Search

International audience; A multi-level analysis of skin/stiffener debonding is used for the fuselage design of future aircraft during postbuckling. The specimens composed of a laminate (the skin) to which an over-thickness (the flange) had been added were subjected to four-point bending, which led to interface failure between the flange and the skin. These tests were performed with several configurations and parameters, such as the orientation of the plies located at the interface, temperature (-50°C, 20°C and 70°C), ageing and manufacturing mode: co-cured or co-bonded. The flange shape (tapered or not) and thickness were also considered. Test data are presented and analyzed and critical configurations are identified. Finite element models were developed and the flange debonding loads computed, firstly by use of cohesive models and then through a fracture mechanics approach (Virtual Crack Closure Technique). In both cases, the Benzeg-gagh-Kenane criterion was selected and proved its efficiency but the fracture mechanics approach was an order of magnitude less time consuming, which will enable future modelling to include larger sizes.

Omega stringers offer interesting structural capabilities and are expected on future aircraft fuselages. In postbuckling mode, the final failure of these structures may occur by stringer debonding between stringer flanges and the skin of the fuselage. In this study, it is demonstrated that the use of fracture mechanics allows to predict skin/omega stringer separation under multiple load cases. Three different load cases and experiments are presented allowing a debonding to start at different locations: at free flange edges or at the inner radius of the omega. Firstly, a skin/stringer configuration subjected to three point bending following the longitudinal axis of the stringer was tested. For this configuration, a numerical study was made and shows the influence of a refined mesh taking into account resin fillets. Secondly, new specimens were obtained by cutting into slices the longitudinal specimen. Those specimens were subjected to four points bending. It has been shown that the upper rolls position of the test jig could modify the debonding location. Numerical models have allowed to determine accurately the debonding location and the associated load level. For some specimens, resin fillets were removed from the flange tips and their effect were assessed numerically and experimentally.

International audience; In this paper, some aspects of a multiscale strategy for the analysis of cracked structures are presented. The purposes of this strategy are, first, to separate the local effects from the global effects in order to keep a macromesh unchanged during the crack's propagation and, second, to enable one to use a proper fine-scale description only where it is required. Here, the propagation aspect is not addressed. We focus on the crack modeling. Two main aspects are discussed: the first is the choice of the macroscale and its associated discretization in order to include the macroeffect of a crack; the second is the use of a decomposition of the domain into substructures and interfaces in order to limit the use of the refined scale only to where it is required.