Stiffness and slip in multi-dowel timber connections with slotted-in steel plates

Abstract

Large multi-dowel connections can provide the strength and ductility required for large, highly-loaded timber structures, but their slip under load is not well understood. This is an important gap in knowledge, because accumulated local displacements at connections represent a large part of the deformation of a timber structure. The empirical relationships used in design codes commonly scale a single-dowel stiffness by the number of dowels, so do not capture the dowel interaction effects of the multi-dowel connections used in larger structures. We present the results of an experimental test series, elastic model and probabilistic numerical analysis investigating the development of stiffness in multi-dowel timber connections with slotted-in steel plates. Novel test methods record the development of stiffness due to each individual connector to show that the stiffness of the complete connection is not proportional to the number of dowels. An elastic stress-function model shows that this is partly due to interaction of the stress field around the dowels. For the first time, this work quantitatively considers the influence of misalignment of dowels due to manufacturing tolerances, and it is shown that this may greatly reduce the overall stiffness of a multi-dowel connection. The test series is used to validate a probabilistic model of this misalignment for the stiffness of such a connection. The model incorporates the nonlinear stiffness and hole opening observed in single-dowel connections to predict the behaviour of the group. The study shows that the random misalignment of dowels in multi-dowel connections reduces the range of displacements over which the connection displays zero stiffness slightly, but that this zone is not eliminated as a result of irreversible hole opening under load, even for a connection with 35 dowels and three steel plates. We conclude that two parameters are important for the design of these connections: the unload-reload stiffness and the zero-stiffness region measured between the zero load intercept of the unload-reload linear fit. With these, a reasonable estimate can be made of the displacement at any serviceability load level in either tension or compression.