Probabilistic assessment of the structural failure of single-lap bonded joints

Abstract

Single-lap bonded joints are a common typology, frequently used when designing structural systems. Several authors have studied their behaviour in different situations, involving both failure of the adhesive and also of the adherends. So far, the design of such joints has been based on deterministic postulates. In this work, a non-deterministic approach is followed. The reliability of the joint is assessed with several limit state approximation methods. Furthermore, sampling methods and a global reliability approach are considered. The algorithms are applied to the reliability analysis of a generic single-lap bonded joint, where the specific uncertainties related with this typology are taken into account, including mechanical properties and geometrical parameters of adherends and adhesive. All of them are defined as random variables and different cases are studied. Results show that first- and second-order second-moment and advanced mean value methods give an estimation to calculate the probability of failure at early stages of design, when precise values are not required. However, methods based on approximations of limit state surface around the most probable point of failure present a better response in terms of accuracy. Global reliability methods offer a good alternative to limit state approximations, as they present similar accuracy and usually require fewer iterations. Sampling methods often need extensive use of computational resources and are suitable for those cases with a large number of random variables, where computational cost is comparable with that required by other methods.