Abstract
Interfacial failure in adhesive joints has been studied. It has been shown by previous researchers that adhesive joints often contain bi-material geometric features that give rise to theoretically singular stress/strain fields even when there are no flaws present. It is reasonable to expect that failure occurs at, or close to, these points. To investigate failure under these conditions and to relate it to well understood failure mechanisms, tests were undertaken on uncracked joints (containing a bi-material singularity) and joint geometries with interfacial cracks of varying lengths. To provide comparative data under different conditions the joint tests were carried out under both mode I and mixed mode loading. The bulk properties of the adhesive used for the joint tests were also assessed. The rate dependent stress-strain behaviour was determined from a range of flat tensile tests. Similar tests were used to determine Poisson's ratio and to investigate the sensitivity of the adhesive to temperature variations. Four point bend tests on the adhesive were also carried out. A new method of processing the data obtained during such a test was developed and this was used to determine the ratio of the yield stress in compression to the yield stress in tension as a function of hardening. The results of these experimental programmes were then used as the basis for a series of finite element analyses. Firstly three-dimensional elastic analyses of the joint used in the experimental programme were carried out. As well as providing information on the stresses in the adhesive, these analyses showed that there was considerable variation across the width of the joint in the effective two-dimensional loading. A method of determining the correct loading was therefore developed and used for subsequent two-dimensional analyses. Three types of detailed two-dimensional analyses were carried out: elastic, elasto-plastic and a group of stress controlled separation analyses in which the material in front of the point of failure was allowed to separate under controlled conditions, thus simulating rupture. The results from these detailed two-dimensional analyses were used to assess the stresses in the joints at the failure loads and to determine the applicability of various failure criteria. Although no failure criterion was found to be universally applicable the results have indicated, clearly, possible directions for future development.