Abstract:
A commercially available, three-dimensional (3-D) finite-element (FE) analysis
program was used to model metal-plate-connected (MPC) joints in wooden trusses.
The model's features included consideration of material properties, teeth-to-grain-to-direction-
of-force orientation, and wood-to-wood contact. Non-linear spring elements,
placed on the wood surface between wood and plate, were used to represent wood-to-teeth
interaction. Each tooth was represented as a single point consisting of three nonlinear
spring elements, accounting for wood-to-teeth stiffness behavior in each of the
three major plate directions. The stiffness properties for the spring elements were
assigned based on tensile joint tests at various teeth-to-grain orientations. Once
incorporated in the model, the per-tooth stiffness need not be adjusted for different
loading conditions applied later to the joint model.
The load-displacement (L-D) results from the model and the experimental
results from tensile and bending tests of splice joints with different teeth-to-grain
orientations showed a good agreement. Simplified models where wood-to-teeth interaction was represented using a reduced number of spring elements showed a
potential for application in truss design.
Furthermore, the behavior of five different types of MPC wood joints from an
actual scissors truss was evaluated through testing, while simulating loads carried by
truss members in service. Strength, stiffness, and failure modes were observed. For
three out of five joints, the steel plate size governed the behavior of the entire joint,
causing a steel-tearing failure mode. Models developed to simulate the five joints
predicted axial L-D behavior of the joints relatively well, whereas the rotational
behavior was not evaluated due to insufficiently defined boundary conditions during
testing.
Finally, an FE analysis of the entire scissors truss was performed. The 3-D
truss model was compared to conventionally accepted, two-dimensional (2-D), beam-element-
based truss FE models with different joint stiffness assumptions: pinned, rigid,
semi-rigid, and fictitious element. The 3-D model predicted the overall experimental
L-D truss behavior most accurately. The difference in displacement predictions by the
2-D models indicated that even more substantial discrepancies might exist in their
predictions of truss forces and moments, which are the basis for truss design
considerations.
