The world is currently in a severe man-made climate crisis, creating the need for using renewable energy sources such as offshore wind farms. Offshore wind turbines do not take up any valuable space on land and can use the greater wind resource at sea. The downside is, however, that they are exposed to a much harsher environment at sea, which could result in more failures and also lead to a smaller accessibility window for maintenance and repair. The structural components of offshore wind turbines are subjected to a corrosive environment and are exposed to cyclic loads that come in the form of wind and waves, which can cause significant corrosion and fatigue damage. Having accurate numerical models that can capture the physics of these damage processes could help reducing uncertainties on the structural performance of these components and therefore improve the position of offshore wind energy as a competitive renewable energy source. In this thesis, two such models are developed @en

This study presents calibration and validation of a cohesive extended finite element model for fatigue crack propagation in ductile materials. The approach relies on a separation between plasticity around the crack tip and fatigue crack growth at the crack tip such that the influence of plasticity on fatigue driving forces is predicted. This implies that characterization of crack growth requires effective Paris parameters. It is shown that the calibrated model can capture fatigue crack growth behaviour in ductile materials for in-phase and out-of-phase biaxial fatigue loading as well as in-phase biaxial loading with an overload.

@en

A numerical model for corrosion pit propagation under mechanical loading is presented. The level set method is used for corrosion front tracking and also enables the domain to be split into a solid and a pit domain. In the pit the diffusion of atoms originating from the dissolution process occurring at the pit front is simulated. The model is capable of automatically capturing lacy cover formation due to the inclusion of activation control, diffusion control, and passivation. In the solid static equilibrium is solved to obtain strains and stresses. A parameter, dependent on the signs of the plastic strain increment and the back stress, is introduced to define the influence of plasticity on the corrosion rate. The model is used to study pit growth under electrochemical and mechanical loading. Under activation control combined with an elastic material response, pits propagate faster under constant loading than under cyclic loading. When plastic deformation occurs, cyclic loading can significantly increase the pit growth rate. Increasing the cyclic load frequency results in faster propagation due to kinematic hardening. Under diffusion control, mechanical loading does not influence the pit growth rate, given that the salt layer leading to diffusion control remains intact.

@en

Structures are subjected to cyclic loads that can vary in direction and magnitude, causing constant amplitude mode I simulations to be too simplistic. This study presents a new approach for fatigue crack propagation in ductile materials that can capture mixed-mode loading and overloading. The extended finite element method is used to deal with arbitrary crack paths. Furthermore, adaptive meshing is applied to minimize computation time. A fracture process zone ahead of the physical crack tip is represented by means of cohesive tractions from which the energy release rate, and thus the stress intensity factor can be extracted for an elastic-plastic material. The approach is therefore compatible with the Paris equation, which is an empirical relation to compute the fatigue crack growth rate. Two different models to compute the cohesive tractions are compared. First, a cohesive zone model with a static cohesive law is used. The second model is based on the interfacial thick level set method in which tractions follow from a given damage profile. Both models show good agreement with a mode I analytical relation and a mixed-mode experiment. Furthermore, it is shown that the presented models can capture crack growth retardation as a result of an overload.@en