Modelling residual stresses and environmental degradation in adhesively bonded joints
2015-06-19T14:29:41Z (GMT) by
The aim of this research was to develop predictive models for residual stresses and environmental degradation in adhesively bonded joints exposed to hot/wet environments. Different single lap joint configurations and a hybrid double lap joint with dissimilar adherends (CFRP/AIIFM73 double lap joint), were exposed to different ageing environments in order to determine the durability of the joints and the effects of ageing on the failure load. Thermal residual stresses in bonded joints were investigated with analytical solutions and finite element modelling, first with a bimaterial curved beam to validate the modelling process and determine the most suitable method for calculating thermal stresses in bonded joints. It was found that none of the analytical solutions and 2D geometric approximations was fully able to describe the 3D stress state in the strip. The incorporation of geometric and material non-linearity into the models was necessary to obtain accurate results. The validated methods were then used predict the thermal residual stresses in bonded lap joints. The thermal stresses were found to be highest in joints with dissimilar adherends. Moisture uptake in bonded joints was investigated using Fickian diffusion modelling. Gravimetric experiments were used to determine the Fickian diffusion parameters for the bulk adhesive and composite adherends. Transient diffusion modelling was used to predict the uptake in bonded joints. It was seen that moisture diffusion is a fully three dimensional process, and the effects of moisture absorption can only be adequately studied using 3D FEA. The effects of swelling from moisture absorption in bonded joints were investigated using coupled stress-diffusion FEA models. Coupled stress-diffusion 3D FEA was used to predict the transient and residual hygroscopic stresses that develop in bonded lap joints as a function of exposure time in accelerated ageing environments, taking into account the effects of moisture on the expansion and mechanical properties of the adhesive and CFRP substrate. It was seen that moisture absorption induces significant stresses in the joints and markedly different behaviour was seen in the cases of absorbent and non-absorbent adherends. Hygro-thermo-mechanical stresses arising from the exposure of single and double lap joints with thermal residual stresses to hot/wet environments were investigated. In the single lap joints, a reduction in the stresses present in the adhesive was predicted, owing to swelling of the adhesive from moisture absorption. In the double lap joint with dissimilar adherends, exposure to hot/wet environments initially reduced the stresses in the joint when dry, followed by an increase in the magnitude of some stress components and reductions in others with increasing levels of moisture absorption. This led to a higher equivalent stress state in the adhesive than when dry. Thermal residual and mechanical strains predictions were validated with internal strains measured by neutron diffraction and surface strains measured by moire interferometry. Comparisons of predicted and measured thermal residual strains showed low levels of strain in joints with similar adherends. The magnitude of strains in the CFRP/AI double lap joint was significant, with the same spatial distribution and magnitude in both measured and predicted strains. The comparison of mechanical strains predicted by FEA and measured strains by moire interferometry showed good agreement. High magnification moire interferometry also confirmed the location of strain concentrations predicted by FEA. A path independent cohesive zone model (CZM) and a coupled continuum damage model were used to predict and characterise damage and failure initiation in bonded joints. Progressive failure prediction was calibrated in the cohesive zone model using the moisture dependent cohesive fracture energy of FM73. There was a reasonably good agreement with the experimental failure loads. This implementation of the cohesive zone model is limited by the ability of the interface elements used, thereby creating mesh dependency. The Gurson-Tvergaard-Needleman (GTN) coupled damage model was used to predict the effects of residual stresses on failure loads. However, this method is difficult to implement, given the numerous parameters required. The failure loads predicted by the GTN model were comparable with the experimental data when the joints were dry or wet. The damage models were capable of predicting the sudden crack growth and propagation seen experimentally.