Global/Local Optimization of Aircraft Wing Using Parallel Processing

Abstract

The structural optimization of a cantilever aircraft wing with stiffeners and curvilinear spars and ribs is described. The decomposition technique for the wing structure is widely used for the optimization of a complex wing. The optimization procedure is divided into two subsystems: the global wing optimization, which determines the geometry and location of spars and ribs, and local panel optimization to further reduce wing weight. Because the design variables that are changed in the global step have an impact on those changed in the local step and vice versa, an iterative process that iterates between the global and local optimizations is employed. Particle swarm optimization and gradient-based optimization are used to perform integrated global/local optimization. Parallel computing is used, implemented using Python, to reduce the central processing unit time. The license cycle-check method and memory self-adjustment method are developed and applied in the parallel processing framework to optimize the use of the resources. The integrated global/local optimization approach has been applied to the NASA Common Research Model wing, which proves the methodology’s application with finite element method analysis of appropriate fidelity. A 40% weight reduction of the Common Research Model wing has been achieved by the global/local optimization at an acceptable computational time.