Assessing the effect of knots in sawn timber utilizing X-ray CT, models for fibre orientation and methods for deep learning

Effective utilization of wood for building and construction requires knowledge both of clear wood properties and of the significance of natural defects such as knots, which have a major negative effect on the strength of sawn timber. Furthermore, such knowledge must be utilized to establish reliable methods for grading.

However, a reliable and thoroughly verified model of fiber direction around knots is not yet available.

Therefore, the proposed project aims to fill this gap by developing, on the basis of detailed and comprehensive experimental data, a model for fiber orientation around knots that represents both the regular and the stochastic nature of the governing geometrical properties of the tree, the branches/knots and the fibers.

Furthermore, the model will in the project be utilized in combination with both finite element modelling and deep learning to determine mechanical properties of sawn timber.Methods to be utilized thus include X-ray computed tomography (CT), laser scanning, numerical and statistical modeling and optimization, finite element modeling and methods for deep learning such as convolutional neural networks.

Samples of wood including knots will be scanned with X-Ray CT and laser, respectively, to reveal three-dimensional geometry of knots, growth surfaces and fiber orientation in 3D and in the planes of wood surfaces, respectively.