Investigation and Optimization of Connections in Timber Assemblies Subjected to Blast Loading

Abstract

The majority of research on high strain-rate effects in timber structures has been limited to the study of the load-bearing members in isolation. Limited work has been conducted on timber connections and full-scale timber assemblies under blast loading, and these have generally been constrained to qualitative observations. In North America, the increasing prevalence of mid- and high-rise timber structures makes them susceptible to blast effects. In addition, questions remain on how to design and optimize these timber assemblies, including the connections, against blast loads, due in part to the limitations on comprehensive design provisions. The effects of far-field blast explosions were simulated using the University of Ottawa shock tube. A total of fifty-eight dynamic tests were conducted on connection-level and full-scale specimens. The research program aimed to investigate the behaviour of heavy-timber connections when subjected to simulated blast loads. The experimental results showed that connections with a main failure mechanism consisting of wood crushing experienced significant increases in dynamic peak load when compared to the static peak load. In contrast, connections where steel yielding and rupturing occurred experienced no statistically significant increase in dynamic peak load. Full-scale glulam specimens with bolted connections designed to yield via wood crushing and bolt bending performed better than those with overdesigned connections. Bolted connections which failed in splitting led to premature failure of the glulam assembly. Reinforcement with self-tapping screws allowed these bolted joints to fail in a combination of bolt yielding and wood crushing, and provided more ductility when compared to unreinforced specimens. Specially designed energy-absorbing connections significantly increased the energy dissipation capabilities of the timber assemblies. The basis of these connections was to allow for connection yielding while delaying failure of the wood member. This was achieved via elastoplastic connection behaviour, which effectively limited the load imparted onto the wood member. Based on the experimental results, limitations in the current Canadian blast provisions were highlighted and discussed. A two-degree-of-freedom blast analysis software was developed and validated using full-scale and connection-level experimental results and was found to adequately capture the system response with reasonable accuracy. Sensitivity analyses regarding the applicability of using single-degree-of-freedom analysis were presented and discussed.