In the past, development of offshore wind energy was mainly focused on ice-free regions, but increasing demand has introduced the need for development in sub-Arctic regions, such as the Southern Baltic Sea. The expanding offshore wind development in this area not only presents new opportunities, but also new engineering challenges. In these regions, sea ice may drift against the foundation of offshore wind turbines, introducing dynamic ice loads which can significantly affect the design and safety of offshore wind turbines. Extensive research has been performed on the ice-structure interaction and the resulting ice-induced vibrations. However, this thesis studies the ice-structure interaction on a macroscale, considering how ice fields interact with offshore wind farms as a whole. The main motivation for this focus is that the presence of multiple structures instead of a single structure will affect the ice drift- and growth within an offshore wind farm. Currently, the designers of offshore structures often assume that the ice drift speed may be correlated to the surface wind speed at 10 m height by a constant factor (Hendrikse, 2024). However, the presence of both a single- and multiple turbines may slow down the ice. Previous research indicates that for low ice velocities, severe loading can occur in the form of intermittent crushing and multi-modal ice structure interaction (Hammer et al., 2023). The presence of multiple turbines may also promote ice growth during winter - leading to stationary ice - as they can function as anchor points for the ice to grow onto. If the wind picks up, the (re)-initiation of the stationary ice sheet may lead to simultaneous loading of all turbines at low relative velocities, resulting in severe ice-induced vibrations (Hammer et al., 2023). This thesis therefore aims to explore the influence of an offshore wind farm on the ice drift speed, leading to the consideration of two scenarios: (1) the drift of an ice floe field that originate from outside the offshore windfarm and (2) the drift of an ice floe that is frozen in around the turbines. This is formulated in the main research question: How does the presence of an offshore wind farm affect the ice drift speed for both an ice floe field that originates from outside the offshore wind farm and a grown-in condition in the Southern Baltic Sea? The first objective of this thesis was to establish if the ice drift speed can indeed be correlated to the surface wind speed by a constant factor in the presence of multiple structures. The second objective was to acquire the combinations of ice thickness and wind speeds for which the motion of a frozen in ice sheet could be (re)-initiated. This thesis applies image processing to the satellite images that were captured from the Baltic Sea during the Copernicus programme. The resulting ice floe size distribution is then coupled to a synthetic floe field generator that randomizes ice floe shapes, orientation and placement within predefined boundaries. This enables the generation of ice floe fields having different sea ice concentrations and maximum floe sizes. These ice floe fields are then converted to SIBIS to enable performing simulations. This thesis only considers ice drift in 2D in the horizontal plane of motion, which implies that bending, rafting and buckling of ice are not modelled, as this concerns ice that is forced up or down due to compressive stress. Rather, the ice-structure interaction is dominated by crushing and splitting. The ice drift is simulated on a timescale up to a few hours and a scaled-down version of a typical offshore wind farm was used due to limitations of computational resources. Whereas typical offshore wind farm layouts consist of 50 turbines, this thesis adopts layouts using an evenly spaced 3 x 3 and 4 x 4 grid, having an inner distance of 1 km. When the floes are initially positioned outside the offshore wind farm, the results shown that the correlation between the ice drift speed and the wind speed is lost during the ice-structure interaction. The lowest ice drift speeds are found to occur for the floes that are directly in contact with the turbines, experiencing an average velocity reduction of 41.6% as compared to the case without turbines present. These floe speeds are in the range where they can promote intermittent crushing and multi-modal icestructure interaction. Furthermore, the results show that ice floe velocities at the turbines are affected by the interaction among ice floes that are not in direct contact with the turbines. Due to the rotation of ice floes around structures, the velocities at a turbine may also be influenced by the presence of neighbouring turbines. These effects lead to a misalignment of the ice- and wind loading, highlighting the importance of macroscale modelling. Results for the frozen-in scenario indicate that (re)-initiation of the stationary ice sheet may occur for several combinations of wind speed and ice thickness that are typical for the Southern Baltic Sea. The critical wind speed across multiple offshore wind farm layouts - and corresponding ice sheet area - was found to depend on the critical wind drag force. This is the wind drag force for which the resistance of the turbines to the motion of ice is exceeded. For the case of an offshore wind farm modelled after the existing Baltic Eagle wind farm layout, it was found that all wind speeds above 26 m/s can lead to simultaneous loading of all turbines. For both scenarios, it was not validated how using a low number of turbines - as compared to a reallife offshore wind farm such as the Baltic Eagle wind farm - affects the results that are presented in this thesis. For the simulations in which the floes originate from outside the offshore wind farm, this thesis identifies several effects - such as the spatial extend of the speed reduction across the floe field - that can be attributed to the presence of very large floes within the ice floe field. Furthermore, it is expected that the shielding of secondary- and tertiary rows of turbines may be related to the obstruction of these very large floes. To further explore these phenomena, a high number of simulations should be performed - that include a high number of turbines - to address the stochastic behaviour of an ice floe field interacting with an offshore wind farm. For the grown in scenario, the sensitivity of the identified critical wind speeds to changes in the offshore wind farm layout and the geometry of the ice sheet has not been validated. Furthermore, this thesis assumes a water current equal to zero, as both the magnitude and the direction of the current in the Baltic Sea were found to be highly varying. However, an offset between the direction of water currents and wind is commonly observed under natural conditions. Hammer et al. (2023) showed that misalignment of wind- and ice loading direction can result in ice-induced vibrations developing for higher ice drift speeds as compared to an aligned scenario. This effect has thus not been incorporated in this thesis. Finally, the crack paths are only drawn in straight lines during splitting and crack propagation is not modelled. It is speculated that this can affect the macroscale ice drift.

Arctic and sub-arctic regions are more and more used to build offshore wind farms. There is a lack of model-scale tests where dynamic sturcutral properties were systematically varied to derive the influence of these properties on boundaries of the three crushing regimes. Those regimes are intermittent crushing, where both the ice load and displacement signal have a sawtooth pattern, frequency lock-in, with its characteristic sinusoidal displacement signal and continuous brittle crushing with smaller random ice loads and structural displacements compared to the two previously mentioned regimes. The first goal of this thesis is to test if an error in the dynamic structural properties (mass, stiffness and damping) has an effect on the transition velocities of the crushing regimes. Secondly, a parameter study has been done to derive trends in the transition velocities when changing dynamic structural properties. Ultimately, in literature proposed dimensionless numbers, which are particularly only taking linear elastic ice parameters into account, have been checked for rightness.
To be able to achieve the above mentioned goals of the thesis, a model-scale test campaign has been performed at the Aalto Ice Tank as part of the SHIVER test campagin. 36 unique single degree of freedom configurations have been tested to derive trends in ice-structure interaction when changing dynamic structural parameters. The most notable element of this test campaign was the usage of a real-time hybrid test setup, which allowed the testing of multiple sets of structural parameters while only one physical structure was used.
Those test signify that there is a relation between mass, stiffness, natural frequency and damping and transition velocities inbetween the three crushing regimes. When mass, stiffness and natural frequency increase, the boundaries of those regimes shift to lower velocities. An increase in damping caused the velocity of the boundary between frequency lock-in and continuous brittle crushing to go up, while the boundary between intermittent crushing and frequency lock-in was not influenced by the change in damping. Furthermore, from the tests it could be conlcuded that dimensionless numbers based on only linear elastic parameters are not correct and more parameters should be taken into account as well.
The thesis is a steping stone into more research into the dependency of ice-induced vibrations on dynamic structural properties with different initial conditions or structural properties. Furthermore in future test campaigns other dimensionless groups could be proposed and validated.