributes 34.9% to the drag decrease.This result shows that the optimization system can optimize both wing and nacelle shape with high efficiency and good results,and coupled optimization gives larger drag reduction than separate optimizations.