ABSTRACT

Suction buckets are a foundation type for offshore structures. They are installed by evacuating the water inside the bucket, which creates an underpressure and thus a vertical installation force. The loading on the cylindrical shell leads to a risk of structural buckling of the structure. Initial geometric imperfections and boundary conditions imposed by the constraining soil influence the buckling resistance significantly. Adequate modeling of the highly nonlinear soil behavior is crucial to capture the overall structural behavior. However, no recommendations on the soil model can be found in design codes and most approaches applied in literature were neither compared nor validated. Within this paper, modeling the soil as continuum with Mohr-Coulomb failure criterion is compared to different nonlinear soil spring approaches. Finite element buckling analyses with different soil conditions and embedment depths were conducted. The comparison of the different modeling approaches shows that a modification of soil spring properties is necessary to apply them for buckling analysis of suction buckets.

INTRODUCTION

With the aim to meet Net Zero carbon emissions, one of the major clean energy sources will be offshore wind energy (IEA 2019). Future wind farms will be further from the coastline in deeper waters and potentially with less favorable soil conditions. At the same time, the capacity of the turbines will increase, which leads to higher loads on the support structure. These design requirements could push monopile manufacturing to the limits. A feasible and material- and weight saving alternative are suction bucket jackets. These are lattice structures mounted on suction buckets. Suction buckets are cylindrical shells with large diameters but significantly shorter height than piles. This makes them suitable for soil conditions where only the upper part is sand or clay, and the lower soil layers contain rocks. Further advantages are that suction buckets have minimal noise emission during installation; they are easy to decommission and reduce material demand compared to pile foundations. A major challenge associated with suction buckets is the installation process. During installation, the lowest part of the suction bucket is embedded by the self-weight of the substructure. Then, the water is evacuated from within the bucket. This pressure differential creates a downward force, which drives the installation downwards and crates seepage, which facilitates the insertion. Simultaneously, the combined circumferential and meridional compression leads to a risk of shell buckling of the suction bucket. In addition to easily quantifiable influencing parameters such as nominal geometric dimensions and material properties, the buckling resistance is significantly influenced by geometric imperfections and boundary conditions, which entail considerably more uncertainty. The soil behaves highly nonlinear and thus adds to the complexity. Soil conditions depend on the site conditions and further, vary over the embedment depth. On the one side, it is difficult to determine soil parameters for the design precisely. On the other side, the consideration of the soil for numerical models used for the design is also challenging. Several approaches exist to consider the soil for buckling analysis. Pinna et al. (2001) used volume elements with a Tresca failure criterion to model clay. Bakmar et al. (2010) implemented a Mohr-Coulomb failure criterion to account for the behavior of sand. Madsen et al. (2013) modeled the soil as linear elastic volume elements, not accounting for any nonlinear behavior of the sand. Hanssen et al. (2013) compared two different approaches to generate nonlinear soil springs: the classical p-y-curves of the API (2007) for laterally loaded piles and soil springs which were calibrated on a sinusoidally displaced bar with an elastic-perfectly plastic soil model representing clay. These resulted in different nonlinear curves depending on the buckling mode and number of buckles. They concluded that both methods lead to conservative results, in the case of the API curves to an underprediction by 30%. However, the bucket considered in this work only have the size of suction buckets built nowadays. Gottschalk (2017) calibrated p-y-curves by simulating a lateral displacement of 100 mm and evaluating the soil resistance at the nodes with the highest soil reaction pressure. These were then applied for all nodes in radial direction. Knudsen et al. (2013), Østergaard, et al (2015) and Vethanayagam et al. (2017) also simulated the horizontal displacement of suction buckets to obtain p-y-curves. The latter resulted in generally applicable p-y-curve equations for suction buckets. The above-mentioned approaches are very different and the consequences for the application on buckling of suction buckets have not yet been quantified. Further, all soil spring approaches were calibrated considering lateral displacement and not buckling.

This content is only available via PDF.
You can access this article if you purchase or spend a download.