Framework for Multi-Level Optimization of Complex Systems

Complex systems, such as those encountered in aerospace engineering, can typically be considered as a hierarchy of individual elements. This hierarchy is reflected in the analysis techniques that are used to analyze the characteristics of the system as a whole. Consequently, a hierarchy of models is to be used, each accounting for different physical scales, components or disciplines. This chapter presents a framework for the multi-level optimization of such complex systems. The framework is demonstrated on maximizing the range of a supersonic business jet via multi-level optimization. Flexibility of the framework is shown considering two different decompositions of the model and two different coordination procedures. The latter, are illustrated using a simple two-bar truss structure. Furthermore, via adjusting algorithm settings a significant reduction of computational costs is shown at the cost of finding a design with marginally lower range.