No abstract

Progressive damage modeling (PDM) is a well-established methodology for the prediction of damage initiation and evolution in composite structures. However, as conventional PDM methodology involves a large post-processing procedure, it is impractical for application in large-scale structures due to the high computational cost it requires. In this study, the local character of nonlinearity, due to the scale of the damage topology compared to the size of the entire structure, is exploited to propose proper modifications in the 'classical' PDM methodology. Specifically, the sub-modeling technique principles are combined and integrated in the PDM methodology and the appropriate modifications required are discussed. Furthermore, two damage prediction indices, which are related to the criticality of damage state at specific sub-areas (material layers) of the structure are introduced, in order to achieve significant reductions of the required computational time. Both the improvements make the application of PDM in large-scale composite structures practically feasible; this is demonstrated in the case of a generic composite shear joint whose numerical model comprises over a million degrees of freedom.

PurposeThe purpose of this paper is to present an efficient engineering methodology for solving the problem of non‐linear (NL) damage and post‐buckling of large‐scale structures, which is of high importance mainly for the aircraft industry.Design/methodology/approachThe methodology takes advantage of the capabilities of finite element substructuring technique in the simulation of large/complex structures and exploits the advantages of local‐global analysis logic. The main innovation deals with the appropriate modification of superelement method, such that it can deal with NL behaviour and efficiently model the entire large‐scale structure. In this study, the proposed methodology is demonstrated in the treatment of geometrical non‐linearity and its efficiency is assessed in the case of a large‐scale fuselage section.FindingsA method capable of solving large‐scale NL problems by taking advantage of the linear response of the different model regions is developed.Research limitations/implicationsFurther development of the proposed method is required for handling other means of non‐linearity.Originality/valueThe proposed approach is advantageous in terms of computational effort over the corresponding conventional ones.

No abstract