Power Production Of Wind Turbine Research Articles

Maximizing wind farm power output with the helix approach: Experimental validation and wake analysis using tomographic particle image velocimetry

AbstractWind farm control can play a key role in reducing the negative impact of wakes on wind turbine power production. The helix approach is a recent innovation in the field of wind farm control, which employs individual blade pitch control to induce a helical velocity profile in a wind turbine wake. This forced meandering of the wake has turned out to be very effective for the recovery of the wake, increasing the power output of downstream turbines by a significant amount. This paper presents a wind tunnel study with two scaled wind turbine models of which the upstream turbine is operated with the helix approach. We used tomographic particle image velocimetry to study the dynamic behavior of the wake under the influence of the helix excitation. The measured flow fields confirm the wake recovery capabilities of the helix approach compared with normal operation. Additional emphasis is put on the effect of the helix approach on the breakdown of blade tip vortices, a process that plays an important role in re‐energizing the wake. Measurements indicate that the breakdown of tip vortices and the resulting destabilization of the wake are enhanced significantly with the helix approach. Finally, turbine measurements show that the helix approach was able to increase the combined power for this particular two‐turbine setup by as much as 15%.

Integral Backstepping Sliding Mode Control for Maximizing the Power Production of Wind Turbines

Wind turbine control has attracted increasing attention, driven in part by evolving challenges due to the growing size and complexity of wind turbines. Addressing these challenges and maximizing wind turbine power production requires the application of advanced nonlinear control methods. Sliding Mode Control (SMC) has emerged as a promising approach in this context. Recent studies have explored the integration of an integral term with SMC, called I-SMC. This technique has been shown to result in system responses that exhibit chattering phenomena with noticeable state errors. This study aimed to address these issues through the introduction of a novel controller known as Integral Backstepping SMC (IB-SMC). This study demonstrated that IBSMC not only ensured the stability of wind turbines but also outperformed other control strategies, even in the presence of disturbances of approximately 30% of the rated electromagnetic torque. To validate the effectiveness of the proposed controller, extensive simulation tests were carried out using MATLAB /Simulink software to evaluate the controller's responsiveness to rapid changes in conditions, as well as its robustness and overall performance. A comparison was carried out between the IBSMC and previous SMCs to evaluate their ability to reduce steady-state error and chattering.

Analytical model for the power production of a yaw-misaligned wind turbine

Wake steering has proven to be effective in enhancing the power output of a wind farm. However, this approach still highly relies on empirical formulas to predict the power production of yawed turbines, limiting its potential in practical applications. In this study, an analytical model is proposed to predict the power production of a yaw-misaligned turbine under uniform inflow conditions. The model is based on the combination of the blade element theory and the momentum theory, with a modification in the latter to account for the disturbance on the spanwise velocity caused by the yawed turbine. A series of large eddy simulations were performed using a utility-scale wind turbine operating at yaw angles |γ|≤30° and tip-speed ratios λ=5–8. The validity of the proposed model is confirmed by the good agreement between the theoretical predictions and the simulation data. Furthermore, the well-known cosine model is shown to describe well the power production of the yawed turbine within the studied parameter range. However, the power-yaw loss exponent is not a constant, but rather a function of the tip-speed ratio. These findings may be useful in yaw optimization and control strategies in wind farms.

Multi-Agent Systems and Machine Learning for Wind Turbine Power Prediction from an Educational Perspective

Artificial intelligence (AI) is an umbrella term that encompasses different fields of study, and topics related to these fields are addressed separately or within the scope of AI. Multi-agent systems (MASs) and machine learning (ML) are the core concepts of AI that are taught during AI courses. The separate explanation of these core research areas is common, but the emergence of federated learning has triggered their combined usage. This paper describes a practical scenario in the energy domain where these technologies can be used together to provide a sustainable energy solution for predicting wind turbine active power production. The projects in the AI course were assigned prior to the step-by-step learning of MASs and ML. These concepts were applied using a wind turbine energy dataset collected in Turkey to predict the power production of wind turbines. The observed performance improvements, achieved by applying various agent architectures and data partitioning scenarios, indicate that boosting methods such as LightGBM yield better results even when the settings are modified. Additionally, a questionnaire about the assignments was filled out by the student groups to assess the impact of learning MASs and ML through project-based education. The application of MASs and ML in a hybrid way proves valuable for learning core concepts related to AI education, as evidenced by feedback from students.

Exploring influence of air density deviation on power production of wind energy conversion system: Study on correction method

Thousands of wind turbines are already installed, providing renewable energy to meet the growing energy demand of consumers. This article investigates the effect of air density deviation, which depends on weather conditions, on power production of operational wind turbines, and examines strategy to mitigate this effect. Throughout the year, the air density in the area where the wind turbines are installed experiences continuous variations. It affects the annual power production of the wind turbine. Therefore, it is essential to discover a method that can reduce the effect of air density changes on power production of the wind turbine without adding extra extender blade. The effect of air density deviation on parameter of the wind turbines is investigated using four years measurement data (with a 10-minute interval) as well as simulation data. To reduce air density effect, the pitch angle and torque gain are readjusted rather than setting it constant based on the air density value. The analysis and simulation study were conducted in Matlab/Simulink® software. The results revealed that the suggested correction in pitch angle and torque gain according air density effectively alleviated the effect of air density variations, and the power production of the wind turbine was increased.

Brief communication: On the definition of the low-level jet

Abstract. Low-level jets (LLJs) are examples of non-logarithmic wind speed profiles affecting wind turbine power production, wake recovery, and structural/aerodynamic loading. However, there is no consensus regarding which definition should be applied for jet identification. In this study we argue that a shear definition is more relevant to wind energy than a falloff definition. The shear definition is demonstrated and validated through the development of a European Centre for Medium-Range Weather Forecasts (ECMWF) fifth-generation reanalysis (ERA5) LLJ climatology for six sites. Identification of LLJs and their morphology, frequency, and intensity is critically dependent on the (i) vertical window of data from which LLJs are extracted and (ii) the definition employed.

Fixed-time fuzzy adaptive control scheme for doubly fed induction generator–based wind turbine

This work focuses on the issue of designing a fixed-time fuzzy adaptive controller for variable-speed doubly fed induction generator-based wind turbine. The main goal is to optimize the wind turbine power production, while improving the performance of the overall control system such as, convergence speed, tracking accuracy, and robustness. For the control design, the backstepping technique is adopted, and fuzzy adaptive systems are employed as universal approximators to estimate continuous functional uncertainties. Using fixed–time stability concept, a detailed mathematical stability proof of the closed-loop control system is performed. Simulation results are presented to testify the efficiency of the designed control approach.

Wake and power prediction of horizontal-axis wind farm under yaw-controlled conditions with machine learning

The main objective of this study is to employ the Extreme Gradient Boosting (XGBoost) machine learning algorithm to predict the power, wake, and turbulent characteristics of horizontal-axis wind farms under yaw-controlled conditions. For this purpose, a series of high-fidelity numerical simulations using LES method are performed over tandem NREL-5 MW wind turbines to generate the input data for training and testing in machine learning analysis. It is observed that XGBoost is more accurate for wake prediction of the yaw-controlled wind farms compared to ANN, which was used in previous studies. The results illustrate that XGBoost can predict the power with a mean deviation of 0.94 % for different yaw angles, while ANN can estimate the power generation with a mean deviation of 2.15 % for various tested yaw angles. At far wake regions (X > 2000 m) of the second wind turbine, the deviations reach below 1 %. Moreover, XGBoost requires a much shorter training time, 87.5 % faster than ANN. The power production of both wind turbines can be predicted more accurately with XGBoost compared to ANN. The wake prediction time of XGBoost is just 0.105 sec, while this time is 4.480 for the ANN model. In conclusion, XGBoost provides a significant reduction in error and training time compared to ANN and deep learning algorithms over yaw-misaligned wind farms.

Local flow and loads estimation on wake-affected wind turbines using graph neural networks and PyWake

The estimation of the flow conditions throughout a modern wind farm is a relevant albeit complex problem involving several different modelling scales. In this context, recent years have seen the rise of the demand - normally industry-driven - of quick, computationally inexpensive, wind farm flow field simulators upon which further calculations might be undertaken (e.g. power output, fatigue loads, etc.). Naturally, the trustworthiness of such models hinges upon their ability to accurately estimate wakes, under the evident trade-off between computational time and accuracy. One such simulator that has been recently gaining traction within the wind community is PyWake, a Python-based flow solver, with several possible add-ons and modularity options. Based on runs from PyWake, this contribution studies the applicability of graph neural networks (GNN) as a simultaneous wind turbine fleet, wind inflow and loads surrogate. As graphs are discrete mathematical sets of dependent objects, one can see how the layout-dependent interaction between wind turbines’ aerodynamics, from which wake arises, lends itself to be modelled as a graph. Thus, we introduce the use of GNNs for layout-agnostic joint modelling of rotor-averaged wind speed, turbulence intensity, power production and damage equivalent loads (DEL) on individual turbines of wind farms. To this end, probabilistic samples of inflow conditions from a Weibull distribution for wind speed, uniform wind direction and conditional normal distributions of wind shear and nacelle yaw angles are generated as main inflow properties in the numerical simulations. Additionally, arbitrary wind farm layouts are created based on varied geometric shapes with random parametrization (varying orientations, length/width ratio). Both the arbitrary layouts and the random inflow conditions are then used as inputs for PyWake. PyWake’s output, namely, rotor averaged wind speed, turbulence intensity, power production and DELs of each individual wind turbine in the wind farms, are used to train a general GNN model. We elect to implement an Encode-Process-Decode GNN as a graph learning model. We compare the performance of various approaches of the relational dependency representation (edge-forming techniques) amongst the nodes (wind turbines) on the graph, such as Delaunay triangulation, KNN, radius-based methods and fully connected scheme. In our analysis, we evaluate the accuracy of GNNs and its ability to generalize their joint predictions for unseen layouts and varying inflow conditions.

Robustness of an Economic Nonlinear Model Predictive Control for Wind Turbines Under Changing Environmental and Wear Conditions

In this letter, the authors have assessed the robustness of an Economic Nonlinear Model-Predictive Controller (ENMPC) aimed at maximizing the power production of wind turbines. The scope of this letter is to quantify the sensitivity of this type of controller concerning wind conditions, climate, wind speed prediction unavailability, and aerodynamic performance degradation. A power production controller’s robustness is crucial for the wind turbine industry due to the extreme variability of external conditions and the wear caused by long-term continuous operativity. Model-Predictive controllers are, in principle, more prone to robustness issues concerning standard controllers, a fact that limits their adoption on actual wind turbines. The analysis is performed with the fully-aeroelastic solver OpenFAST considering a wide set of realistic load cases. It is demonstrated that the ENMPC previously developed is robust to wind prediction unavailability and change in wind turbulence intensity. Conversely, it is not robust to the modelling error due to aerodynamic degradation. Indeed, a reduction in generated power concerning the reference controller is observed, especially for operating region two and end-life blades. Finally, a significant increase in power production is achieved considering the external temperature variation thanks to the ENMPC’s direct handling of the generator temperature constraint.

A novel Economic Nonlinear Model Predictive Controller for power maximisation on wind turbines

Reducing the Levelized Cost of Energy (LCoE) is one of the main objectives of the wind turbine industry. There are several ways to achieve this goal: reducing construction and installation costs, reducing Operating&Maintenance costs, or increasing the power output. In this work, an Economic Nonlinear Model Predictive Control strategy is developed to maximise the power production of wind turbines. A novel three-states, non-linear Reduced Order Model is developed to predict aerodynamic power, rotor thrust and generator temperature with suitable accuracy. The control action is obtained from a constrained optimisation problem that uses the developed model, where the objective is the maximisation of the integral of the aerodynamic power. A set of constraints (including a bound on the generator temperature and the rotor thrust) are imposed. First, the turbine model is validated against high-fidelity simulations, then the controller performance and robustness are assessed in the entire wind range of operation, obtaining a significant increase of average power. Apart from the assessment of the controller performance in OpenFAST, the controller robustness is verified, introducing errors in the estimation of incoming wind, up to the case of a complete lack of information. The controller (freely downloadable from a dedicated repository) is effective in all the operating regions without the need for logical switches. Moreover, thanks to the optimised numerical solver adopted, it can be applied to actual wind turbines (which require real-time algorithmic performance).

Large eddy simulation for the effects of ground roughness and atmospheric stratification on the wake characteristics of wind turbines mounted on complex terrains

The wake characteristics of wind turbine mounted on complex terrains are of great significance in wind energy utilization. The ground roughness and atmospheric stratification are two key factors for this issue and their coupling effect has not been fully studied before. In this research, the wake distributions of wind turbine mounted on two typical complex terrains and a real complex terrain are investigated using large eddy simulations and actuator disk model with rotation. The ground roughness and atmospheric stratification are simulated by adding source term in the governing equations of fluid dynamics to generated temperature gradient and vegetation canopy. It is found that the effects of ground roughness and atmospheric stratification on the wake of wind turbine are superimposed or counteracted depending on the shape of complex terrain. The presence of atmospheric stratification reduces the power of wind turbine drastically by more than 40% and the rough ground causes a small decrease in the power of wind turbine on typical complex terrains. For the multiple wind turbines on a real terrain with rough ground, the influence of atmospheric stratification on the wake characteristics and power production of each wind turbine strongly affected by the micrositing of wind turbines. The usual practice in industry of multiplying the oncoming wind profile by the turbine power curve overestimates the power production of wind turbines on complex terrains with atmospheric stratification by up to 670% compared with the numerical results.

Effect of blade contamination on power production of wind turbines

Wind turbines suffer from considerable power losses because of contamination on their blades, that can be due to erosion, wear, smog, insect, sand and dust particle impact. Blade contamination, its effects on the flows over the wind turbine blades and consequent power production losses form the main focus of the present study. These effects are mainly due to increased roughness on the blades leading to earlier laminar-turbulent transition and consequently, thicker boundary-layers on the blades. Early laminar-turbulent transition leads to a larger part of the flow over a blade being turbulent, thus increasing skin friction drag. Thicker boundary-layer on a blade results in blade profile being effectively modified, rendering the flow over the blade depart from ideal. In the present study, the effects of blade contamination on power output of contaminated wind turbine blades is investigated numerically using an in-house computational tool. Blade Element Momentum Method (BEM) combined with the Panel Method is used to calculate the local velocity and angle of attack at the blade sections, together with the power produced by the blade. Trajectories of particles causing contamination are calculated using Lagrangian approach, also yielding the impingement pattern of the particles on the blade surface, i.e. particle collection efficiency distribution. The effects of roughness on the boundary-layer flow are investigated by using an Integral Boundary-Layer Method, which yields the characteristics of the boundary-layer, i.e. laminar-turbulent transition location, increased skin-friction and thickening of the boundary-layer. The blade shape is modified due contamination thickness, the local height of which is assumed to be proportional to the local collection efficiency. Also, the roughness height distribution used in the boundary-layer calculations is assumed to be equal to the contamination thickness distribution on the blades. Power production and consequent losses of wind turbines with contaminated wind turbine blades are studied with respect to variations in particle size, wind speed and roughness height.

Deploying Taller Turbines in Complex Terrain: A Hill Flow Study (HilFlowS) Perspective

Terrain-induced flow acceleration is presented for the summertime, peak power season at Lawrence Livermore National Laboratory’s Site 300 for the Hill Flow Study (HilFlowS). HilFlowS, designed as an adjunct field campaign to the Department of Energy’s Second Wind Forecasting Improvement Project (WFIP2), provides wind profile observations at a second location in complex terrain for validating numerical atmospheric model simulations and for better understanding flow behavior over hills for wind power generation. One unique feature of HilFlowS was the inclusion of an undergraduate university student who helped plan and execute the experiment as well as analyze wind data from two remote sensing laser detection and ranging (lidar) instruments deployed along parallel ridgelines. HilFlowS examines the trend of building higher into the atmosphere for the purpose of increasing wind turbine power production and evaluates the wind resource in the Altamont Pass Region of Northern California for a set of wind turbines of differing hub-heights and rotor-disk diameters found in the area. The wind profiles show strongly channeled onshore flow above both hills, enhanced by strong subsidence aloft, which produces a wind maximum (Umax) around z = 10 m and strong negative shear throughout all of the evaluated rotor-disks for much of the summer wind season. Under these conditions, shear becomes more negative with increasing hub-height and increasing rotor-disk size. Rotor-disk equivalent wind speed (Uequiv), a measure of the average wind speed across the entire rotor-disk, is compared to hub-height, rotor length, and rated capacity factor for the set of turbines. Uequiv is most closely related to turbine hub-height and is negatively correlated given the low altitude of Umax. Based on these results, building the largest capacity, large rotor-disk wind turbine at the lowest possible hub-height appears to provide turbines in the Altamont with a fast, near-surface, onshore wind resource during the peak power season.

Full-coupled analysis of offshore floating wind turbine supported by very large floating structure with consideration of hydroelasticity

A multi-bodies hydroelastic model is developed and coupled to the full aero-hydro-servo-elastic analysis of a VLFS (very large floating structure) floating wind turbine. The proposed hydroelastic model discretizes the VLFS into a set of interconnected submodules. The hydrodynamic load acting on the entire VFLS is lumped to the mass center of submodules whilst the structural flexibility is represented by the finite element model used to connect submodules. The multi-bodies character makes the developed hydroelastic model robust to the floating structure and convenient to be embedded into coupled dynamic modelling of offshore floating wind turbines. The established model is validated against model test first and subsequently coupled to the numerical model of a moored VFLS floating wind turbine. The wind turbine power production, VFLS flexible deformation and structural loads are investigated. The analysis results reveal the significance of hydroelasticity on wind turbine power production and structural dynamics.

Experimental study on application of nacelle-mounted LiDAR for analyzing wind turbine wake effects by distance

The blade rotation of wind turbines causes the wake effect, which affects energy production of downstream wind turbines, and its magnitude and intensity vary with distance. To optimize the wind turbine layout for maximum efficiency, it is necessary to quantitatively evaluate the wake effect by distance. Using a nacelle LiDAR, it is possible to concurrently measure wind conditions at multiple positions. This study presents the results of analyzing the wake effects by distance, behind a 3 MW wind turbine, using a nacelle LiDAR. Wake wind data measured at 10 points by a nacelle LiDAR, mounted on the nacelle of a 1.5 MW downstream wind turbine, were analyzed against freestream wind data obtained from a 70-m-high met mast installed near the turbine. Variations in wind conditions in the wake region were analyzed as a function of distance and wind direction, and the influence of these factors on the power production of the downstream wind turbine was estimated using SCADA. The variations in wind speeds and turbulence intensities at multiple distances between 0.9D and 4.8D were quantitatively evaluated. Additionally, it was confirmed that the wake effects reduced the power performance by 23% at 5.7D under a wind speed of 8.5 m/s.

A reduced-order model for the near wake dynamics of a wind turbine: Model development and uncertainty quantification

Wind turbine power production and its variability are affected by unsteady, turbulent structures that are induced in the wake and emerge over a range of disparate length scales. However, the inclusion of the unsteady dynamics in wake models remains difficult in kinematic modeling where steady-state conditions are assumed. We develop and quantify the uncertainty of an unsteady wake model by leveraging model-order reduction. The wake is modeled as a dynamical system based on dynamic mode decomposition and compressive sensing, where the system is reduced by designating an objective function to select energetic turbulent structures represented by dynamic modes. A series of large-eddy simulations are undertaken using the actuator line model and the actuator surface with nacelle model for training and testing data. Differences in the turbine parameterizations of near wake simulations are used to identify how modes are related to turbulent structures and are selected through compressive sensing. The results showed that the flow field can be constructed with few modes. The modes are the most energetic and have frequencies related to pertinent features such as tip and hub vortices. While the error in the training phase is dependent on the number of modes, the relative error remains less than 6% and is largely independent of the number of modes during the prediction phase if certain modes are retained. The error accumulates over time near the turbine blades. However, error accumulation does not have a significant impact on the prediction of instantaneous velocity in the wake further downwind.

Time-Averaged Wind Turbine Wake Flow Field Prediction Using Autoencoder Convolutional Neural Networks

A convolutional neural network (CNN) autoencoder model has been developed to generate 3D realizations of time-averaged velocity in the wake of the wind turbines at the Sandia National Laboratories Scaled Wind Farm Technology (SWiFT) facility. Large-eddy simulations (LES) of the SWiFT site are conducted using an actuator surface model to simulate the turbine structures to produce training and validation datasets of the CNN. The simulations are validated using the SpinnerLidar measurements of turbine wakes at the SWiFT site and the instantaneous and time-averaged velocity fields from the training LES are used to train the CNN. The trained CNN is then applied to predict 3D realizations of time-averaged velocity in the wake of the SWiFT turbines under flow conditions different than those for which the CNN was trained. LES results for the validation cases are used to evaluate the performance of the CNN predictions. Comparing the validation LES results and CNN predictions, we show that the developed CNN autoencoder model holds great potential for predicting time-averaged flow fields and the power production of wind turbines while being several orders of magnitude computationally more efficient than LES.

Evaluation of operational strategies on wind turbine power production during short icing events

Cold climates around the world are seeing increasing investment in wind power generation. The benefits of cold regions, however, come with unique challenges such as the accumulation of ice on wind turbine blades, which impacts the aerodynamic properties of the blades and, in turn, power production. One decision that must be made when facing an icing event is whether to maintain operation of the turbine or to pause the rotor until the event has passed. Powering through an icing event allows the operator to maintain power production, but with an increase in ice accumulation and associated adverse effects. Pausing the wind turbine results in less ice accumulation and may therefore provide higher power production once operation is resumed. The aim of this research was to complete a comparative assessment of these two strategies with regards to overall impact on power production. A numerical Ice and Power Model was developed, validated, and employed to assess power production impacts on a 1.5 MW wind turbine for five 2-h icing events with varying atmospheric parameters. The simulation results on short icing events provide evidence that while pausing the turbines does indeed result in significantly less ice accumulation, the positive impact on power production compared to the power-through strategy is insufficient to justify the shut-down strategy. This is found to be due to the overpowering impact of surface roughness from ice crystal beading on blade aerodynamics, which occurs whether or not the blades are rotating.

A new method for simulating multiple wind turbine wakes under yawed conditions

Counter-rotating vortices generated in wake steering not only deform the turbine wake, but also can make the wake trajectory of a non-yawed downwind turbine deviate from its rotor centerline, referred to as “secondary wake steering” phenomenon. Recent studies have also shown that the vortex interactions become clearer when the wind farm includes multiple turbines. However, in the common analytical models for active yaw control, the effects of these vortices are not considered. Evidently, this omission can lead to a decrease in model prediction accuracy. To compensate for it, a new analytical wind farm model is proposed. It adopts a physical-based momentum conserving wake superposition method to deal with the interaction of multiple wakes, in which, not only combining the streamwise velocity deficit of each individual yawed wind turbine, but also the transverse velocity from different wakes. Additionally, an “added yaw angle” is defined for a downwind turbine operating in upstream yawed turbine wakes, to reflect the change in local wind direction it perceives. For validation purposes, the LES wind field obtained from the SOWFA tool is used as a reference, and the newly proposed model is found to agree well with LES results and outperforms the representative conventional analytical model in almost all test cases. The new model can successfully reproduce the “secondary wake steering” phenomenon in the overlapped wake, and provides significant improvements in predicting power production of wind turbines.