Yield optimization and surface image-based strength prediction of beech

Wood is a strongly anisotropic and heterogeneous material with natural defects that are affecting the uniform scatter of the fiber patterns. European beech (Fagus sylvatica) is the species, used for this study. The logs of these species have generally complicated shapes with frequent curvatures, which is in contrast to most of the softwood species with relatively straight log shapes. From structural point of view, these species have fewer knots and natural features, but stronger fiber deviations compared to different softwood species that have complicated knot configurations. This study consists of two parts: 1) log reconstruction and optimization of the cutting pattern, and 2) board reconstruction and strength prediction. Due to the complex structural pattern of hardwoods, the visual grading method is a relatively weak strength predictor for these species. The aim of this study is to develop a numerical method based on the finite element (FE)-analysis to provide a better prediction for the tensile strength of the boards. The analysis covers the scatter of 200 beech boards. By resembling the tensile test setup numerically, the stress concentration factors (SCFs) are calculated, considering the average and maximum stresses around the imperfections. SCFs in combination with the longitudinal stress wave velocity are the numerical parameters, used in the nonlinear regression model for tensile strength prediction. The nonlinear model is checked for different combinations of the numerical parameters to estimate and visualize the potential of the virtual predictions. Performance of the novel criteria is compared to the typical grading criteria (knottiness and the dynamic MoE (MoEdyn)), and is shown that the coefficient of determination is higher, when using the virtual methods for tensile strength predictions