Dynamic modelling of offshore monopile decommissioning

Abstract

The offshore wind sector in Europe has grown significantly
over the past 20 years. Following this surge in number of installations, the
first generation of turbines is reaching its end-of-life (EOL) phase and
thought needs to be put into end-of-life strategies. As the market for lifetime
extension is still in its infancy and re- powering is economically unappealing,
decommissioning is the most obvious solution. The tower, Rotor Nacelle Assembly
(RNA) and electrical cables will most likely be removed by reverse-installation
methods, as well as the Transition Piece (TP). Removal of the monopile
foundation however, is considerably more diffi- cult due to the large embedded
length of the pile in the soil. The common way of decommissioning such piles in
industry (mostly the Oil & Gas (O&G) industry) is to cut them at or
below the seabed and leave the remaining stumps in place. This research
investigates more sustainable methods that remove the foundation completely, as
this is more in line with the modern circular economy and regulations, i.e.,
OSPAR, are getting more stringent, focusing on complete removal of all offshore
structures in the future. To remove the entire monopile foundation using only a
blunt uplift force, tremendous amounts of force are required. It is thus
desired to lower this extraction force by some technique to reduce soil
resistance. This research investigates the possibility of reducing soil
resistance surrounding monopiles by utilizing pitch and heave vessel motions.
Such motions can induce an harmonic force on the monopile during the first mo- ments
of lifting. The hypothesis is that low frequency oscillations of about 0.1 Hz
can induce plastic strain accumulation of the soil, resulting in permanent
displacement of the pile. This hypothesis is tested by con- structing two
models, one describing vessel motion in MATLAB, and the other describing the
pile-soil inter- action under cyclic loading, which is modelled in the open
source software OpenSees. An extensive market study is performed to find a
representative maximum set of dimensions of the first generation of monopiles installed
offshore. Monopiles with a grouted connection to the transition piece are
considered as first gener- ation monopiles, and it is these monopiles that are
of interest to the study because they need to be removed the first. The vessel
considered in this research is the Pioneering Spirit and the crane its Jacket
Lift System (JLS), due its large lifting capacity. After delineating to the
monopile size and the vessel, assumptions are made regarding the extraction
process. A cranemaster is added to the system to avoid slack in the lines and a
connection tool between the crane wire cables and the monopile is selected,
assuming a rigid connection between the two. The results indicate that applying
a low frequency cyclic load to a monopile can result in the accumulation of plastic
strain of the soil and, consequently, to larger monopile displacements than in
monotonic load cases of the same magnitude. In the simulations performed in
this research, this effect, called cyclic degradation, is observed to be more
intense in sands than in clays. It is observed that plastic strain is largest
when large forces are applied to the model, hence it is desired to maximize the
combination of tension and amplitude of the applied cyclic force. The effect of
cyclic degradation however, is stronger if the forcing frequency is increased
to a value near the resonance frequency of the pile-soil system. This value
lies in the region of 4 to 8 Hz, depending on the force input characteristics
because of the non-linear behavior of soils. Achieving fre- quency
multiplication of the input force has been considered in both active and
passive ways. It is concluded that only active frequency multiplication may
deliver the needed cyclic degradation, since passive frequency multiplication
diminishes the amplitude of the cyclic force, driving up the resonance
frequency of the pile- soil system even further. The applicability of the
pile-soil model, and thus these conclusions, is only valid during the first
moments of lifting, when displacements are relatively small. The results
obtained in this thesis are the result of a relatively conservative soil
dynamics model and a non- conservative hydrodynamical model. The reliability of
the results could be further improved if the two models are combined in a
co-simulation, where at every time-step the pile displacement and vessel motion
are deter- mined. Additionally, expanding the OpenSees model so that accurate
representations of large pile displace- ments are included, will allow for
calculation of longer simulations, where the pile is completely removed from
the soil. The hydrodynamic model can be improved by using an entire sea-state
as input.