Variable amplitude fatigue of adhesively bonded joints.

Adhesive bonding is the most attractive joining technique for many aerospace applications. One of the most important reasons for this is the superior fatigue performance of bonded joints compared to traditional joining techniques, fatigue being considered as one of the most important design concerns for aerospace structures. Previously, efforts have been made to develop lifetime prediction methods for bonded joints under constant amplitude (CA) fatigue. Although CA fatigue conditions can be assumed for some real structures, a more complex load history is likely to be expected for many aerospace applications. However, a problem arises from the fact that materials can behave very differently under variable amplitude (V A) loading than they do under CA conditions. Cycles with different stress levels can interact and thus lead to acceleration or retardation of the fatigue process. Despite this, there is very little work on the V A fatigue of bonded joints. Therefore, in this work, the effect that V A fatigue has on the initiation and propagation of damage in bonded composite joints was studied and a predictive methodology for joints SUbjected to complex loading regimes was developed. In this study, both double lap joints (DLJ) and double-cantilever beam (DCB) specimens were tested. The applicability of the Palmgren-Miner (P-M) rules and numerical crack growth integration (NCGI) to bonded joints subject to a block-loading V A spectrum were investigated. The P-M rules failed to predict the fatigue life of double lap joints and NCGI failed to predict both the experimentally observed cohesive and interlaminar crack growth in bonded composite joints under V A loading. In all cases, severe fatigue crack growth acceleration was reported. This is obviously of some concern as traditional predictive methods will tend to overestimate the actual fatigue life of bonded components. In order to improve the ability to predict V A fatigue in bonded joints two novel useful predictive methodologies were developed: the 'Linear Cycle Mix' (LCM) model for uncracked joints and the 'Damage Shift' model for double cantilever beam (DCB) joints. The LCM method is based on the observation that the mean stress variations, i.e. transitions from a CA stage to another stage having a higher mean stress value, can be responsible for fatigue crack growth accelerations (i.e. the 'cycle mix' effect). This method proved to be a considerable improvement on traditional cumulative damage laws. The Damage Shift model requires the modification of NCGI to incorporate the effect of the damage zone induced by the overloads. This study showed that the method can be used to explain unusual sudden crack jumps during the initial stages of V A cycling and fatigue crack growth acceleration due to overloads. It is suggested that the Damage Shift model may be applicable to a variety of complex fatigue spectra.