In this paper, some of the results are presented from a series of quasi-static tests on CLT wall panels conducted at FPInnovation-Forintek in Vancouver, BC. CLT wall panels with various configurations and connection details were tested. Wall configurations included single panel walls with three different aspect ratios, multi-panel walls with step joints and different types of screws to connect them, as well as two-storey wall assemblies. Connections for securing the walls to the foundation included: off-the-shelf steel brackets with annular ring nails, spiral nails, and screws; combination of steel brackets and hold-downs; diagonally placed long screws; and custom made brackets with timber rivets. Results showed that CLT walls can have adequate seismic performance when nails or screws are used with the steel brackets. Use of hold-downs with nails on each end of the wall improves its seismic performance. Use of diagonally placed long screws to connect the CLT walls to the floor below is not recommended in high seismic zones due to less ductile wall behaviour. Use of step joints in longer walls can be an effective solution not only to reduce the wall stiffness and thus reduce the seismic input load, but also to improve the wall deformation capabilities. Timber rivets in smaller groups with custom made brackets were found to be effective connectors for CLT wall panels. Further research in this field is needed to further clarify the use of timber rivets in CLT.
The existing models for the prediction of the splitting failure of dowel-type connections loaded perpendicular to grain are determined generally based on crack growth of the entire member cross-section. These models can be appropriate for stocky or rigid fasteners installed through the full thickness of the wood member. However, for slender dowel-type fasteners such as timber rivets, particularly when the penetration depth of the fastener does not cover the whole member thickness, the crack formation is different. Observations from current tests in thick members show that the crack growth across the grain occurs to a depth corresponding to the effective embedment depth of the fastener and propagates along the grain until it reaches its unstable condition. The design method presented in this paper to predict the connection splitting capacity takes into account the observed two possible failure modes of wood: either partial or full width splitting. In the proposed method, the effect of geometry parameters such as connection width and length, fastener penetration depth, loaded and unloaded edge distances, end distance, and member thickness as observed by others are considered. Results of the tests undertaken with laminated veneer lumber (LVL) and glulam manufactured from New Zealand Radiata Pine (RP) and data available from literature confirm the validity of this new method and show that the proposed design approach can be used advantageously in comparison to other existing models for timber rivet connections under transverse loading.