This work is the analyse of the results of several test campaigns using the composite fatigue know-how to understand, the composite fatigue of bonded joints. The static stress distribution of bonded joints is also considered in order to understand the behaviour choosing the best descriptive variables. As soon as fatigue of composites is a complex phenomenon, no crack initiation is allowed . The approaches employed to provide fatigue insensitive thresholds are empirical taking a global approach to analyze the results in a first step. Modified Wöhler curves have been adjusted following these directives. As first approach, they have been demonstrated to give good approaches of the phenomenon. Nevertheless it is difficult to establish accurate fatigue thresholds due to the lack of accuracy of the approach since it is based on global loads applied on the coupon and not on the detailed bonded joints stresses. The environmental conditions during the test, the type of adhesive, the specimen geometry and the type of load applied have been statistically demonstrated to be parameters modifying the mean-life of the bond layer. The remaining scatter could be attributed to the manufacturing process of the joints as well as the intrinsic test dispersion and the configuration discrepancy. After the analyses with a global approach, the following work has been focused on the establishment of a fatigue law based on local bonded joints stress. That is, a local law capable to approach the mean-life of an arbitrary bond layer. The proposed local criterion is based on static bonded joints analyses associated to the classical SN curves approaches. Until now, this criterion has been approached, with an acceptable precision, the fatigue meanlife of T-stringers in tension-compression and has validated the non-fatigue of omega-stringers of the test campaign.

Outgoing