| dc.description.abstract | The assessment of structural reliability for offshore wind turbines (OWTs) is challenging. The environmental conditions and loads on a wind turbine have large variability. Wind and wave act simultaneously on the structures, giving rise to combined effects that are difficult to assess. Often a time-domain simulation of the load components is required to assess the loads and the combination of loads, which is rather time-consuming. Hence, the minimum of simulations that still produces sufficiently accurate results is beneficial.
The objective of this thesis is to test different methods for estimating the fatigue reliability of a wind turbine structure and particularly seeking methods where a minimum of such load-response simulations is required while maintaining a high level of accuracy. Achieving this objective would require a simplified model of the wind turbine and its substructure, in this case, a jacket structure. This simplified model would allow for implementing an effective way of calculating the loads affecting the system and the response of the wind turbine structure. In this thesis, the fatigue of the substructure of the wind turbine due to wind loading is studied based on pre-performed damage simulations.
Gaussian processes are a relatively new approach to obtaining structural reliability, is presented using a statistical regression model to estimate fatigue damage. Predicting the total wind-induced fatigue damage in the structure is the aim of this research.
The Gaussian process method is compared with other methods, such as Monte Carlo Integration and Bins Method, which requires hundreds if not thousands of load-response simulations, which is time-consuming and costly. The results obtained show that this adopted approach is applicable to evaluating structural reliability analysis using a small number of training datasets which contributes in saving time and cost. | |
