Offshore wind turbines, given their structural properties, are expected to experience severe ice-induced vibrations. However, full-scale events neither have been observed nor published yet and thus predictions of existing numerical models could not yet been validated. Model-scale experiments, aiming to investigate ice-induced vibrations of offshore wind turbines, have been inconclusive as structures with low natural frequencies would exceed the capacity of test facilities (i.e., size and weight limits), while geometrical scaling introduced scaling effects (e.g., buckling) compromising the validity of conducted experiments.

In the absence of full-scale testing capabilities, the main goal of this work was thus to demonstrate how offshore wind turbines behave under dynamic ice loads in smallscale experiments. In total four research questions (RQ) have been formulated and are addressed in this thesis, collectively serving to achieve the primary objective of this thesis.

RQ1: How can ice-induced vibrations of vertically sided offshore structures be scaled?

RQ2: How can offshore structures with low and multiple eigenfrequencies be tested in
ice tank experiments?

RQ3: What types of ice-induced vibrations can an offshore wind turbine experience?

RQ4: What is the effect of wind-ice misalignment on the development of ice-induced
vibrations of offshore wind turbines?@en

The study investigated the use of a Hardware-in-the-Loop (HiL) technique applied in model ice experiments to enable the analysis of offshore structures with low natural frequencies under dynamic ice loading. Traditional approaches were limited by facility capacities and ineffective downscaling of the geometry of the offshore structures. The goal of the present study was to overcome these challenges and to enhance the understanding and explore the applicability of a hybrid testing technique in model ice experiments. To achieve the objective, 204 Hardware-in-the-Loop simulations in model Ice (HiLI) were analyzed. Results showed robust behavior and good performance of the HiLI due to minimal variation in measured delay, normalized root mean square error, and peak tracking error and low magnitudes of such parameters despite alterations in factors such as the choice of the numerical structural model, physical prototype, measurement system, and ice type. Notably, the performance of the HiLI was affected when testing with warm model ice or scaling for harsh ice conditions, attributed to a reduced signal-to-noise ratio and instability of the system, respectively. Experimental identification of the critical delay, along with the application of an analytical stability criterion, revealed that the instability observed, was likely induced by reducing the structural stiffness of the numerical structural model to fulfil the scaling requirements when testing for harsh ice conditions. Additionally, the study showed improved HiLI performance when the physical prototype was in contact with the model ice. This observation was further analyzed and is assumed to be caused by the coupling between the ice and physical prototype, causing a coupled and thus increased eigenfrequency of the physical prototype-ice system.

@en

A modeling approach to simulate ice-induced vibrations of vertically sided offshore structures in ice tank experiments is presented. The technique combines replica modeling with the preservation of kinematics during ice-structure interaction. The technique was chosen based on the theoretical understanding that ice-induced vibrations are caused by an energy exchange between the structure and the ice. The mechanism is controlled by primarily four aspects: the kinematics during ice-structure interaction, the degree to which the ice can resist higher loading at low velocities prior to failure (velocity effect), the existence of a transition speed from ductile-to-brittle failure, and the mean ice load level. A model ice type which resulted in a velocity effect and provided a transition speed comparable to that of sea ice was developed and used during ice tank experiments. A scaling factor, derived from the comparison between the mean brittle crushing ice load of the full-scale event and the in-situ measured mean brittle crushing model ice load, was applied to scale structure properties of a numerical model. This model was implemented during real-time hybrid simulations in model ice to preserve kinematics during the ice-structure interaction. To verify the proposed scaling approach, rigid indenter experiments covering velocities from 0.1 mm s⁻¹ to 500 mm s⁻¹ and dynamic ice-induced vibration experiments of structures with varying aspect ratios (8 and 12) and shapes (cylindrical and rectangular) were conducted. Neither the aspect ratio nor shape appeared to influence the development of ice-induced vibrations significantly. The approach was qualitatively validated by reproducing full-scale ice-induced vibrations as experienced by the Molikpaq platform and Norströmsgrund lighthouse.

@en

A field campaign in the Vallunden lagoon in the Van Mijenfjorden on Spitsbergen was conducted to gather data on sea ice restoration by artificial flooding. Sea ice thickening was initiated by pumping sea water from below the first-year sea ice onto the surface without removing the covering snow layer. Part of the data was collected by four thermistor strings, two radiation sensors, and one anemometer. All measurement systems were left in the field until recovery of the floating systems in summer. Data provided by the measurement devices were received remotely to gather data before, during, and after the flooding phase (including the melting for as long as the sensors were sending data). Furthermore, coring systems were used to extract 88 ice cores for analysis of temperature, density and bulk salinity profiles along the full length of the ice cores before, during and within four days after flooding. The data set can be used to investigate physical processes involved in the ice growth before, during and after flooding. The data can be used to understand the development, growth and melting of snow ice. The radiation data can be used to analyze the (reflected) radiation of the initial, flooded and melting ice. Data gathered during the melting can be used to investigate the melting of thickened sea ice with different initial conditions prior to the onset of melting. Data on bulk salinity can be used to investigate short-term salt migration. Combing the different insights, growth- and melting models of sea ice including snow and snow ice can be validated. The understanding of melt-water drainage events could be improved and flow models for simulation of artificial flooding of snow-covered first-year sea ice could be further developed using the data.

@en

Increased activity in planning offshore wind farms in the northern Baltic Sea has renewed interest in studying the effect of ice cones on ice failure mechanisms. In preparation for future experiments with steep ice cones, preliminary ice basin experiments were performed at the Aalto Ice and Wave Tank to investigate how model ice fails against a 3D-printed cylindrical and a conical structure representative of wind turbine foundations. The main motivation behind the two structures is to test ice loads on a monopile foundation and a monopile foundation fitted with an upward-bending ice cone. Each structure was tested at eight different velocities in three ice sheets with varying mechanical properties, including a newly developed, crushing-optimized model ice. The force exerted by the ice on these rigid structures was measured using a six-axis load cell. The results show that ice undergoes mixed-mode failure on the cone in the form of bending, crushing, and spalling, when tested in crushing-optimized ice. Based on the observations and results, it is recommended that model-scale experiments, focused on mixed-mode ice failure, use model ice with a representative compressive to flexural strength ratio, scaled flexural and compressive strength, and the ability to fail in brittle crushing. If these criteria cannot be met, it may be possible to combine test results from different ice sheets, each focused on one ice failure mechanism. Additionally, this study successfully used 3D-printed structures, which present a new and more accessible method of preparing scale models.
@en

The effect of misalignment between wind- and ice loading direction on the development of ice-induced vibrations of offshore wind turbines has been investigated experimentally. In the experiments a hybrid test setup was used to study the structural response to combined loading from physical model ice and numerically applied wind. The motivation for this study was the high uncertainty in the design of offshore wind turbine support structures in cold regions, caused by scarcity of full-scale and model-scale data on ice-structure interaction. Test results revealed that misaligned scenarios result in the development of sustained ice-induced vibrations in the ice load direction. The test results also showed that ice-induced vibrations can develop up to higher ice drift speeds for misaligned scenarios than for aligned scenarios. Both observations are considered to be related to low total damping in the ice drift direction for a misaligned scenario. Further comparison between a 90°-misaligned operational and an aligned idling scenario revealed that wind-induced structural displacements perpendicular to the ice drift direction do not cause the ice to fail. On the contrary, it was shown that the ice constrains the wind-induced motion for low relative velocities between ice and structure. For high relative velocities, wind-induced displacements approach those in open water as the ice fails in rapid succession at the sides of the structure during crushing. The analysis of a misaligned scenario with a smaller misalignment angle revealed that vibrations occur perpendicular to the ice drift direction and are characterized by relatively low amplitude and high frequency. The ice, being in contact with the structure, neither prevented those vibrations nor failed.

@en

For offshore wind farms which are planned in sub-arctic regions like the Baltic Sea and Bohai Bay, support structure design has to account for load effects from dynamic ice-structure interaction. There is relatively high uncertainty related to dynamic ice loads as little to no load- and response data of offshore wind turbines exposed to drifting ice exists. In the present study the potential for the development of ice-induced vibrations for an offshore wind turbine on monopile foundation is experimentally investigated. The experiments aimed to reproduce at scale the interaction of an idling and operational 14 MW turbine with ice representative of 50-year return period Southern Baltic Sea conditions. A real-time hybrid test setup was used to allow the incorporation of the specific modal properties of an offshore wind turbine at the ice action point, as well as virtual wind loading. The experiments showed that all known regimes of ice-induced vibrations develop depending on the magnitude of the ice drift speed. At low speed this is intermittent crushing and at intermediate speeds is ‘frequency lock-in’ in the second global bending mode of the turbine. For high ice speeds continuous brittle crushing was found. A new finding is the development of an interaction regime with a strongly amplified non-harmonic first-mode response of the structure, combined with higher modes after moments of global ice failure. The regime develops between speeds where intermittent crushing and frequency lock-in in the second global bending mode develop. The development of this regime can be related to the specific modal properties of the wind turbine, for which the second and third global bending mode can be easily excited at the ice action point. Preliminary numerical simulations with a phenomenological ice model coupled to a full wind turbine model show that intermittent crushing and the new regime result in the largest bending moments for a large part of the support structure. Frequency lock-in and continuous brittle crushing result in significantly smaller bending moments throughout the structure.

@en

The authors regret their errors in the production of the legend labels and marker colors in Fig. 15 on page 11. The correct legend labels and marker colors are provided in the figure below:[Formula presented] The authors would like to apologize for any inconvenience caused.

@en

Cyclic crushing experiments with a haversine velocity waveform were performed on passively confined, freshwater columnar ice specimens for a variety of velocities and frequencies. The aim of the experiments was to study the ice deformation and failure behavior in crushing when loaded at a predefined displacement pattern closely resembling the frequency lock-in regime of ice-induced vibrations. The focus of the experiments was on the development of load and ice deformation behavior at the grain and ice specimen scales during each cycle. To this end, the deformation and failure of the ice were observed with crossed-polarized light to highlight the microstructure in-situ during cyclic crushing. It was shown that there are dichotomous mechanical behaviors of the damaged and confined ice during a single crushing cycle: brittle at high velocity and non-brittle at low velocity. At low velocity, ice fracture was interrupted and stress relaxation occurred until the predefined velocity began increasing in the cycle. The stress relaxation in the load was accompanied by stress-optic effects in the ice. It was found that a load peak-velocity hysteresis developed in each crushing cycle: peak loads following the non-brittle behavior were temporarily higher than the peak loads of the brittle behavior. The temporary load peak enhancement tended to increase with increasing duration of stress relaxation, i.e. the peak enhancement tended to increase with decreasing velocity and frequency. Negligible peak enhancement and stress relaxation duration were observed for the highest frequency and mean velocity tested of 2 Hz and 10 mm s⁻¹, respectively. For tests with a minimum velocity of 1 mm s⁻¹, no stress relaxation was observed in the load measurement. Preliminary results from deviating from the haversine velocity waveform by increasing the minimum velocity showed that the stress relaxation duration decreases, but the non-brittle peak load does not decrease. It is speculated that ice anelastic ice behavior could account for the rapid stress relaxation at low velocity. It is unclear what causes the hysteresis, although it is speculated that dynamic strain aging might play a role. The change in ice behavior during the experiments demonstrates a mechanism which develops rapidly and might therefore incite the development of the frequency lock-in regime of ice-induced vibrations of vertically-sided structures.

@en

Drifting sea ice failing in crushing against vertically-sided offshore structures can cause ice-induced vibrations. Offshore structures are typically founded on slender structures to minimize the load effect of waves and currents. In combination with large top masses, those offshore structures often provide sufficient compliance for ice-induced vibrations to develop. Although modern offshore structures can be expected to experience ice-induced vibrations in higher structural modes, this phenomenon is rarely considered during experiments and numerical analysis of dynamic ice-structure interaction. Inspired by this challenge, we investigated experimentally how the sole change of mode shape amplitude relation between higher structural modes at the water level of a multi-degree-of-freedom structure influences the development of frequency lock-in. Experiments of four different multi-degree-of-freedom structures in cold model ice have been performed in Aalto Ice and Wave Tank. To allow full control over the eigensystem during testing, modal representations of structures were implemented in the numerical domain of a hybrid test setup. When changing the mode shape amplitude, the total structural stiffness at the ice action point and modal damping as a fraction of critical were kept constant between the four structures. We found that the structure experienced sustained frequency lock-in vibrations in a frequency corresponding to the mode shape amplitude of artificially high magnitude. When mode shape amplitudes of two eigenmodes were equalized, the structure experienced oscillations in the frequency of the mode with lower frequency or lower damping mainly. It was found that ice-induced vibrations of multi-degree-of-freedom structures are highly dependent on the relative velocity between the ice and structure and thus on the superposition of higher mode oscillations with lower mode oscillations.@en

A series of ice penetration tests with a rigid structure, with controlled oscillation, and with a single-degree-of-freedom structure were performed to investigate the peak load-velocity dependence for ethanol-doped model ice during a test campaign at the Aalto Ice and Wave Tank. For the rigid structure and controlled-oscillation tests, the ice drift speed ranged between 1 and 150 mm s⁻¹. In the controlled-oscillation tests, amplitudes of oscillation between 0.40 and 15.90 mm and frequencies of oscillation between 0.143 and 4 Hz were prescribed such that the relative velocity between ice and structure never became negative. A constant ice drift deceleration experiment with a single-degree-of-freedom structure was performed to investigate the development of frequency lock-in and intermittent crushing in the model ice and compare the results with the rigid structure and controlled-oscillation tests. It was found that the peak load-velocity dependence identified in the rigid structure tests was not always uniquely defined as identified in the controlled-oscillation tests because the loading history affected the peak load at ice failure. A rapid strengthening of the ice developed at low relative velocity and carried over to high relative velocity until the ice failure dissipated the strengthening effect. The strengthening effect, observed in the rigid structure and controlled-oscillation tests, was also observed during frequency lock-in and intermittent crushing in the single-degree-of-freedom structure test. The observations in the present study indicate that the so-called velocity and compliance effects in ice-structure interaction originate from the same strengthening effect. It then follows that peak loads on compliant structures cannot exceed peak loads on rigid structures in the same ice conditions, with the only difference being that the peak loads on compliant structures occur at apparently higher far-field ice drift speeds due to the change in relative velocity.

@en

Basin tests were performed at the Aalto Ice Tank to gather data on ice-structure action and interaction from ice failing against a vertically sided cylindrical pile. The tests were performed with a real-time hybrid test setup, which combined physical and numerical components to simulate a range of test structures in real-time. The dataset includes results from tests with offshore wind turbine structures, structural models representing a series of single- and multi-degree-of-freedom oscillators, and scaled dynamic models of the Norströmsgrund lighthouse and the Molikpaq caisson structure. In addition, forced vibration tests and rigid structure tests were performed. Ice loads and structural response were measured with accelerometers, displacement sensors, potentiometers, strain gauges and load cells and the ice-structure interaction process was filmed from three different camera angles. The resulting raw data have been categorized and stored as unfiltered time series. A total of 259 different tests are included in the dataset. The model ice formation procedure and the test temperature were aimed at creating model ice that mimics the material behavior of full-scale saline ice during crushing failure, with a specific focus on the transition from brittle to ductile behavior. The data can be used for validation of models for dynamic ice-structure interaction. The offshore wind turbine data can be used to study the effect of wind loading on the interaction with ice and the effect of the specific dynamic properties of wind turbine structures with monopile foundations on the ice-structure interaction process. The forced-oscillation data can be used to quantify the time and speed dependant aspects of ice loading. The Norströmsgrund lighthouse and the Molikpaq data can be used as a reference comparison to full-scale data on ice loads.

@en

Ice-induced vibrations of offshore wind turbines on monopile foundations were investigated experimentally at the Aalto Ice Tank. A real-time hybrid test setup was developed allowing to accurately simulate the motion of a wind turbine in interaction with ice, incorporating the multi-modal aspects of the interaction and the effect of simultaneous ice and wind loading. Different vibration patterns were observed where some could be described based on the common terminology of intermittent crushing or continuous brittle crushing. However, not all resulting vibrations could be described accordingly. A combination of several global bending modes interacting with the ice resulted in high global ice loads and structural response. Such response is likely typical for an offshore wind turbine, owing to the dynamic characteristics of the structure. The type of interaction observed during the tests would be most critical for design as the largest bending moments in critical cross-sections of the foundations occur for this regime. A classification of ice-induced vibrations is proposed which encompasses the experimental observations for offshore wind turbines on the basis of the periodicity in the structural response at the ice action point. @en

For the topic of predicting ice-induced vibrations of vertically sided offshore structures, the rate-dependent ductile-to-brittle transitional deformation and failure behavior of ice is critical but remains superficially understood. To investigate this knowledge gap, a test setup has been designed which allows for in-situ crossed-polarization imaging of passively confined ice thick sections subjected to compressive loading. The test setup is designed to recreate the scenario of a cross-section at the leading edge of an ice sheet which is laterally confined by surrounding ice and fails in crushing against a structure. The setup comprises a linear actuator which drives a flat plate into a confinement box containing the ice thick section, which is passively confined orthogonal to the plane of loading by thick fused silica glass plates. The ice is illuminated through the glass plates with crossed-polarized light, which highlights the microstructure of the ice. Freshwater ice of columnar grain structure is prepared in the ice laboratory at Delft University of Technology, and quantified in terms of its microstructure. The ice thick sections in the test setup are subjected to a range of deformation rates at different temperatures. While similar experiments have been performed, this setup provides novelty by accentuating the dynamic microstructural deformation in-situ with crossed-polarized light. Moreover, this microstructural deformation is observed for global deformation rates relevant for ice-induced vibrations of offshore structures. A description of the test setup is presented along with preliminary experimental results.@en

The concept of floating vegetation-based islands for the bioremediation of aquatic ecosystems is well known. Less so, their hydrodynamic capabilities regarding the damping performance, positional stability and water-structure interactions. To this end, physical model tests with fully organic, reed-based gabions were carried out in a large-scale facility in this study. The initial, reflected, and transmitted waves were recorded and analyzed regarding transmission and reflection coefficients. A motion tracking system was utilized to allow for an investigation regarding the motion of the artificial floating islands under waves. The results show that the artificial floating islands significantly dampen shorter waves with a wave period of T ≤ 2.25 s. The transmission of the incident waves is reduced by 50% for the smallest wave periods (T = 1.5 s). The incident waves are reflected between 20 and 50% for the same wave period. The incident wave energy is dissipated by up to 85% for the smallest wave height and period (H = 0.10 m, T = 1.5 s). The comparable performance regarding more traditional floating breakwaters is discussed as well as the width of the structure as the key parameter for the layout of artificial floating islands in rivers and still waters regarding the damping performance.

@en

In bodies of water where ice is not an annual occurrence, such as in the Southern Baltic Sea, the design of offshore wind turbines is complicated by the difficulty involved in estimating the relevant ice parameters (thickness, velocity, and strength) and their corresponding probabilities of occurrence (return periods). In this paper, the use of a Copernicus reanalysis product is evaluated for its applicability in preparing drift ice thickness distributions in the design phase of offshore wind turbines. An area surrounding the Kriegers Flak wind farm site in the Southern Baltic Sea is used as a case study. The drift ice thickness statistics of ice within the region which could potentially drift into the site were weighted according to drift directions, based on the wind direction frequency in the area. We found that between 1993-2017, drift ice at Kriegers Flak mainly occurred in 1996 with 0.1 m maximum ice thickness, in good agreement with estimations reported in the literature. Ice thickness probabilities have been created from the 1996 winter data and used as input for a fatigue damage analysis of an offshore wind turbine. The additional steps required to improve the suitability of Copernicus reanalysis data for use as input into design calculations are discussed.@en

With the recent surge in development of offshore wind in the Baltic Sea, Bohai Sea and other ice-prone regions, a need has arisen for new basin tests to qualify the interaction between offshore wind turbines and sea ice. To this end, a series of model tests was performed at the Aalto ice basin as part of the SHIVER project. The tests were aimed at modeling the dynamic interaction between flexible, vertically-sided structures and ice failing in crushing. A real-time hybrid test setup was used which combines numerical and physical components to model the structure. This novel test setup enabled the testing of a wide range of structure types, including existing full-scale structures for which ice-induced vibrations have been documented, and a series of single-degree-of-freedom oscillators to obtain a better understanding of the fundamental processes during dynamic ice-structure interaction. The tests were primarily focused on the dynamic behavior of support structures for offshore wind turbines under ice crushing loads. First results of the campaign show that the combination of the use of cold model ice and not scaling time and deflection of the structure can yield representative ice-structure interaction in the basin. This is demonstrated with experiments during which a scaled model of the Norströmsgrund lighthouse and Molikpaq caisson were used. The offshore wind turbine tests resulted in multi-modal interaction which can be shown to be relevant for the design of the support structure. The dataset has been made publicly available for further analysis.@en

With the ongoing development of offshore wind in cold regions where the foundations are exposed to sea ice, there is a strong need for data to validate the numerically predicted dynamic interaction between ice and structure used for design. Full-scale data is non-existent and only a limited number of experimental campaigns in ice tanks have been conducted for this specific problem. When compared to traditional structures subjected to sea ice loading like lighthouses and oil and gas platforms, the motion of the turbines at the ice action point is both in line with the ice drift direction but also significantly across due to the interaction of the turbine with the wind. Furthermore, the structure being slender overall and having a large top mass results in a very particular set of modes of oscillation where at least both the first and second global bending mode are expected to interact with the ice. To capture this complexity, a real-time hybrid test setup has been designed for basin tests in the SHIVER project and is presented in this paper. The setup uses two integrated linear actuators to control the motion of a rigid pile in two dimensions. Loads at the ice-action point are measured and used in a numerical model where these are combined with virtual loads, for example wind loading, to determine the response of the structure which is then applied in the physical setup by the actuators. The system allows to test a wide range of combinations of structural stiffness, mass, and damping, including structural properties typically associated with the relevant modes of oscillation of offshore wind turbines.@en

In the FATICE project we have addressed the fatigue damage on fixed offshore structures exposed to drifting ice. This is an important challenge in the development of energy production from offshore wind in the Baltic and involves at least five element: a) define ice statistics, b) predict the structural response (ice-structure interaction simulations), c) estimate the fatigue damage and d) carry out scale-model tests. We have used the Copernicus database and simple analytical equations to define the large-scale ice statistics and studied down-scaling to structural scale by comparing with ice load data on the Norströmsgrund lighthouse (LOLEIF and STRICE data). The VANILLA model allows for ice-structure interaction simulations and has been validated against the full-scale LOLEIF and STRICE data and against the model-scale ice in HSVA. The fully coupled and the traditional methods are compared. In the fatigue estimations studies the assumption of linear damage accumulation is challenged and load combinations from wave, wind and ice studied by assessing simulated time-series of the different loads. The main results is that sea ice cause the higher loads than wind and waves do , but the cumulative frequency of ice loads is much smaller than for wind and waves. The traditional model-scale ice tends to be too soft and/or too viscous so that a realistic breaking pattern combined with realistic force-time series is not been obtained for large aspect ratios. HVA has developed a crushing model ice (ICMI) in which the ice crystals are larger and the texture more uniform.@en