Highlights ► Experiments on 64 dowel-type steel-to-timber connections are performed. ► The effects of connections width, wood density, dowel roughness, and lateral reinforcement are studied. ► Individual stages during the loading process are identified. ► The suitability of the current design rules laid down in Eurocode 5 is assessed and deficiencies revealed.

Abstract Dowel-type connections are commonly used in timber engineering for a large range of structural applications. The current generation of design rules is largely based on empiricism and testing and lacks in many parts a stringent mechanical foundation. This often blocks an optimized use of the connections, which is essential for the design of economically efficient structures. Moreover, it severely limits the applicability of the design rule, such as restrictions regarding splitting behavior or maximum ductility (e.g. maximum allowable deformations) are missing. Therefore, the demands due to a large and quickly evolving variety of structural designs in timber engineering are not reflected. The aim of this work is to study the load-carrying behavior of the connection in detail, including all loading stages, from the initial contact between dowel and wood up to the ultimate load and failure. Distinct features during first loading as well as during unloading and reloading cycles are identified and discussed. The knowledge of the detailed load-carrying behavior is essential to understanding the effects of individual parameters varied in relation to the material and the connections design. The suitability of the current design rules laid down in Eurocode 5 (EC5) is assessed and deficiencies revealed. Tests on 64 steel-to-timber dowel-type connections loaded parallel to the fiber direction were performed. The connections were single-dowel connections with dowels of twelve millimeter diameter. The test specimens varied in wood density and geometric properties. Additionally, the effects of dowel roughness and lateral reinforcement were assessed. The experiments confirmed that connections of higher density show significantly higher ultimate loads and clearly evidenced that they are more prone to brittle failure than connections using light wood. The latter usually exhibit a ductile behavior with an extensive yield plateau until final failure occurs. With increased dowel roughness, both, ultimate load and ductility are increased. The test results are compared with corresponding design values given by EC5 for the strength and the stiffness of the respective single-dowel connections. For connections of intermediate slenderness, EC5 provided conservative design values for strength. Nevertheless, in some of the experiments the design values overestimated the actual strengths considerably in connections of low as well as high slenderness. As for the stiffness, a differentiation according to the connection width is missing, which gives useful results only for intermediate widths. Furthermore, the test results constitute valuable reference data for validating numerical simulation tools, which are currently a broad field of intensive interest.