Author: Gómez Royuela, J.L.
Supervisor: Majano-Majano, A. & Xavier, J.
University: Higher Tech. School of Architecture of Madrid (ETSAM) | Tech. University of Madrid (UPM)
Abstract
Mechanical dowel connections are a widely used solution in timber structures due to their ease of assembly and disassembly, rapid execution, and high load-bearing capacity. Their potential to develop ductile failure behaviour makes them particularly interesting. However, under certain conditions, stress concentrations perpendicular to the grain and shear stresses can arise, potentially leading to brittle failure by splitting, especially in dowel connections loaded at an angle to the grain.
Splitting failure in these types of connections is directly related to the fracture processes of the material. However, there is no general consensus on how to adequately address this issue, either in terms of analytical expressions for estimating their load-bearing capacity or in terms of characterising the fracture properties of the material. Eurocode 5 provides a very simplified expression, based on fracture mechanics, to determine the splitting capacity of dowel joints loaded at an angle to the grain. However, this expression does not consider the influence of many geometrical parameters of the connection, material properties, and is limited in its applicability to softwood.
The primary objective of this thesis is to contribute to overcome these limitations in the case of European beech (Fagus sylvatica L.) beams loaded perpendicularly to the grain using dowel connections, by adopting a non-linear fracture mechanics approach. To this end, both experimental work and numerical simulations are carried out to investigate the influence of different parameters related to the connection geometry. In particular, the impact of using one or two dowels to transfer the load to the beam, the spacing between the dowels in the longitudinal direction of the beam, and their proximity to the supports are analysed. Additionally, the influence of beam slenderness and the effect of arranging a second connection symmetrically with respect to the centre of the beam are also studied.
In the numerical simulations carried out to reproduce the splitting failure in beams, cohesive zone models are employed to simulate both damage initiation and crack propagation. These models require the implementation of the elastic and fracture properties of the material, which have been obtained experimentally in this thesis. The twelve elastic constants that define the orthotropic compliance matrix of the European beech are determined by compression tests with strain measurements using digital image correlation techniques. Fracture properties in loading modes I and II are obtained from DCB (Double Cantilever Beam) and ENF (End-Notched Flexure) tests, respectively. The cohesive laws describing the fracture behaviour of the material are obtained by applying the direct method using the CBBM (Crack Based Beam Method) procedure to determine the fracture energy. This method has the advantage of not requiring crack growth measurement during the test, which is particularly beneficial for wood. The numerical results are compared with the experimental results and with those obtained from the formula included in Eurocode 5 and two other analytical reference models for the estimation of the splitting capacity.
Based on the work carried out, the influence of the studied parameters on the splitting capacity is demonstrated, highlighting the need to update the current code expression. The comparison between the experimental, numerical and theoretical failure loads reveals that the analytical models tend to estimate splitting capacity conservatively, particularly the expression included in Eurocode 5.
Finally, the proposed numerical model proves to be an effective tool to reliably predict the splitting capacity of European beech beams with dowel connections and different configurations.
This doctoral thesis is presented as a compendium of publications. Its content is articulated through four contributions made by the author and published in indexed scientific journals (JCR Q1).