Integration of a global/local modelling approach into the multi-scale optimization method of thin-walled structures: application to a wing-box architecture

Air traffic passenger demand is constantly growing of about 5% per year. This will led to a substantial increase in the emission of gases that contribute to global warming, in contrast with the general demand to reduce the environmental impact of aviation. Thus, new aircraft structural configurations have been studied and, among these ones, there is the PrandtlPlane (PrP) configuration which follows a Prandtl’s intuition of the so called “Best Wing System”. The PARSIFAL project aims to design an innovative aircraft, based on the PrandtlPlane configuration, in order to meet the general demand of reducing the environmental impact of air transportation. The realization of an innovative aircraft entails the development of new design methods in order to compensate for the lack of statistical data required for the preliminary stages. This thesis has been carried out within the general framework of an EU funded project named PARSIFAL. The main goal of the present work is the development of a numerical optimization strategy aimed at generating, evaluating and improving the structural efficiency of thin-walled aeronautical structures. The strategy consists of two main numerical tools: the ERASMUS (EvolutionaRy Algorithm for optimiSation of ModUlar Systems) optimization tool and the APDL (Ansys Parametric Design Language) Finite Element software.Particular focus is placed on the design of a conventional wing-box structure, used as a benchmark for this work, in which all the geometric parameters concerning the inner layout are provided as design variables within the ERASMUS optimization software.To realize this procedure, Finite Element (FE) models have been generated by scripting parametric codes through the use of APDL software. In this way, the design variables are chosen by the ERASMUS software at each optimization cycle, within a pre-defined range, and used as input parameters for the generation of the FE models of different wing structures.The objective of the optimization design procedure is the minimization of the structural mass, under certain mechanical and technological requirements which are formulated as constraints for the optimization process. Main structural components are verified for each of the generated FE models through specific analyses performed at two different design scales. Thus, a static analysis is solved for the overall wing-box structure, i.e. for the global model, while a linear eigenvalue buckling analysis is performed on the second level local model, represented by a particular bay. The local model is extracted automatically from the previous global model according to a precise criterion, based on the most loaded repeating unit. Moreover, the global-local FE design approach is used to change the element types of the stiffener members (from beam to shell) and to refine the mesh size when passing from the global to the local.Implementing the Multi-scale Optimization Method in this way, it has been possible to and a minimum weight layout for the wing-box structure through the evaluation of several configurations. The analyses carried out in this thesis have verified the parametric code used for generating the FE models. Moreover, such analyses also allowed to highlight important aspects to be taken into account for the future applications