Search

Data-driven approaches have gained interest recently in the field of wind energy. Data-driven online estimators have been investigated and demonstrated in several applications such as online loads estimation, wake center position estimations, online damage estimation. The present work demonstrates the application of machine learning algorithms to formulate an estimator of the internal loads acting on the bearings of the drivetrain of onshore wind turbines. The loads estimator is implemented as a linear state-space model that is augmented with a non-linear feed-forward neural network. The estimator infers the loads time series as a function of the standard measurements from the SCADA and condition monitoring systems (CMS). A formal analysis of the available data is carried out to define the structure of the virtual sensor regarding the order of the models, number of states, architecture of neural networks. Correlation coefficient of 98% in the time domain and matching of the frequency signature are achieved. Several applications are mentioned and discussed in this work such as online estimation of the forces for monitoring and model predictive control applications.

This paper presents a feasibility analysis of vertical wake steering for floating turbines by differential ballast control. This new concept is based on the idea of pitching the floater with respect to the water surface, thereby achieving a desired tilt of the turbine rotor disk. The pitch attitude is controlled by moving water ballast among the columns of the floater. This study considers the application of differential ballast control to a conceptual 10 MW wind turbine installed on two platforms, differing in size, weight, and geometry. The analysis considers the following: (a) the aerodynamic effects caused by rotor tilt on the power capture of the wake-steering turbine and at various downstream distances in its wake; (b) the effects of tilting on fatigue and ultimate loads, limitedly to one of the two turbine-platform layouts; and (c) for both configurations, the necessary amount of water movement, the time to achieve a desired attitude, and the associated energy expenditure. Results indicate that – in accordance with previous research – steering the wake towards the sea surface leads to larger power gains than steering it towards the sky. Limitedly to the structural analysis conducted on one of the turbine-platform configurations, it appears that these gains can be obtained with only minor effects on loads, assuming a cautious application of vertical steering only in benign ambient conditions. Additionally, it is found that rotor tilt can be achieved on the order of minutes for the lighter of the two configurations, with reasonable water ballast movements. Although the analysis is preliminary and limited to the specific cases considered here, results seem to suggest that the concept is not unrealistic and should be further investigated as a possible means to achieve variable tilt control for vertical wake steering in floating turbines.

Energiekontor AG is a pioneering developer and operator of wind parks with technical innovation as one of its core values. ventodyne GmbH is a young dynamic company specialized in boosting the productivity of wind parks by wind farm control (WFC). Energiekontor and ventodyne have teamed up to bring WFC to the Withernwick wind farm, an onshore park with 13 closely-spaced wind turbines. In this document we report on the results of a six-month test campaign, where two of the Withernwick turbines were equipped with a wake-steering park controller. Toggling the park controller on and off provides clear and quantitative evidence of the effectiveness of this solution. The results also confirm the accuracy with which ventodyne estimated the gain in annual energy production (AEP) prior to field deployment.

This paper explores the potential benefits brought by the integration of lidar-assisted control (LAC) in the design of a wind turbine. The study identifies which design drivers can be relaxed by LAC, as well as by how much these drivers could be reduced before other conditions become the drivers. A generic LAC load-reduction model is defined and used to redesign the rotor and tower of three representative turbines, differing in terms of wind class, size, and power rating. The load reductions enabled by LAC are used to save mass, increase hub height, or extend lifetime. For the first two strategies, results suggest only modest reductions in the levelized cost of energy, with potential benefits essentially limited to the tower of a large offshore machine. On the other hand, lifetime extension appears to be the most effective way of exploiting the effects of LAC.

This paper describes a method to identify the heterogenous flow characteristics that develop within a wind farm in its interaction with the atmospheric boundary layer. The whole farm is used as a distributed sensor, which gauges through its wind turbines the flow field developing within its boundaries. The proposed method is based on augmenting an engineering wake model with an unknown correction field, which results in a hybrid (grey-box) model. Operational SCADA (supervisory control and data acquisition) data are then used to simultaneously learn the parameters that describe the correction field and to tune the ones of the engineering wake model. The resulting monolithic maximum likelihood estimation is in general ill-conditioned because of the collinearity and low observability of the redundant parameters. This problem is solved by a singular value decomposition, which discards parameter combinations that are not identifiable given the informational content of the dataset and solves only for the identifiable ones. The farm-as-a-sensor approach is demonstrated on two wind plants with very different characteristics: a relatively small onshore farm at a site with moderate terrain complexity and a large offshore one in close proximity to the coastline. In both cases, the data-driven correction and tuning of the grey-box model results in much improved prediction capabilities. The identified flow fields reveal the presence of significant terrain-induced effects in the onshore case and of large direction and ambient-condition-dependent intra-plant effects in the offshore one. Analysis of the coordinate transformation and mode shapes generated by the singular value decomposition help explain relevant characteristics of the solution, as well as couplings among modeling parameters. Computational fluid dynamics (CFD) simulations are used for confirming the plausibility of the identified flow fields.

This paper describes the design and characterization of a scaled wind turbine model, conceived to support wake and wind farm control experiments in a boundary layer wind tunnel. The turbine has a rotor diameter of 0.6 m and was designed to match the circulation distribution of a target conceptual full-scale turbine at its design tip speed ratio. In order to enable the testing of plant-level control strategies, the model is equipped with pitch, torque and yaw actuation and is sensorized with integrated load cells, rotor azimuth and blade pitch encoders. After describing the design of the turbine, its steady-state performance and wake characteristics are assessed by conducting experiments in two different wind tunnels, in laminar and turbulent conditions, collecting wake data with different measurement techniques. A large-eddy simulator coupled to an actuator-line model is used to develop a digital replica of the turbine and of the wind tunnel. For increased accuracy, the polars of the low-Reynolds-number airfoil used in the numerical model are tuned directly from measurements obtained from the rotor in operation in the wind tunnel. Results indicate that the scaled turbine performs as expected: measurements are repeatable and consistent, and the wake appears to have a realistic behavior in line with expectations and with a similar but slightly larger scaled model turbine. Furthermore, the predictions of the numerical model are well in line with experimental observations.

Wind turbines are designed to minimize the cost of energy, a metric aimed at making wind competitive with other energy-producing technologies. However, now that wind energy is competitive, how can we increase its value for society? And how much would a societal gain cost other stakeholders, such as investors or consumers? This paper tries to answer these questions from the perspective of wind turbine design. Although wind turbines produce green renewable energy, they also generate various impacts on the environment, as all human endeavours. Among all impacts, the present work adopts the environmental effects produced by a turbine over its entire life cycle, expressed in terms of CO2-equivalent emissions. A new approach to design is proposed, whereby Pareto fronts of solutions are computed to define optimal trade-offs between economic andenvironmental goals. The new proposed methodology is demonstrated on the redesign of a baseline 3 MW wind turbine at two locations in Germany, differing for typical wind speeds. Among other results, it is found that, in these conditions, a 1 % increase in the cost of energy can buy about a 5 % decrease in the environmental impact of the turbine. Additionally, it is also observed that in the specific case of Germany, very low specific-power designs are typically favored, because they produce more energy at low wind speeds, where both the economic and environmental values of wind are higher. Although limited to the sole optimization of wind-generating assets at two different locations, these results suggest the existence of new opportunities for the future development of wind energy where, by shifting the focus slightly away from a purely cost-driven short-term perspective, longer-term benefits for the environment (and, in turn, for society) may be obtained.

A new method is described to identify the aerodynamic characteristics of blade airfoils directly from operational data of the turbine. Improving on a previously published approach, the present method is based on a new maximum likelihood formulation that includes both errors in the outputs and in the inputs, generalizing the classical error-in-the-outputs only formulation. Since many parameters are necessary to meaningfully represent the behavior of airfoil polars as functions of angle of attack and Reynolds number, the approach uses a singular value decomposition to solve for a reduced set of observable parameters. The new approach is demonstrated by identify high quality polars for small-scale wind turbines used in wind tunnel experiments for wake and wind farm control research.

The formulation of Parametric Online Rainflow Counting implements the standard fatigue estimation process and a stress history in the cost function of a Model Predictive Controller. The formulation is tested in realistic simulation scenarios where the states are estimated by a Moving Horizon Estimator and the wind is predicted by a lidar simulator. The tuning procedure for the controller toolchain is carefully explained. In comparison to a conventional MPC in a turbulent wind setting, the novel formulation is especially superior with low lidar quality, benefits more from the availability of a wind prediction, and exhibits a more robust performance with shorter prediction horizons. A simulation excerpt with the novel formulation provides deeper insight into the update of the stress history and the fatigue cost parameters. Finally, in a deterministic gust setting, both the conventional and the novel MPC - despite their completely different fatigue cost - exhibit similar pitch behavior and tower oscillation.

This paper presents a feasibility analysis of vertical wake steering for floating turbines by differential ballast control. This new concept is based on the idea of pitching the floater with respect to the water surface, thereby achieving a desired tilt of the turbine rotor disk. The pitch attitude is controlled by moving water ballast among the columns of the floater. This study considers the application of differential ballast control to a conceptual 10 MW wind turbine installed on two platforms, differing in size, weight and geometry. The analysis considers: a) the aerodynamic effects caused by rotor tilt on the power capture of the wake-steering turbine and at various downstream distances in its wake; b) the effects of tilting on fatigue and ultimate loads, limitedly to one of the two turbine-platform layouts; and c) for both configurations, the necessary amount of water movement, the time to achieve a desired attitude and the associated energy expenditure. Results indicate that – in accordance with previous research – steering the wake towards the sea surface leads to larger power gains than steering it towards the sky. Limitedly to the structural analysis conducted on one of the turbine-platform configurations, it appears that these gains can be obtained with only minor effects on loads, assuming a cautious application of vertical steering only in benign ambient conditions. Additionally, it is found that rotor tilt can be achieved in the order of minutes for the lighter of the two configurations, with reasonable water ballast movements. Although the analysis is preliminary and limited to the specific cases considered here, results seem to suggest that the concept is not unrealistic, and should be further investigated as a possible means to achieve variable tilt control for vertical wake steering in floating turbines.

This paper describes the design and characterization of a scaled wind turbine model, conceived to support wake and wind farm control experiments in a boundary layer wind tunnel. The turbine has a rotor diameter of 0.6~meters, and was designed to match the circulation distribution of a target conceptual full-scale turbine at its design tip speed ratio. In order to enable the testing of plant-level control strategies, the model is equipped with closed-loop pitch, torque and yaw control, and is sensorized with integrated load cells, as well as with rotor azimuth and blade pitch encoders. After describing the design of the turbine, its performance and wake characteristics are assessed by conducting experiments in two different wind tunnels, in laminar and turbulent conditions, collecting wake data with different measurement techniques. A large-eddy simulator coupled to an actuator-line model is used to develop a digital replica of the turbine and of the wind tunnel. For increased accuracy, the polars of the low-Reynolds airfoil used in the numerical model are tuned directly from measurements obtained from the rotor in operation in the wind tunnel. Results indicate that the scaled turbine performs as expected, measurements are repeatable and consistent, and the wake appears to have a realistic behavior in line with expectations and with a similar slightly larger scaled model turbine. Furthermore, the predictions of the numerical model are well in line with experimental observations.

The aim of this paper is to analyze to which extent wind tunnel experiments can represent the behavior of full-scale wind turbine wakes. The question is relevant because on the one hand scaled models are extensively used for wake and farm control studies, whereas on the other hand not all wake-relevant physical characteristics of a full-scale turbine can be exactly matched by a scaled model. In particular, a detailed scaling analysis reveals that the scaled model accurately represents the principal physical phenomena taking place in the outer shell of the near wake, whereas differences exist in its inner core. A large eddy simulation actuator line method is first validated with respect to wind tunnel measurements, and then used to perform a detailed comparison of the wake at the two scales. It is concluded that, notwithstanding the existence of some mismatched effects, the scaled wake is remarkably similar to the full-scale one, except in the immediate proximity of the rotor.

This paper formulates laws for scaling wind turbine rotors. Although the analysis is general, the article primarily focuses on the subscaling problem, i.e., on the design of a smaller-sized model that mimics a full-scale machine. The present study considers both the steady-state and transient response cases, including the effects of aerodynamic, elastic, inertial, and gravitational forces. The analysis reveals the changes to physical characteristics induced by a generic change of scale, indicates which characteristics can be matched faithfully by a subscaled model, and states the conditions that must be fulfilled for desired matchings to hold. Based on the scaling laws formulated here, the article continues by considering the problem of designing scaled rotors that match desired indicators of a full-scale reference. To better illustrate the challenges implicit in scaling and the necessary tradeoffs and approximations, two different approaches are contrasted. The first consists in a straightforward geometric zooming. An analysis of the consequences of zooming reveals that, although apparently simple, this method is often not applicable in practice, because of physical and manufacturing limitations. This motivates the formulation of scaling as a constrained optimal aerodynamic and structural matching problem of wide applicability. Practical illustrations are given considering the scaling of a large reference 10 MW wind turbine of about 180 m in diameter down to three different sizes of 54, 27, and 2.8 m. Results indicate that, with the proper choices, even models characterized by very significant scaling factors can accurately match several key performance indicators. Additionally, when an exact match is not possible, relevant trends can at least be captured.

This paper compares the accuracy of four different FLORIS model implementations characterized by increasing levels of accounted physics. The predictions of each model were tested against experimental observations obtained from wind tunnel tests of a cluster of three scaled wind turbines. The experiments were conducted for a wide range of environmental conditions, and mainly with yawed turbines. The results showed a significant improvement in the accuracy of the base model by adding more details to the physical effects of the simulation, such as secondary steering and a heterogeneous flow field. It is also interesting to note that the FLORIS predictions become particularly accurate when considering inflow-dependent power losses in yaw-misaligned conditions.

In this study a new yaw mechanism for a small existing wind turbine model is designed and tested. The special requirement for this yaw device is its compact design due to the small length scaling factor of the model turbine. Such a dense design is crucial in order for the wake of the model turbine not to be influenced by the yaw mechanism. Furthermore, it has to have a fast system response and high accuracy to take into account the time scaling. Different concepts are investigated and a design located at the tower base is realized. To limit its drag, the structure is being concealed in an aerodynamically shaped cover. Wind tunnel measurements confirmed that the wake of the turbine is not influenced by the structure. A system analysis, using a Simulink model of the turbine helps finding the right motor and shows that the device has a fast system response with high accuracy. Consequently, the designed yaw mechanism is suited for the small wind turbine model, which can be used for experimental wake and wind farm control studies in the future.

The performance of an open-loop wake steering controller is investigated with a new unique set of wind tunnel experiments. A cluster of three scaled wind turbines, placed on a large turntable, is exposed to a turbulent inflow and dynamically changing wind directions, resulting in dynamically varying wake interactions. The changes in wind direction were sourced and scaled from a field-measured time history and mirrored on to the movement of the turntable. Exploiting the known, repeatable and controllable conditions of the wind tunnel, this study investigates the following effects: fidelity of the model used for synthesizing the controller, assumption of steady state vs. dynamic plant behavior, wind direction uncertainty, the robustness of the formulation in regard to this uncertainty, and a finite yaw rate. The results were analyzed for power production of the cluster, fatigue loads and yaw actuator duty cycle. The study highlights the importance of using a robust formulation and plant flow models of appropriate fidelity, and the existence of possible margins for improvement by the use of dynamic controllers.

A previously published wind sensing method is applied to an experimental dataset obtained from a 3.5 MW turbine. The method is based on a load-wind model that correlates once-per-revolution blade load harmonics to rotor-equivalent shears and wind directions. Loads measured during turbine operation are used to estimate online – through the load-wind model – the inflow at the rotor disk, thereby turning the whole turbine into a sort of generalized anemometer. The experimental dataset consists of synchronous measurements of loads, from blade-mounted strain gages, and of the inflow, obtained from a nearby met mast. As the mast reaches only to hub height, a second independent method is used to extend the met-mast-measured shear above hub height to cover the entire rotor disk. Part of the dataset is first used to identify the load-wind model, and then the performance of the wind observer is characterized with the rest of the data. Although the experimental setup falls short of providing a thorough validation of the method, it still allows for a realistic practical demonstration of some of its main features. Results indicate a good quality of the estimated linear shear both in terms of 1 and 10 min averages and of resolved time histories, with mean average errors around 0.04. A similarly accuracy is found in the estimation of the yaw misalignment, with mean errors typically below 3∘.

This paper presents the field validation of a method to estimate the local wind speed on different sectors of a turbine rotor disk. Each rotating blade is used as a scanning sensor that, travelling across the rotor disk, samples the inflow. From the local speed estimates, the method can reconstruct the vertical wind shear and detect the presence and location on an impinging wake shed by an upstream wind turbine. Shear and wake awareness have multiple uses, from turbine and farm control to monitoring and forecasting. This validation study is conducted with an experimental data set obtained with two multi-MW wind turbines and a hub-tall met-mast. Practical and simple procedures are presented and demonstrated to correct for the possible miscalibration of sensors. Results indicate a very good correlation between the estimated vertical shear and the one measured by the met-mast. Additionally, the proposed method exhibits a remarkable ability to locate and track the motion of an impinging wake on an affected rotor.

As wind turbines in a wind farm interact with each other, a control problem arises that has been extensively studied in the literature: how can we optimize the power production of a wind farm as a whole? A traditional approach to this problem is called induction control, in which the power capture of an upstream turbine is lowered for the benefit of downstream machines. In recent simulation studies, an alternative approach, where the induction factor is varied over time, has shown promising results. In this paper, the potential of this dynamic induction control (DIC) approach is further investigated. Only periodic variations, where the input is a sinusoid, are studied. A proof of concept for this periodic DIC approach will be given by the execution of scaled wind tunnel experiments, showing for the first time that this approach can yield power gains in real-world wind farms. Furthermore, the effects on the damage equivalent loads (DEL) of the turbine are evaluated in a simulation environment. These indicate that the increase in DEL on the excited turbine is limited.

The paper describes the development of a scaled model of complex terrain, suitable for terrain-wind turbine interaction wind tunnel studies, taking into account flow similarity criteria. The size and the geometry of the experimental model of the complex terrain were refined using results of CFD simulations in order to achieve the best possible flow similarity and avoid edge effects arising from the finite (relative to the rotor size) terrain geometry. Moreover, Particle Image Velocimetry was used to survey the flow field on a longitudinal plane along the terrain center line. Flow measurements with and without a wind turbine model enabled to quantitatively evaluate the speed up produced by the terrain in the region of the wind turbine and the effect of the terrain on the wake characteristics of the wind turbine model.

To account for the problem of an azimuthally constant induction in the BEM method, which influences on incorrectly predicted aerodynamic loads in the yawed flow, a skewed-wake model implementation to the BEM method has been performed. The numerical aerodynamic loads have been compared with the wind tunnel data of the NREL Phase VI and against another numerical campaign. At first, the model predictions have been validated against experimental data performed with aligned flow conditions, showing a reasonable match. Successively, the model predictions are validated against experimental results obtained with the wind turbine yawed. Results show, a possible better prediction of loads at yawed flow with Skewed-Wake correction, however the method does not overall correlate better, compared to the BEM method with implemented local variability of the induction factor.

Careful validation of the modelling and control actions is of vital importance to build confidence in the value of coordinated wind farm control (WFC). The efficiency of flow models applied to WFC should be evaluated to provide reliable assessment of the performance of WFC. In order to achieve that, FarmConners launches a common benchmark for code comparison to demonstrate the potential benefits of WFC, such as increased power production and mitigation of loads. The benchmark builds on available data sets from previous and ongoing campaigns: synthetic data from high-fidelity simulations, measurements from wind tunnel experiments, and field data from a real wind farm. The participating WFC models are first to be calibrated or trained using normal operation periods. For the blind test, both the axial induction and wake steering control approaches are included in the dataset and to be evaluated through the designed test cases. Three main test cases are specified, addressing the impact of WFC on single full wake, single partial wake, and multiple wake. The WFC model outcomes will be compared during the blind test phase, through power gain and wake loss reduction as well as alleviation of wake-added turbulence intensity and structural loads. The probabilistic validation will be based on the median and quartiles of the observations and WFC model predictions. Every benchmark participant will be involved in the final publication, where the comparison of different tools will be performed using the defined test cases. Instructions on how to participate are also provided on farmconners.readthedocs.io.