Kokeellinen tutkimus ja probabilistinen mallinnus sormijatkosten ja liimapuupalkkien mekaanisista ominaisuuksista

Sustainable development requires reducing the environmental footprint of construction, and engineered wood products, which are based on naturally grown material, offer a solution. Among these products, glued laminated timber (GLT) beams are extensively utilized in structural engineering owing to their adaptability and favorable strength-to-weight ratio. These beams are produced from timber boards that are joined end-to-end using finger joints (FJs). The mechanical performance of GLT beams largely depends on the properties of both the individual timber boards and the FJs. Due to variation in the mechanical properties of these components, the mechanical behavior of GLT beams can exhibit considerable variation. Consequently, developing and validating accurate predictive models is crucial for reliably estimating the mechanical properties of GLT beams and optimizing the efficient use of this material in construction. The main aim of this thesis is to contribute to the probabilistic modeling and validation of GLT beams by advancing prediction models for FJs, thereby enabling a more realistic representation in beam level simulations and enhancing the understanding of how local material variability and defects within the beams affect their mechanical behavior. Extensive experimental investigations were conducted on 227 FJs and 62 structural-scale GLT beams with well-known setups. For the FJ prediction models, Bayesian statistical methods were applied, explicitly accounting for censored data arising from failures outside the FJs. This approach provided more reliable estimates of FJ tensile and bending properties based on grading indicators of timber boards. At the beam level, digital image correlation was employed to investigate the impact of local defects on mechanical behavior, providing insight into the roles of knots and FJs within GLT beams. These findings were incorporated into the probabilistic numerical model, which was validated against the structural-scale GLT beam experiments. The model demonstrated high accuracy in predicting GLT beam bending stiffness (R2 ≈ 0.97) and reasonable accuracy in predicting bending strength (R2 ≈ 0.60). The validated model can support parametric studies and optimization of GLT beams, enabling more reliable designs and improved material efficiency. Another goal of this thesis was to investigate the potential influence of layup modification to improve the mechanical properties of GLT beams, experimentally and numerically.Both approaches indicated that layup modification can improve bending mechanical properties, highlighting an additional pathway for optimizing GLT beam design.