The increasing structural and economic complexity of offshore wind turbine support structures has amplified the need for reliability assessments that can adequately represent the uncertainties associated with long-term deterioration mechanisms, such as fatigue and corrosion. Conventional structural reliability models typically rely on simplifying assumptions, including statistical independence among components, which may underestimate or overestimate system-level risks. To reduce epistemic uncertainties and integrate data from multiple sources—including inspections and structural health monitoring (SHM)—Bayesian updating with structural reliability methods (BUS) has been widely investigated. However, existing BUS frameworks, especially those based on dynamic Bayesian networks (DBNs), suffer from exponential computational growth with the number of components, limiting their applicability to realistic structural systems. This thesis introduces a novel multi-level BUS framework (mBUS) that enables tractable Bayesian updating of deteriorating structural systems while preserving physical interpretability and avoiding the need for surrogate modeling or correlation optimization. The proposed approach decomposes the system into component-level deterioration models while accounting for statistical dependencies via shared physical parameters, enabling efficient posterior inference even in high-dimensional settings. The formulation is validated against DBN models through a series of randomized scenarios and an illustrative example, demonstrating comparable accuracy with significantly reduced computational burden. Two application cases illustrate the practical value of the method. The first explores inspection and maintenance planning for a monopile substructure using heuristic policy search, revealing substantial cost reductions when system-level effects are incorporated. The second focuses on end-of-life decision-making for a monopile foundation, showing how failure probability and remaining useful life (RUL) estimates are sensitive to system-level correlations. Overall, the mBUS framework provides a robust and scalable basis for probabilistic structural assessment, offering a realistic alternative for engineering decision-making under uncertainty in offshore wind applications.

MAXWind – MAintenance, Inspection and EXploitation Optimization of Offshore Wind Farms subjected to Corrosion-Fatigue