Surrogate modeling of the variability of metocean conditions in space and in time during hurricanes is a crucial task for risk analysis on offshore structures such as offshore wind turbines, which are deployed over a large area. This task is challenging because of the complex nature of the meteorology-metocean interaction in addition to the time-dependence and high-dimensionality of the output. In this paper, spatio-temporal characteristics of surrogate models, such as Deep Neural Networks, are analyzed based on an offshore multi-hazard database created by the authors. The focus of this paper is two-fold: first, the effectiveness of dimension reduction techniques for representing high-dimensional output distributed in space is investigated and, second, an overall approach to estimate spatio-temporal characteristics of hurricane hazards using Deep Neural Networks is presented. The popular dimension reduction technique, Principal Component Analysis, is shown to perform similarly compared to a simpler dimension reduction approach and to not perform as well as a surrogate model implemented without dimension reduction. Discussions are provided to explain why the performance of Principal Component Analysis is only mediocre in this implementation and why dimension reduction might not be necessary.

Fixed-bottom offshore wind turbines (OWTs) are typically located in shallow to intermediate water depth, where waves are likely to break. Support structure designs for such turbines must account for loads due to breaking waves, but predictions from breaking wave models often disagree with each other and with observed behavior. This variability indicates the need for a better understanding of each model’s strengths and limitations, especially for different ocean conditions. This work evaluates and improves the accuracy of common breaking wave criteria through comparison to Computational Fluid Dynamics (CFD) simulations of breaking waves. The simulated ocean conditions are representative of potential U.S. East Coast offshore wind energy development sites, but the discussion of model accuracy and limitations can be applied to any location with similar ocean conditions. The waves are simulated using CONVERGE, a commercial CFD software that uses a Volume of Fluid (VOF) approach and includes adaptive mesh refinement at the evolving air-water interface. First, the CFD model is validated against experimental data for shoaling and breaking wave surface elevations. Second, 2D simulations of breaking waves are compared to widely-used breaking wave limits (McCowan, Miche, and Goda) for different combinations of wave height, wavelength, water depth, and seafloor slope. Based on these comparisons, the accuracy and limitations of each breaking limit model are discussed. General usage guidelines are then recommended.

A mooring and anchoring concept for floating offshore wind turbines is introduced in which each anchor moors multiple floating platforms. Several possible geometries are identified and it is shown that the number of anchors for a wind farm can be reduced by factors of at least 3. Dynamic simulation of turbine dynamics for one of the candidate geometries and for two directions of wind and wave loading allows estimation of multiline anchor forces the preview the types of loads that a multiline anchor will need to resist. Preliminary findings indicate that the peak demand on the anchor may be reduced by as much as 30% but that anchors used in such a system will need to be able to resist multi-directional loading.

Approximately 75% of installed offshore wind turbines (OWTs) are supported by monopiles, a foundation whose design is dominated by lateral loading. Monopiles are typically designed using the p-y method which models soil-pile resistance using decoupled, nonlinear elastic Winkler springs. Because cyclic soil behavior is difficult to predict, the cyclic p-y method accounts for cyclic soil-pile interaction using a quasistatic analysis with cyclic p-y curves representing lower-bound soil resistance. This paper compares the Matlock (1970) and Dunnavant & O’Neill (1989) p-y curve methods, and the p-y degradation models from Rajashree & Sundaravadivelu (1996) and Dunnavant & O’Neill (1989) for a 6 m diameter monopile in stiff clay subjected to storm loading. Because the Matlock (1970) cyclic p-y curves are independent of the number of load cycles, the static p-y curves were used in conjunction with the Rajashree & Sundaravadivelu (1996) p-y degradation method in order to take number of cycles into account. All of the p-y methods were developed for small diameter piles, therefore it should be noted that the extrapolation of these methods for large diameter OWT monopiles may not be physically accurate; however, the Matlock (1970) curves are still the curves predominantly recommended in OWT design guidelines. The National Renewable Energy Laboratory wind turbine analysis program FAST was used to produce mudline design loads representative of extreme storm loading. These design loads were used as the load input to cyclic p-y analysis. Deformed pile shapes as a result of the design load are compared for each of the cyclic p-y methods as well as pile head displacement and rotation and degradation of soil-pile resistance with increasing number of cycles.