FE-Modelling of a Joint for Cross-Laminated Timber

Sammanfattning: Woodbe Engineering AB is a freshly started company that has developed a new type of joint for cross-laminated timber (CLT). The joint does not include any metallic fasteners, which improves sustainability, the ergonomics for the workers and time efficiency. The joint is designed to connect floor and wall elements in multi-storey buildings, by milling a dovetail in the floor element, and a fitting track in the wall element using a CNC machine. Before the product can be used on the market, it needs to be verified. This verification can either be done using physical tests, calculations, or a combination of both. The company has performed experimental small-scale tests, where the load-bearing capacity was tested. Later this year, large scale tests are to be performed. The purpose of this work is to develop a simulation model that can predict the results of the physical test. A simulation model that yields accurate results can be a good substitution for physical testing, due to a lower cost, better time efficiency, and parameters that can easily be changed. CLT is made up of several layers of wooden plates with different directions. The wood itself is quite complex to model. It has different properties in different directions, both ductile and brittle fracture modes and a large scatter of material properties. To capture this behaviour, a material model which incorporates orthotropic elasticity with linear fracture mechanics has been used. The behaviour of the material model has been evaluated with tests in both tension and compression in different directions. The accuracy of the material model was investigated by a simulation of the small-scale tests where the load-bearing capacity and the mode of fracture was investigated. A simulation of the large-scale experiment has also been conducted, where predictions of the load-bearing capacity and the first mode of failure was investigated. Also, a calculation script has been developed, which calculates the shear stress in the dovetail.  The results of the simulations clearly show the capability of the material model. Load-displacement graphs show ductile and brittle behaviour in compression and tension respectively. The strength is the highest along the fibres of the wood, with a fast decrease as the angle is increased. The simulation of the small-scale tests showed the initiation of rolling shear damage in the bottom transverse layer of the dovetail at a load level of 87 kN. The load continued to rise until a maximum load of 112 kN, while the damaged region grew upwards into the next layer. As compared to the physical tests, the mean maximum capacity of the joint was 125 kN, where rolling shear cracks could be found in the upper transverse layer in all tested specimens. Some of the tested specimens showed damage initiation at a load level of 84 kN. For the larger experiment, the same mode of damage was initiated at a load level of 161 kN which continued to rise until a maximum load level of 165 kN. The calculated values of the shear stress showed a critical shear force of 26 kN per dovetail. This value is 60 and 63 % of the simulated critical shear forces. The results of the simulation are in good agreement with the reference experiment in terms of damage initiation and maximum load. However, a large scatter of material properties, approximations of material orientations and interactions between individual layers results in a low level of predictability in terms of damage evolution and ductility in the material.