Advancing Probabilistic Risk Assessment of Offshore Wind Turbines on Monopiles

Offshore Wind Turbines (OWTs) are a unique type of engineered structure. Their design spans all engineering disciplines, ranging from structural engineering for the substructure and foundation to electrical or mechanical engineering for the generating equipment. Consequently, the different components of an OWT are commonly designed independently using codified standards. Within the OWT design process, financial cost plays an important role as a constraint on decision making, because of the competition between prospective wind farm operators and with other forms of electricity generation. However, the current, independent design process does not allow for a combined assessment of OWT system financial loss. Nor does it allow for quantification of the uncertainties (e.g., wind and wave loading, materials properties) that characterise an OWT’s operations and which may have a strong impact on decision making. This thesis proposes quantifying financial losses associated with an OWT exposed to stochastic wind and wave conditions using a probabilistic risk modelling framework, as a first step towards evaluating Offshore Wind Farm (OWF) resilience. The proposed modelling framework includes a number of novel elements, including the development of site-specific fragility functions (relationships between the likelihood of different levels of damage experienced by an OWT over a range of hazard intensities), which account for uncertainties in both structural capacity and demands. As a further element of novelty, fragility functions are implemented in a closed-form assessment of financial loss, based on a combinatorial system reliability approach, which considers both structural and non-structural components. Two important structural performance objectives (or limit states) are evaluated in this thesis: 1) the Ultimate Limit State (ULS) which assesses the collapse of an OWT due to extreme wind and wave conditions, such as those resulting from hurricanes; and 2) the Fatigue Limit State (FLS), which addresses the cumulative effects of operational loading, i.e., cracks growing over the life of the structure until they threaten its integrity. This latter limit state is assessed using a novel machine learning technique, Gaussian Process (GP) regression, to develop a computationally-efficient surrogate model that emulates the output from computationally-expensive time-domain structural analyses. The consequence of the OWT failing is evaluated by computing annualised financial losses for the full OWT system. This provides a metric which is easily communicable to project stakeholders, and can also be used to compare the relative importance of different components and design strategies. Illustrative applications at case-study sites are presented as a walk-through of the calculation steps in the proposed framework and its various components. The calculation of losses provides a foundation from which a more detailed assessment of OWT and OWF resilience could be developed.