For many years now, composite materials have been used for aeronautical and space structures. In order to join composite parts together or with metallic parts, classical bolted assembly technologies are generally employed, after being slightly adapted to these new configurations. Thus, for composite bolted joints the load transfer is mainly performed through the contact between the fastener’s pin and the composite hole. This type of contact generates high local stress concentration around the hole, which then requires to be reinforced. One solution consists in adding a metallic ring inside the hole, which allows reducing the load along a part of the composite hole. Different processes can be used: expansion, bonding or simple setting.CapAero developed EB² (Expanded Bonded Bushing), an innovative technique for ring reinforcement by expansion and bonding. This process provides the advantage of not requiring any specific device for bonding, since the adhesive lies in its microencapsulated form and is already part of the semi-product. The expansion of the pre-coated ring, with a mandrel, mainly aims at crushing the capsules inside the adhesive and mixing the two components together, but also at setting up a light radial preload inside the composite. The combination of expansion and bonding gives the joint better stiffness and failure strength for bearing loads, since the presence of the adhesive allows the load reduction all around the hole, and not only along a part of it.For the metallic case, the ring design usually relies on hardening criteria but for composite joints, a damage initiation is not desirable. Consequently, the design rules are totally different. For now, no design tools dedicated to composite yet exist. A particular design method has then to be defined, and its application requires the control of the expansion rate applied. Indeed, a low expansion does not allow improving mechanical performance, but a too high expansion can lead to premature bearing damage in the composite. Given the potential combination of manufacturing tolerances in stakes in the joint, only taking into consideration the nominal optimal geometry is not enough.The present study first introduces the development of an axisymmetric finite element model, for a robust predesign. This model is used to run a high number of simulations, which takes into account different manufacturing variabilities. According to specifications, the analysis of residual stresses and circumferential strains allows the choice of proper ring geometries. The finite element model is then numerically validated by a previously developed 3D finite element model which reproduces accurately the expansion process. Finally, the model predictions are compared to the results of experimental tests, for several values of expansion rate.The experimental test run has two objectives: the first one is to validate the numerical model, by measuring the expansion load and strains around the hole, for numerous geometries of specimens and for different variations of processes. The second one is to identify which expansion rate generates the best adhesive activation. Thus, after expansion, push-out tests are run on specimens, in order to estimate the joint quality by comparison of push-out loads between all configurations.The study brings conclusions about the relevance of such a tool and its limitations, and also the perspectives for this work, which could ultimately allow the study of bolted structures.< Réduire