The market for offshore wind turbines has been growing rapidly in the last decade, with over 1.2 GW of new effect installed in 2012 worldwide. There are many advantages to installing wind turbines offshore as opposed to onshore, however the total cost of offshore turbines per installed effect is still larger than onshore. Still the advantages regarding wind climate, visual impact, planning issues and noise are so important, that offshore development is being done on a large scale in especially Europe. Popularly said, most people like wind turbines if they obey the not-in-my-backyard-concept.
Due to the political demand for offshore wind farm development, great effort is put into trying to reduce the overall cost of the wind production in the whole lifetime. This has resulted in many suggestions to constructions, installation, operation, maintenance and decommissioning that can reduce the total expenses. One of the largest overall expenses is the foundation of the wind turbines, since the offshore environment causes a complex combination of loads from wind, wave, ice and operation. The bucket foundation is a new foundation concept developed during the last decade at Aalborg University that is meant to greatly reduce the cost of the foundation during the whole lifetime. The bucket foundation is a cylindrical steel caisson, open in the bottom and closed at the top with a skirt length to diameter ratio between 0.5 and 1.0. The foundation is installed by applying suction inside the caisson, and can thus be installed using no heavy equipment. When installed, the foundation is a hybrid between a monopile and a gravitational foundation, utilizing both the weight of the soil in the caisson and the soil pressure on the outside of the skirt. As the suction bucket concept is new, great effort is put into validation of the design process, with the goal of getting it approved by independent risk evaluation and classification organizations such as Det Norske Veritas. This master thesis is a part of the research work done involving the bucket foundation, and is in some parts a further development of work done previously at Aalborg University. The overall theme of this project is thus the bucket foundation concept. An analytical tool used to determine the ultimate capacity of a cylindrical pile foundation is the p-y curves, which were developed using experiments done on very slender cylindrical piles with a length to diameter ratio of 34.4. These p-y curves have been used for monopile foundations as well, with a length to diameter ratio around 5, however this application is problematic, due to the large difference in slenderness, causing the behaviour of the foundation to be inconsistent with the original p-y experiments. The inconsistency is even greater with the bucket foundation, and can therefore not be applied. The first part of this thesis uses 18 numerical finite element models of the bucket foundation in saturated sand in the drained condition with varying soil strength and geometry to calculate the soil response for a horizontal displacement needed to develop a new p-y formulation, which is valid for the bucket foundation. From the 18 models it has been possible to calibrate a generalized mathematical model, which allows for easy calculation of the soil response for a drained sand with a given angle of friction and geometry under a given horizontal displacement. The mathematical model is a first draft, and further work should be done to better validate and calibrate the model. In the research work in geotechnics, experimental model testing is often used. To validate the experimental results it is often attempted to recreate the test results via numerical modeling. With scaled models the stress level in the soil is however scaled accordingly as well. This is problematic, since the behaviour of the soil is not linear at small stress levels as predicted by the Mohr-Coulomb failure criterion. This causes the failure load obtained in a scaled experiment to be larger than predicted by the numerical modeling. This is naturally also a big challenge in relation to the research done with the bucket foundation. The solution to this problem could be to use a material model that takes the stress-dependent behaviour of the soil into account. This is achieved by implementing a stress-dependent strength material model in PLAXIS 3D, which is a widely used commercial finite element code. After calibrating the material model to Aalborg University Sand No. 1, the material model exhibited correct behaviour and is able to predict results from triaxial at low confining pressures. When comparing the results from the material model to actual bucket foundation test results, it is obvious that the finite element model needs more work, although compared to the widely used Mohr-Coulomb model, the material model predicts a larger failure moment, just as expected.
One of the disputes of the bucket foundation design is the behaviour of the foundation when subjected to impulsive loads. These loads can occur offshore under a number of scenarios such as breaking waves, freak waves or emergency stops of the turbine. The approach today is to find the ultimate capacity in drained and undrained condition and then take the lowest value. For a dense sand, often encountered offshore, the drained capacity will in most cases be lower than the undrained. Previous research at Aalborg University has shown that the approach is very conservative, when dealing with impulsive loads, as the behaviour of the foundation for a high loading rate is close to fully undrained. As a forced displacement of a high velocity is applied, a significant pore pressure build up takes place inside and around the caisson, which greatly increases the capacity of the foundation. The previous tests investigating the behaviour with different loading rates have been executed with equipment capable of applying a forced displacement at 10 mm/s with a range of 40 mm. Since then new equipment has been installed, making it possible to displace the bucket foundation with velocities up to 500 mm/s over a range of 500 mm. The third part of this thesis involves the first four succesful tests with this new equipment. Great work has gone into installing and setting up the new equipment, which has lead to a lot of good experience to be used in the further research. The results from the four tests with 150 mm of forced displacement at rates from 0.1 mm/s to 100 mm/s showed the same trend as found in previous test. A strength increase from the slowest to the fastest test of more than 20 times was observed, while the build up of pore pressure reached values close to the cavitation limit of the setup at -290 kPa. The pore pressure distribution, which is a picture of the failure mechanism, showed negative pressure development at all measuring points, with the largest change inside the caisson.