ROS-3D

Objectives:

This project focuses on two areas: (i) the development of effective three-dimensional deterministic models for offshore wind turbine foundations subjected to axial and lateral loads, and (ii) the development of probabilistic models to calculate the probability of failure of offshore foundations, taking into account the three-dimensional spatial variability of soil properties.

Methodologies adopted:

– For the development of mechanical models, a large-diameter monopile offshore foundation was modeled using Abaqus finite element software. Two types of modeling were considered in the mechanical model: a “wished in place” model and a more sophisticated model that takes into account the foundation installation process. The latter model uses the coupled Eulerian-Lagrangian approach implemented in the Abaqus software.

•  For probabilistic analyses, three effective probabilistic methods based on the meta-model Kriging technique were developed in this work: GSAS, AK-MCSm, and AK-MCSd. The GSAS method combines Kriging with global sensitivity analysis, with the aim of providing better meta-model enrichment compared to the conventional Kriging approach. The AK-MCSm approach uses multi-point enrichment, which is very useful when computer facilities capable of parallel computing are available. Finally, the AK-MCSd approach uses the dependency between Krigeage predictions and reduces computation time compared to the conventional Krigeage approach.

The results from the various simulations are as follows:

• For the determination of the ultimate vertical bearing capacity, the results obtained using the numerical approach in Abaqus and those obtained using the API approach are consistent. The contribution of the strength of the monopile shaft and base to the total strength of the monopile has been determined. The results showed a major contribution of 60-80% for the shaft and a contribution of 20-40% for the base. In addition, it was noted that the shaft reaches failure first. It was shown that the soil inside the monopile remains integral with the monopile when the latter is loaded. The bearing capacity of the monopile is therefore calculated as the sum of the resistance of the shaft and that of the base of the monopile (soil + ring). This is in line with API recommendations.

•  The results for the large-diameter monopile subjected to lateral loading showed that the monopile undergoes rigid rotational movement around a point located around the lower third of the embedded length of the monopile. The results from modeling the monopile installation process in sand showed that the plugging phenomenon increases with decreasing monopile diameter, increasing sand density, and increasing installation force. For large-diameter monopiles, the plugging phenomenon was found to be unlikely. In addition, the results showed that the stress state around the monopile is significantly altered by the installation of the monopile. This change could have a significant effect on the response of the monopile.

• With regard to the probabilistic approaches developed, these methods have been shown to be highly efficient in terms of computation time compared to the classic Monte Carlo probabilistic approach and classic probabilistic methods based on kriging. The AK-MCSm method has proven to be the most efficient in terms of computing time. However, it requires the availability of computer facilities capable of parallel computing. AK-MCSd has proven to be more relevant than the GSAS approach because it is based on a more efficient enrichment methodology in terms of computing time.With regard to the probabilistic approaches developed, these methods have been shown to be highly efficient in terms of computation time compared to the classic Monte Carlo probabilistic approach and classic probabilistic methods based on kriging. The AK-MCSm method has proven to be the most efficient in terms of computing time. However, it requires the availability of computer facilities capable of parallel computing. AK-MCSd has proven to be more relevant than the GSAS approach because it is based on a more efficient enrichment methodology in terms of computing time. Furthermore, it has been shown that the effect of 1-D spatial variability in the vertical direction has a significant influence on the probability of failure as long as the vertical autocorrelation distance is less than the length of the monopile. In addition, it has been shown that the effect of horizontal spatial variability appears as soon as the horizontal autocorrelation distance becomes less than 2.5 D, where D is the diameter of the monopile. In the case of three-dimensional soil variability, the use of a conventional random field discretization method (e.g., EOLE) leads to a memory problem. To solve this problem, the Turning Band Method (TBM) has been proposed. This method has proven to be very effective for discretizing a three-dimensional random field with very short autocorrelation distances (i.e., in the case of highly heterogeneous soil). However, this method has a drawback related to the large number of random variables it requires to represent a random field. This hinders its combination with a Kriging-based approach, which can only handle a limited number of random variables.

Figure: Effect of spatial variability on the probability of failure

PRESENTATION OF THE PROJECT AT THE WEAMEC SEMINAR “GEOTECHNICAL & GEOPHYSICAL FOR MRE APPLICATIONS”

https://youtu.be/5oNVX-SvAT0