Study of fracture properties of wood using high-speed video imaging and neural networks

Abstract

In this study, the Duration of Load (DOL) for crack initiation and propagation, crack speed, and load carrying capacity were investigated for three Rates of Loading (ROL) and four sizes of notched wood beams using high-speed video imaging and neural networks. For the smallest ROL, there was a distinct volume effect on DOL to initiation which was almost inhibited at the largest ROL. The DOL for crack propagation for all volumes appeared to be random. The crack propagation was a wave phenomenon with positive and negative speeds that varied with the rate of loading. The study showed that the crack initiation load, peak load, and their respective gross stresses were independent of ROL but were nonlinealry correlated with volume and the smallest volume maintained the highest stress. The stresses followed the Weibull's weakest link theory. Artificial Neural Networks (ANN) revealed meaningful trends for the combined effect of physical and geometric variables on the loads and stresses. Fracture toughness was insensitive to ROL and realtively constant for the three larger volumes. However, the smallest size produced the largest fracture toughness, which was explained by a neural network model that showed that the width had the greatest influence on fracture toughness highlighting plane stress conditions. The study showed the usefulness of ANN for analyzing interaction among many variables affecting wood fracture behaviour and their potential to become reliable predictors of load carrying capacity including maximum load and stress and fracture toughness under the uncertain influence of these variables.