Presentations

Abstract

Design of Offshore Wind Turbines (OWT) implies aero-servo-hydro-elastic models in which many sources of uncertainties are involved: the wind and wave conditions, on the structure, and on the material resistance. All these uncertainties should be considered when designing a turbine. In international standards [1], this is generally achieved by means of a deterministic design and safety factors introducing significant conservatism in the design. This motivated the recent development of new methods for probabilistic design where the environmental conditions joint probability is estimated [2] and all the uncertainties are propagated for several limit states. A novel integrated workflow has been developed accordingly in HIPERWIND H2020 European project [3]. This presentation focuses on the Ultimate Limit State (ULS) considering rare events producing extreme loads on the wind turbine components. Reliable ULS design of OWT is challenging for computational feasibility, as the target failure probability is small [4], and because a complex combination of all uncertainties must be considered. For this purpose, we rely on a dedicated Sequential Sampling approach [6] where metamodels are built from a sequence of Design of Experiment which are enriched in environmental/model regions. This approach allows to estimate accurately the failure probability with few simulations runs. The final goal of this presentation is to illustrate how much reduction of unnecessary material can be achieved by considering uncertainties with a demonstration on case study from Teesside offshore wind farm operated by EDF in UK. The gain is obtained by mass reduction with reduction of diameter and thickness of the tower and monopile. We consider constraints on the natural frequencies of the structure, the shell buckling, and the failure probability of the tower basis moment. Lastly, the diameter/thickness ratio is constrained for manufacturability. This optimization step is prohibitively expensive as it requires a failure probability estimation for every design considered. For this case study, it was however verified that bending moments, throughout the structure, were preserved across a relatively wide range of diameter and thickness. Consequently, for a given set of environmental and model condition, the Von Mises stress involved in ULS bending of tower and monopile can be assumed to depend only on the thickness and diameter. This result is explained by static equilibrium consideration of the whole structure. Taking advantage of this simplification, we kept the same Design of Experiments to evaluate ULS change when reducing tower and monopile material. The presentation will give quantification of the obtained gains showing the interest to replace deterministic design by probabilistic ones. [1] International Electrotechnical Committee (2019) IEC International Standard 61400-3: Wind energy generation systems. [2] Vanem, E, et al. 2023. A joint probability distribution for multivariate wind-wave conditions and discussions on uncertainties (OMAE 2023) [3] Dimitrov, N. et al. 2024. End-to-end wind turbine design under uncertainties: a practical example. (TORQUE 2024) [4] International Electrotechnical Committee (2023) IEC TS 61400-9 ED1: Wind energy generation systems. [5] Gramstad, O. et al. 2020. Sequential sampling method using Gaussian process regression for estimating extreme structural response. Marine Structures. 72. 102780.