Post-elastic behavior of bolted connections in wood

A nonlinear finite element model is developed to study the behavior of single- and double-bolted timber connections with relatively low member thickness-to-fastener diameter ratios. These structural joints tend to fail in a brittle fashion. The established model is capable of predicting the post-elastic deformations of the connections locally and globally, and the unequal load fractions transferred by each bolt in a two-fastener connection.; The ADINA software package is used to generate the model. The problem presents two types of non-linearity. The first is due to geometric effects caused by increased sliding contact between the bolts and the oversized holes in the wood member. Contact is simulated using the Lagrange Multiplier algorithm available in ADINA. This algorithm enforces the compatibility of surface displacements at the wood member which is the contactor surface, and Coulomb frictional conditions over the contact segments, with a coefficient of friction of 0.7. The second non-linearity results from the material post-elastic behavior adjacent to the contact points. A plasticity-based compressive constitutive material model is developed to represent wood as elasto-plastic orthotropic according to the Hill yield criterion in regions of bi-axial compression. Linear elastic orthotropic response is applied otherwise. The model is incorporated as a user-supplied material model of ADINA to carry out the analysis.; The performance of the program and the effectiveness of the elasto-plastic material model are first verified with examples from the literature. Thereafter, numerical simulations of the post-elastic deformations of one- and two-bolt connections are compared to experimental results from tensile tests undertaken on glued-laminated timber connections with stocky bolts and subject to monotonic loading. Characterization tests are undertaken to obtain stiffness and strength for compression and tension parallel and perpendicular to grain, and shear, required as input in the new material model. Five configurations of single-bolt connections and four configurations of double-bolt connections have been tested with different combinations of end distance, edge distance, and spacing between bolts. Reasonable agreement is found between numerical and experimental load vs. strains obtained using single axis strain gages and strain rosettes mounted in regions of stress concentrations. The model is capable of tracing the post-elastic global deformation as obtained from experimental load vs. displacements of LVDT's mounted on specimens. The non-linearity of these load-slip curves is primarily caused by the inelastic wood deformation beneath the bolts at low load levels. This behavior explains the redistribution of load proportions among the bolts in a double-bolt connection. Under test configurations forces are unequally shared between the bolts even at the ultimate load. As a consequence, joint capacity as estimated with the proposed elasto-plastic material model is bounded by the so-called European Yield Model, which assumes equal load distribution among the bolts at ultimate, and the linear elastic model. Observed shear-out brittle failures of wood bolted connections are believed to be caused by excessive combined shear and tension perpendicular to grain along the sides of the contact zone.