Tensile Strength Prediction of Softwood Glulam Lamellas Using Virtual Vibration Technique

The strength of wooden boards is related to its natural defects and the ability of a stress wave to propagate around them. Anisotropy, heterogeneity, and the strong moisture dependency of wood make it difficult to predict its strength. Covering the ordinary quality range of wood, a numerical simulation model was developed for strength prediction of timber, and 250 boards were numerically simulated. In this study, by virtually reconstructing a three-dimensional (3D) geometrical model of each case based on the surface information of the knots, and by predicting the fiber patterns, the material properties of wood were virtually predicted. Thus, the strength predictions were done solely based on the surface information of the knots. Strength variation in a knot-free board is only dependent on the variation of the actual density of the board. As the actual density of wood may not be available under different environmental conditions for measuring the dynamic modulus of elasticity (MoEdyn), the average density of each set was used as one of the only input parameters for simulations. The resonance kind stress wave propagation and its return were calculated in the reconstructed boards. Numerical results of the finite-element (FE) stress wave analysis were used in a linear regression analysis for prediction of the tensile strength based on the information of the calculated time of the stress wave. The predicted results were benchmarked against the measured values in the laboratory. Performing a multiple regression analysis, the virtual results provided much higher strength predictions than the geometrical parameters available from scanners. However, strong knot interactions in lower-quality boards also affect the strength predictions. This study provides a comprehensive system (starting from the geometrical reconstruction of the boards to the virtual analysis for calculation of the virtual MoEdyn) for the strength prediction of wood that is based solely on the surface images. The developed model makes it possible to predict the tensile strength of timber with relatively high accuracy, which is approximately at the same level as current grading machines.