Perspectives and Challenges Related Offshore Wind Turbines in Deep Water

In the coming decades, energy from offshore wind turbines is expected to be an important energy source in electric power systems. To reduce emissions of greenhouse gases into the atmosphere, the CO₂ emissions from electric power systems must be considerably reduced. The European Union (EU) has established the “Green Deal” (https://ec.europa.eu/info/strategy/priorities-2019-2024/european-green-deal_en (accessed on 8 April 2022) with ambitions to be the first carbon-neutral continent and aim for carbon neutrality before 2050. This implies a major redesign of the present energy system with improved energy efficiency, and replacing fossil fuels with renewables and electrification [1]. The transition from a fossil-fuel-based energy system to a renewable system may reduce the total global need for primary energy, even if, according to the UN sustainable development goals (https://sdgs.un.org/goals (accessed on 8 April 2022)) (SDGs), more people should have access to clean and affordable energy (SDG 7). This can be obtained as renewable energy sources such as solar energy, wind energy and hydropower, which produce electricity directly and with higher efficiency than electricity produced by fossil fuel systems. The options for increasing hydropower production in industrialized areas such as Europe are limited; therefore, solar and wind energy will play an increasingly important role. The two are partly complementary both in terms of time and space. During winter, there may be little solar energy, but good wind conditions; in Southern Europe, the sun conditions are generally good, while in Northern Europe, the wind conditions are superior (https://globalatlas.irena.org/workspace (accessed on 8 April 2022)).

In 2021, the International Energy Agency [1] (IEA) issued a report discussing possible pathways towards carbon neutrality in Europe by 2050. A massive upscaling of the installed wind power is part of this pathway. In a special report on offshore wind energy [2] it is implied that offshore wind energy may become the largest contributor to the European electricity system by 2040. The EU’s strategy on offshore renewable energy [3] the EU Commission) illustrates the upscaling needed: from 12 GW installed capacity in 2019 (Great Britain not included) to 300 GW by 2050. This implies installation of 6 GW/year on average from 2020 to 2030 and 12 GW/year on average from 2030 to 2050. In addition, non-EU European countries, in particular Great Britain, have ambitious plans for developing offshore wind energy. In the USA, the development has been slower, but ambitious plans are evolving on both the east and west coasts. In Asia, in particular China, Japan and South Korea, there are also ambitious plans for offshore wind energy development [4].

One essential reason for the projected future position of offshore wind in the electricity supply is the enormous amount of available wind energy resources. Estimates on the available technical resources show resources an order of magnitude larger than the present worldwide consumption of electric energy [2,5]. Technical resources may be estimated in many ways. Starting from Geophysical resources, the technical resources are limited by what is assumed to be technically and economically feasible. Furthermore, the usable resources are limited by subtracting areas with obvious conflicts as protected areas, ship routes, military areas, important fishing areas, etc. Even after such exercises, the offshore wind resources are vast.

Two decades ago, the estimates of available technical resources from offshore wind energy were frequently limited by a certain (short) distance from shore and shallow water depths (typically less than 50 m), allowing for bottom-fixed foundations. If floating wind turbines and water depths down to 1000 m are included, estimates of the global technical potential for offshore wind energy has been given as 330,000 TWh/year (330,000 TWh/year corresponds to an installed capacity of 75,343 GW at a capacity factor of 0.5) of which 230,000 TWh/year is in water depths greater than 60m [5]. Including water depths down to 2000 m, which presently seems very optimistic, estimates of a global potential in excess of 420,000 TWh/year have been given [6]. Both of these estimates are an order of magnitude larger than the 2019 global electricity consumption of 23,000 TWh/year [7].

From the above figures, and because many coastlines around the world are missing a wide and shallow ocean shelf, wind turbines need to be installed in water depths deeper than approximately 50 m, calling upon floating support structures. One of the early studies of floating wind turbines was presented by [8,9]. They presented a system aiming for 100–300 m water depth. The floating structure was designed as a vertical cylinder with a horizontal circular disc at the bottom. The vertical cylinder had a draft of 28.5 m and a diameter of 19 m. The disc will contribute to added hydrodynamic inertia in vertical direction as well as added damping. The wind turbine had a rotor diameter of 60 m corresponding to a rated power of 1.4 MW. The concept was not very different from several of the concepts considered more recently. Model tests using both wind and wave loading were conducted. However, as many more recent researchers also have experienced, Tong and Cannell found that performing model testing with an accurate modelling of both wind and wave loading is a very challenging task. An important finding was that extreme loads in the tower and mooring system do not necessarily occur during extreme wind and wave conditions, but rather during certain operational conditions with large wind thrust.

After the work by Tong and Cannell, it was not until 2009 that the world’s first multimegawatt floating wind turbine was installed, the Hywind Demo spar with a 2.3 MW turbine [10]. By that time, a great deal of skepticism was expressed related to the feasibility of floating offshore wind turbines. The turbine was meant to demonstrate that a standard multimegawatt turbine mounted upon a floating foundation would work. The demonstration project was planned for two years. During this period, a great deal of data was collected to be used in order to verify the numerical analysis and feasibility of the concept [11,12,13]. The project was a success, demonstrating a high capacity factor, operational reliability and dynamic behaviour in accordance with the predictions. Today, 13 years after installation, the turbine is still in operation. The fundamental skepticism has abated, but there remains a significant amount of work to be undertaken in order to optimize the floating wind turbines and to arrive at reliable solutions fitted for mass production that can deliver electricity at an acceptable price. One track towards reduced costs has been to increase the size of the turbines.

During the almost 30 years since the work by Tong and Cannell, and the 13 years since the deployment of the Hywind Demo floating wind turbine, state-of-the-art wind turbines have dramatically increased in size. Present-day state-of-the-art turbines have a diameter of up to 240 m [14]. This means that the blade tip reaches more than 260 m above sea level. Such dimensions challenge both current knowledge and design standards. What do we know about the mean wind profile at these heights? Can we still assume that the logarithmic or exponential wind profiles are reasonable approximations of the real wind profile in all situations? The answer is obviously no. The logarithmic or exponential wind profiles are developed using standard boundary layer equations and assuming a neutrally stratified atmospheric boundary layer. Often, these assumptions are not valid. The turbulence also introduces new challenges. As will be addressed in more detail below, the floating wind turbines have natural frequencies at very low frequencies, in a frequency range neglected by the designers of bottom-fixed wind turbines. By investigating various methods for generating numerical wind spectra, using standard spectra, large eddy simulations and measured wind spectra, very different results are obtained in the low frequency range [15]. Even larger uncertainties exist for the vertical and lateral coherence of the turbulent flow field [15]. For the vertical coherence, some support may be found in observations from meteorological masts. However, little information exists for the horizontal coherence and at heights relevant for state-of-the-art wind turbines. Various models and assumptions about the coherence may result in very different dynamic wind loads on a floating wind turbine [16]. For bottom-fixed wind turbines, the simulation of a dynamic response due to wind loading has normally been performed with numerically generated turbulent wind fields with a duration of 10 min. This is acceptable as most natural periods of the turbines have been below a couple of seconds. However, for a floating turbine, the longest natural periods, corresponding to rigid body motion of the floater in the mooring system, may be in the order of a couple of minutes. The duration of the simulations should thus be at least one hour, preferably longer. In real wind measurements, it is hard to find stationary conditions with a duration of several hours as the weather is steadily changing.

The present Special Issue of Energies addresses various important issues related to floating offshore wind turbines. It does not address the issues related to the wind loads mentioned above, but focuses instead on the modelling of the elastic structure, wave loading, model testing, and maintenance. All these issues are extremely important for optimum design of floating wind turbines, a prerequisite to the reduced cost of energy.

Ishihara and Liu [17] study the dynamic response of a semi-submersible floating wind turbine in combined waves and current. Using the Morison equation, the classical problem of determining the proper values for the inertia and drag coefficients appear. Ishihara and Liu propose splitting drag coefficient into an oscillatory component and a steady component. The coefficients depend upon the Reynolds number (Rn) as well as the Keulegan-Carpenter number (KC) for the oscillatory component. Model tests, full-scale measurements as well as Large Eddy Simulations (LES) are employed to estimate the 3 proper values of the coefficients. They observe a significant increase in dynamic mooring line loads when combining waves and current versus the case of waves alone. Considering that the fatigue damage in a steel mooring line, wire or chain typically increases with the third power of the dynamic load amplitude, it is extremely important to have good estimates on the mooring load. Ishihara and Liu consider the mean hydrodynamic loads; if the mean wind loads were added, a significantly higher mean tension would be present. The dynamic load in the mooring lines is a non-linear function of the mean and dynamic displacement of the floater. If the line is heavily loaded and the transverse drag coefficient of the line is large, as for chains, the phenomenon “drag locking” may occur for the dynamic motions. In case of “drag locking”, the apparent dynamic stiffness of the mooring line is much higher than the quasistatic stiffness corresponding to the actual mean load. Asymptoticly, the apparent dynamic stiffness approaches that of the elastic stiffness of the line [18]. The combination of large mean mooring line loads and the requirement of a low natural period in surge and pitch is a serious design challenge, particularly for shallow water conditions. Thus, both fatigue and extreme loads in the most severely loaded mooring lines should be investigated carefully and one should look for innovative designs that can reduce the dynamic stiffness and thus the extreme loads.

Liu and Ishihara [19] study the same semisubmersible platform and apply the same computational method as in the 2020 article mentioned above. In the 2021 article they include a method to compute the dynamic sectional loads on the various structural components. They also study the impact of using an elastic model of the platform. As may be expected, using an elastic but stiff representation of the semisubmersible does not have any significant impact on the motions and mooring line loads as compared to a rigid body representation. Numerical analysis and model tests compare favorably with respect to sectional loads in the elastic model. A simple optimization of the structure is carried out by studying the impact of adding various struts and bracings. Liu and Ishihara provide a fast and fairly accurate computational method that may be used for more extensive optimization, as optimizing main dimensions, pontoon cross section, etc. The cost function in the optimization could be the steel weight. Challenges in the optimization occur when it comes to determine constraints for motions, natural periods, etc. The next step in such an optimization should be to include dynamic wind loads.

Lu, Sun and Xue [20] address optimization of maintenance of offshore wind farms. This is a topic of increasing importance as the number of offshore wind turbines world-wide increases, and these turbines are located further from the shore. The need for condition monitoring and good estimates on “time-to-failure” for various components will have increased attention. From an operator point of view, production loss due to failing turbines and cost of maintenance and replacement of components must be balanced. The maintenance of offshore wind turbines also introduces uncertainty related to accessibility or “weather windows”, i.e., situations where wind and wave conditions make it possible to access the turbine for a sufficiently long time to complete the required work. Monte Carlo simulations are well suited for such studies, and this is what Lu, Sun and Xue use. They find the probability distribution of the duration of “waiting on weather”, i.e., the duration of periods with wave heights over a certain threshold. To estimate the proper statistical distributions, real measurements of significant wave height are used. This approach is reasonable for the optimization task at hand. A possible follow-up of this study is to use the time records of waves, preferably over many years, in the simulations. That would make it possible to include two additional factors: the increased uncertainty in the weather forecast with the length of the forecast, and the increased uncertainty of wave height forecasts for low wave conditions. In real life, the decision is go/no go for an operation. How should the two uncertainties mentioned be accounted for in the decision? Frequently, the uncertainty is accounted for by including a so-called “alpha-factor” in standards for marine operations [21]. A more systematic and improved principle for accounting for the uncertainty would be welcomed.

Walia et al. [22] discuss in detail the model set-up and numerical analysis for a 5 MW wind turbine on a tension leg platform (TLP). Their work clearly demonstrates many of the seemingly trivial challenges that arise when making a model scale representation of an elastic support structure for wind turbines. The basic idea behind the TLP is to restrict the motions in heave, roll and pitch by introducing a very stiff tether. Thus, the natural frequencies of these modes are in the same range as the natural frequencies of the first bending modes of the tower. Such proximity in natural frequencies calls for extra awareness in the design and modelling of the structure. Walia et al. discuss several issues related to the modelling and measurement set-up for the model scale tests. They also make a numerical model of the system and compare natural frequencies and modal damping in model tests versus the numerical results. Additional viscous damping is introduced in the numerical model to fit the model tests. Herein is a dilemma for model tests. The viscous damping is normally larger for the model than for the full-scale system. Thus, certain resonant responses may be underestimated in model testing. Additionally, the coupling effects and the off-diagonal terms in the viscous damping matrix should be carefully considered. Such coupling terms may introduce dynamic effects not immediately observed by studying pure eigenmodes. It will be of great interest to see how model tests and numerical simulations compare when the complete system—which include an operating wind turbine—is studied. As mentioned above, to obtain realistic dynamic responses, the wind loads must be modelled properly, and must account for vertical shear, turbulence intensity and coherence in a realistic manner. Additionally, the turbine torque and blade pitch must be controlled to avoid excessive motions in surge and yaw at wind speeds above rated [23,24,25].

Sarkar and Fitzgerald [26] use Kane’s method for modelling a floating wind turbine on a spar foundation. Kane’s method for modelling multibody systems uses separate coordinate systems for each body and concepts as the partial velocities, where the total velocities are obtained by summation of partial velocities multiplied by a generalized speed. In the work of Sarkar and Fitzgerald, it is shown in detail how elastic modes, hydrodynamic forces and vibration dampers can be implemented using Kane’s method. They claim that the solution of the dynamics using Kane’s method is easier to implement in a computer code and faster to solve than more classical approaches. In the literature, many floating offshore wind concepts have been proposed. Most of them are at the conceptual stage. In order to improve the designs and make them ready for use in the offshore wind industry, comprehensive optimization analyses must be performed. In order to do this, it is crucial to determine the availability of efficient methods for modelling and analyzing the dynamic behaviour of the turbines, as well as properly accounting for the elasticity. It is, therefore, a positive to observe that new approaches are tested with this in mind, and that the modelling involved is well-documented. An important follow-up of this study would be the implementation of dynamic wind loads and motion controllers.

Compared to bottom-fixed offshore wind turbines, floating wind turbines have a more complex dynamic behaviour involving a much wider range of natural frequencies. Mooring lines and motion controllers used to avoid excessive low frequency motions are new components not used for bottom-fixed wind turbines [25]. We see a steady growth in the size of the turbines and thus a growth in the size of the support structures as well. With increased size, radiation–diffraction effects, slamming, wave drift loads, etc. should be accounted for. Here, methods developed for the offshore oil and gas and maritime industries are available and should be implemented. There are new challenges that are related to the size of the rotors. When the tip of the rotor reaches more than 250 m above sea level, it is necessary to have an improved understanding of the wind profile and turbulent structure of the wind. This may change our approach to implementing wind loads in numerical models. It is, therefore, a positive to see that several of the articles in this Special Issue have focused upon a proper modelling of the wave-induced dynamics; this enables us to take the next steps towards advanced modelling of wind loads. An improved understanding of dynamic wind loads will also necessitate a rethink of the design of the controller systems.