An Integrated Global/Local Optimization Framework for Subsonic Wing

An optimization framework has been developed for the aircraft wings with curvilinear spars and ribs (SpaRibs), and stiffeners. The development of additive manufacturing technologies like electron-beam-free-form-fabrication (EBF³), electron-beam melting (EBM) and friction stir welding (FSW) have opened up the opportunities for the design concept of reinforcing the wing structure using curvilinear stiffening members. The global aircraft design is characterized by multiple disciplines: structures, aeroelasticity and buckling. The wing structural optimization can be divided into two subsystems: the global optimization which optimizes the geometry of wing skins and SpaRibs; and the local optimization which focuses on the local panel thicknesses and stiffeners. The local panels are optimized considering the stress and buckling constraints. In the stiffened panel optimization, the curvilinear or straight stiffeners are added on the local panels to improve the buckling resistance. The geometry of SpaRibs has influence on the optimization of stiffened panels. The study of the interaction between the global optimization and the local optimization is usually computationally expensive. An approximate approach is implemented for the stiffened panel optimization to obtain approximately optimal panels using the polynomial curve fitting of a series of optimized panel thicknesses, so as to reduce the computational time. Particle swarm optimization is used in the integration of global/local optimization to optimize the SpaRibs. The optimization framework is developed in MATLAB and Python environment to incorporate the MSC.PATRAN for geometry modeling, and MSC.NASTRAN for finite element analysis. The integrated global-local optimization approach has been applied to subsonic NASA common research model (CRM) wing which proves the methodology’s application scaling with medium fidelity FEM analysis. The results have been presented, showing a significant wing weight reduction can be achieved at an acceptable computational cost.