A Probabilistic multi-objective optimization framework for offshore wind turbine jacket substructure under aero-hydrodynamic loading using surrogate models

The significant inherent uncertainties in the design and optimization of offshore wind turbines make it necessary to adopt a probabilistic approach to ensure structural reliability and performance. This study proposes a probabilistic multi-objective optimization approach for the offshore wind turbine (OWT) jacket based on dynamic responses predicted by surrogate models. The proposed surrogate models-based framework emphasizes incorporating probabilistic performance criteria not only as constraints but also directly within the objective function to achieve a comprehensive probabilistic optimization, which has not been used yet on the OWTs. Wind-wave-induced dynamic responses were evaluated through finite element analysis (FE) using Abaqus software, in which the structure was subjected to the modified intensifying constrained new wave (MICNW) and intensifying wind records. These responses, which include the maximum drift at the tower-top and jacket-top under various sea states, as well as the maximum stress under the ultimate limit state (ULS), were utilized to develop three artificial neural network (ANN) models as surrogate models. The proposed framework simultaneously minimizes two conflicting objectives, which are initial material cost and expected annual loss (EAL), while the failure probability due to stress as the probabilistic constraint was determined through reliability assessment based on importance sampling. The Pareto optimal solutions were obtained using the NSGA-II algorithm combined with surrogate models. In the post-processing step, further verification was also carried out to check the frequency, ULS stress, and fatigue capacity of optimized candidate solutions using modal analysis and time-domain simulations, respectively. Finally, the results show that the optimized OWT jacket substructure is improved in terms of initial material cost and vulnerability of the OWT system under various sea states compared to the initial design, while satisfying the reliability criteria related to ULS, together with frequency, stress criteria, and fatigue requirements. The proposed optimization framework can be further developed for application in the initial engineering design stages of OWT projects.