This paper presents experimental, theoretical and numerical studies of wood-CFRP beams bonded with polyurethane (PUR) adhesive. The analyses include two types of CFRP (carbon fibre-reinforced polymer) strengthening configurations and pure glue laminated timber beams as a reference. Through detailed analyses of a double-lap connection on blocks with and without CFRP strips, the authors state that neglecting the cohesive stiffness of adhesive layers may lead to an overestimation of an overall beam’s stiffness. This is significant with wood–CFRP connections, which showed values two times lower than with wood–wood connections. Theoretical modelling of the equivalent area used in a theory of composites provided much stiffer behaviour of the beams than in laboratory experiments. It proves that a PUR glue eliminates the possibility of using simple models that assume a perfect connection between bonded parts. These conclusions led the authors to use the finite element method (FEM) to take into account the cohesive stiffness. The FEM, based on the properties obtained from a double-lap joint analysis, allowed for the precise prediction of the elastic stiffness of the beams.

The article presents research on modelling fracture in softwood bent elements. This kind of timber is the one most exploited for construction. Authors present a brief review on the subject with emphasis on three basic attempts: Linear Elastic Fracture Mechanics (LEFM), Continuum Damage Mechanics (CDM) and Hill’s Function (HF). Proposed 3D solution bases on Hill’s Function applied in the ABAQUS FEM code. The new idea includes isolating theoretical compression and tension zones in a model. Then, it is possible to distinguish between compressive and tensile strength and predict a real behaviour of bent elements. Introducing general dependencies between material properties leads to the need of determining only longitudinal elastic modulus (EL) and modulus of rupture (MOR). It is practical because these parameters are the main reported in a scientific and technical literature. Authors describe all of the assumptions in details. The experimental tests and Digital Image Correlation method (DIC) validate the FEM model.

This paper describes a method of predicting the direction of crack propagation implemented by user subroutines in the Simulia-Abaqus FEA system with the use of the extended finite element method (X-FEM). This method is based on displacements and stresses according to Westergaard’s solution of Griffith’s crack problem. During the calculations, in each crack increment, the algorithm reads the stresses and displacements in the model around the crack tip, calculates the criterion values at the read points, reduces them to a unit distance from the crack tip, fits a polynomial to these points, and finds the minimum of the function closest to the last propagation angle. The algorithm also decides when the crack grows, depending on a chosen criterion. Four criteria have been implemented to predict the direction of failure propagation: the maximum principal stress criterion, the Ottosen–Podgórski criterion, the new criterion described here based on the minimum component values of the displacement vector, and the maximum circumferential tensile stress (MTS). These criteria were verified in two tests: the three-point bending test of the notched beam and the anchor pull-out test. For these tests, the criterion built into Simulia Abaqus does not correctly define the crack path, which causes the crack propagation direction to “rotate” when simulating the fracture. The criteria developed here, in most cases, determine the crack path and the maximum force very well compared to real laboratory tests.

The paper presents numerical and experimental research on glulam delamination in a double lap connection with predominant shear stresses. Laboratory tests and wide literature survey enabled to determine timber and glue joint parameters. Cohesive zone theory, generally used for epoxy matrix and fiber reinforced composites, was adopted to modelling glue layer delamination in glulam elements. Numerical models were validated with laboratory tests.