Numerical modelling of adhesive joints debonding under fatigue loading

Abstract

Composite and hybrid metal/composite structures are nowadays present not only in the aerospace industry, but thanks to continuous performance improvement and cost reduction, also many more industrial fields are approaching the use of multimaterial structural elements. This requires, in turn, extensive use of adhesive bonding and a more and more sophisticated capability to simulate and predict the strength of bonded connections where, for this purpose, analytical methods are being progressively integrated or replaced by finite element analysis (FEA). To ensure the safety of the resulting structures, it is imperative to understand their fatigue behaviour. Thus the rise of the application of adhesive bonding has gone hand in hand with the development of models capable to predict the fatigue life that is related to the initiation and propagation of defects starting at free edges of joining regions or other features, such as throughthickness holes.The cohesive zone model (CZM) has found a wide acceptance as a tool for the simulation of debonding in adhesively bonded joints. This model is commonly used for the simulation of the quasi-static fracture problems, especially in the case of interface cracks such as in bonded joints and delamination in composites. The possibility to simulate the growth of a crack without any remeshing requirements and the relatively easy possibility to manipulate the constitutive law of the cohesive elements makes the cohesive zone model attractive also for the fatigue crack growth simulation.The purpose of this work is to develop the two-dimensional cohesive zone model presented by Pirondi and Moroni.A cohesive damage model has been implemented by means of the USDFLD and URDFIL subroutines in the commercial software Abaqus, in order to take into account the damage produced by fatigue loading.The CZ models were tested on various joint geometries characterized by different mixed mode ratios, in order to verify accuracy, robustness and performance in terms of computational time.