A Data-Driven, Cooperative Approach for Wind Farm Control: A Wind Tunnel Experimentation

This article proposes an active power control strategy for dynamically maximizing power production in waked wind farm (WF) based on an accelerated asynchronous distributed calculation scheme (DCS). By adopting a two-degree-of-freedom (2Dof) active power control strategy based on model predictive control (MPC), the generator speed and pitch angle of wind turbines (WT) can be coordinated to enhance the long-term WF-level cumulative power production and power reserve subject to strong wake effects. The proposed strategy is significantly different from existing model-free control strategies, and can quickly respond to time-varying wind speeds through online feedback of both WT mechanical status and dynamic wake. The proposed strategy can greatly improve the efficiency of solving large-scale optimization problems through two designs: first, using asynchronous DCS to reduce computation time and strengthen robustness to communication failures. Second, the Nesterov acceleration strategy is introduced to significantly reduce the number of iterations and enhance the convergence rate. Additionally, a novel control command issuance strategy is devised to address the issue of wake delay. In a 1000s simulation, the proposed strategy achieves a remarkable improvement of 10.81% and 18.29% in cumulative power production and power reserve of WF, respectively.

Abstract. This paper studies a closed-loop wind farm control framework for active power control (APC) with a simultaneous reduction of wake-induced structural loads within a fully developed wind farm flow interacting with the atmospheric boundary layer. The main focus is on a classical feedback control, which features a simple control architecture and a practical measurement system that are realizable for real-time control of large wind farms. We demonstrate that the wake-induced structural loading of the downstream turbines can be alleviated, while the wind farm power production follows a reference signal. A closed-loop APC is designed first to improve the power-tracking performance against wake-induced power losses of the downwind turbines. Then, the nonunique solution of APC for the wind farm is exploited for aggregated structural load alleviation. The axial induction factors of the individual wind turbines are considered control inputs to limit the power production of the wind farm or to switch to greedy control when the demand exceeds the power available in the wind. Furthermore, the APC solution domain is enlarged by an adjustment of the power set-points according to the locally available power at the waked wind turbines. Therefore, the controllability of the wind turbines is improved for rejecting the intensified load fluctuations inside the wake. A large-eddy simulation model is employed for resolving the turbulent flow, the wake structures, and its interaction with the atmospheric boundary layer. The applicability and key features of the controller are discussed with a wind farm example consisting of 3×4 turbines with different wake interactions for each row. The performance of the proposed APC is evaluated using the accuracy of the wind farm power tracking and the wake-induced damage equivalent fatigue loads of the towers of the individual wind turbines.

The flow speed-up generated by windbreaks can be used to increase the power production of wind turbines. However, due to the increased drag imposed by the windbreaks, their use in large wind turbine arrays has been questioned. We use large eddy simulations to show that windbreaks can increase the power production of large wind farms. A crucial finding is that windbreaks in a wind farm should be much lower than for a single turbine case. In fact, the optimal windbreak for an isolated turbine can reduce wind farm performance. The optimal windbreak height in a wind farm namely depends on the right balance between flow speed-up over the windbreak and the drag imposed by all windbreaks in the farm. The increased performance is a result of the favorable total pressure flux created by the windbreaks.

In order to meet the increasing demand of wind energy utilization, wind turbines (WTs) are developing toward the trend of large size and large capacity. In such a trend, various advanced yaw control strategies have been proposed to improve large WTs' comprehensive performance, but the analysis and summary of these strategies are still lacking. Therefore, it is necessary to have a review of yaw control, which not only enables readers to understand the current status of yaw control research but also promotes the development of wind energy technology. This paper presents a review of the current situation of yaw control for WTs, focusing on the mechanical/aerodynamic parts. The mechanical part is concerned with the WT yaw system and its effect on the fatigue load of the WT, and the aerodynamic part involves the wind energy capture and wake redirection to reduce the impact on adjacent WTs. In this review, the existing yaw control methods are classified in term of three control objectives: (1) increasing the wind energy capture of a single WT, (2) reducing the fatigue load of a single WT, and (3) maximizing the total power production of the whole wind farm and optimizing the wind farm fatigue load. On this basis, the control mechanism, the control algorithm, and the results are presented and analyzed in detail. Meanwhile, the advantages and disadvantages of the existing achievements are discussed. In addition, in a conclusion of the review, the future research direction has been identified.

During recent years, wind energy has moved from an emerging technology to a nearly competitive technology. This fact, coupled with an increasing global focus on environmental concern and a political desire of a certain level of diversification in the energy supply, ensures wind energy an important role in the future electricity market. For this challenge to be met in a cost-efficient way, a substantial part of new wind turbine installations is foreseen to be erected in big onshore or offshore wind farms. This fact makes the production, loading and reliability of turbines operating under such conditions of particular interest.

This paper presents a preliminary study of a model-free approach based on spiral dynamic algorithm (SDA) formaximizing wind farms power production. The SDA based approach is utilized to find the optimal control parameter of each turbine to maximize the total power production of a wind farm. For simplicity, a single row wind farm model with turbulence interaction between turbines is used to validate the proposed approach. Simulation results demonstrate that the SDA based method produces higher total power production compared to the particle swarm optimization (PSO) and game theoretic (GT) based approaches.

SummaryWind energy has shown a remarkable potential for fulfilling the increasing world's energy demand in a clean and sustainable way. The wind energy industry has installed increasingly sophisticated and larger wind turbines particularly in offshore regions to capture such energy as efficiently and cost effectively as possible. The rapid growth in size and capacity of wind turbines together with harsh climate conditions and limited accessibility in offshore regions all result in higher failure rates and increased maintenance requirements and costs. Such difficulties motivate the use of advanced fault detection and diagnosis and fault‐tolerant control schemes in wind farms to improve their reliability and availability. Given the importance of this issue, this paper uses a fuzzy model reference adaptive control approach in a cooperative framework that is oriented to the design and development of a novel fault‐tolerant cooperative control scheme in a wind farm. This scheme handles decreased power generation faults in a wind farm caused by turbine blade erosion and debris buildup on the blades over time. The effectiveness and performance of the proposed scheme is demonstrated by a series of simulations on an advanced large offshore wind farm benchmark model in the presence of wind turbulence, measurement noise, load variations, and realistic fault scenarios.

Wind farms with counter-rotating wind turbines

Development of a Wind to Power Model for Wind Farm Power Production Forecasting

Through wind farm control, the negative impact of the wake effect on power production could be mitigated. In previous research on wind farm control, the centralized optimization method was mostly employed due to its simplicity. However, centralized optimization control methods will face difficulties for large wind farms, in which more complex wake coupling relationships among wind turbines will make the optimization highly inefficient. In order to boost the efficiency of the optimization, the weighted graph wake decoupling (WGWD) method is proposed in the current research, which decomposes the complex optimization problem into accessible sub-problems solved in parallel. A wake weighted graph portraying the wake coupling relationship among wind turbines has been established. The decoupling approach applied on a weighted graph partitions the wind farm into uncoupled subsets of wind turbines, simplifying the optimization problem. By using distributed optimization method, the yaw angles of wind turbines in each subset have been optimized to maximize the power generation of the subset. The results show that the proposed method has high computational efficiency and can also improve wind farm total power production. Taking advantage of its high efficiency, the proposed wind farm optimization control method is suitable for future real-time active yaw control of large wind farms.

This paper describes the development of a wind turbine (WT) component lifetime fatigue load variation map within an offshore wind farm. A case study on the offshore wind farm Horns Rev I is conducted with this purpose, by quantifying wake effects using the Dynamic Wake Meandering (DWM) method, which has previously been validated based on CFD, Lidar and full scale load measurements. Fully coupled aeroelastic load simulations using turbulent wind conditions are conducted for all wind directions and mean wind speeds between cut-in and cut-out using site specific turbulence level measurements. Based on the mean wind speed and direction distribution, the representative 20-year lifetime fatigue loads are calculated. It is found that the heaviest loaded WT is not the same when looking at blade root, tower top or tower base components. The blade loads are mainly dominated by the wake situations above rated wind speed and the highest loaded blades are in the easternmost row as the dominating wind direction is from West. Regarding the tower components, the highest loaded WTs are also located towards the eastern central location. The turbines with highest power production are, not surprisingly, the ones facing a free sector towards west and south. The power production results of few turbines are compared with SCADA data. The results of this paper are expected to have significance for operation and maintenance planning, where the schedules for inspection and service activities can be adjusted to the requirements arising from the varying fatigue levels. Furthermore, the results can be used in the context of remaining fatigue lifetime assessment and planning of decommissioning.

<strong class="journal-contentHeaderColor">Abstract.</strong> Offshore wind farms are more commonly installed in wind farm clusters, where wind farm interaction can lead to energy losses; hence, there is a need for numerical models that can properly simulate wind farm interaction. This work proposes a Reynolds-averaged Navier-Stokes (RANS) method to efficiently simulate the effect of neighboring wind farms on wind farm power and annual energy production. First, a novel steady-state atmospheric inflow is proposed. This inflow model is well suited for RANS simulations of large wind farms because it does not lead to the development of nonphysical wind farm wakes. Second, a RANS-based wind farm parametrization is introduced, the actuator wind farm (AWF) model, which represents the wind farm as a forest canopy and allows to use of coarser grids compared to modeling all wind turbines as actuator disks (ADs). When the horizontal resolution of the RANS-AWF model is increased, the model results approach the results of the RANS-AD model. A double wind farm case is simulated with RANS to show that replacing an upstream wind farm with an AWF model only causes a deviation less than 1 % in terms of wind farm power of the downstream wind farm. Most importantly, a reduction in CPU hours of 74.4 % is achieved, provided that the AWF inputs are known, namely, wind farm thrust and power coefficients. The reduction in CPU hours is further reduced when all wind farms are represented by AWF models; namely 89.3 % and 99.9 %, for the double wind farm case and for a wind farm cluster case consisting of three wind farms, respectively. For the double wind farm case, the RANS models predict different wind speed flow fields compared to output from simulations performed with the mesoscale Weather Research and Forecasting model (WRF), but the models are in agreement with the inflow wind speed of the downstream wind farm. The double wind farm case is also simulated with the TurbOPark engineering wake model. Similar wake shapes are reproduced by TurbOPark but the model predicts a larger wind farm wake magnitude compared to RANS and WRF. TurbOPark predicts much better results when its ground model is switched off and a wake expansion coefficient of 0.06 is used. The RANS-AD-AWF model is also validated with SCADA measurements in terms of wind farm shape; the model captures the trend of the measurements for a wide range of wind directions, although the SCADA measurements indicate more pronounced wind farm wake shapes for certain wind directions.

High-resolution simulations are conducted with the Weather Research and Forecasting Model to evaluate the sensitivity of wake effects and power production from two wind farm parameterizations [the commonly used Fitch scheme and the more recently developed Explicit Wake Parameterization (EWP)] to the resolution at which the model is applied. The simulations are conducted for a 9-month period for a domain encompassing much of the U.S. Midwest. The two horizontal resolutions considered are 4 km × 4 km and 2 km × 2 km grid cells, and the two vertical discretizations employ either 41 or 57 vertical layers (with the latter having double the number in the lowest 1 km). Higher wind speeds are observed close to the wind turbine hub height when a larger number of vertical layers are employed (12 in the lowest 200 m vs 6), which contributes to higher power production from both wind farm schemes. Differences in gross capacity factors for wind turbine power production from the two wind farm parameterizations and with resolution are most strongly manifest under stable conditions (i.e., at night). The spatial extent of wind farm wakes when defined as the area affected by velocity deficits near to wind turbine hub heights in excess of 2% of the simulation without wind turbines is considerably larger in simulations with the Fitch scheme. This spatial extent is generally reduced by increasing the horizontal resolution and/or increasing the number of vertical levels. These results have important applications to projections of expected annual energy production from new wind turbine arrays constructed in the wind shadow from existing wind farms.

Wind farm power production is known to be significantly affected by turbine wakes. When mesoscale numerical models are used to predict power production, the turbine wakes cannot be resolved directly because they are sub-grid features and therefore their effects need to be parameterized. Here we propose a new wind farm parameterization that is based on the Jensen model, a well-known analytical wake model that predicts the expansion and wind speed of an ideal wake. The Jensen parameterization is implemented and inserted into two commonly-used atmospheric numerical models: the Weather Research and Forecasting (WRF) Model and the Model for Prediction Across Scales (MPAS). In addition, the internal variability in wind speed and direction within a wind farm, the wind direction uncertainty, and the superposition of multiple wakes are taken into account with an innovative approach. The proposed approach and parameterization are tested against observational data at two offshore wind farms: Lillgrund (small in size and tightly spaced) and Anholt (large and widely spaced). Results indicate that power production is predicted more accurately with the Jensen than with the Fitch wind farm parameterization, which is the only one available in WRF. Power predictions with the Jensen parameterization are similar in WRF and MPAS. The sensitivity to grid resolution is small and the bias is generally low and negative. In conclusion, we recommend that the Jensen wind farm parameterization be used in the WRF and MPAS models, especially for coarse resolution, high turbine density, and wind directions aligned with the turbine columns.

Wind farm power production is known to be significantly affected by turbine wakes. When mesoscale numerical models are used to predict power production, the turbine wakes cannot be resolved directly because they are sub-grid features and therefore their effects need to be parameterized. Here we propose a new wind farm parameterization that is based on the Jensen model, a well-known analytical wake model that predicts the expansion and wind speed of an ideal wake. The Jensen parameterization is implemented and inserted into two commonly-used atmospheric numerical models: the Weather Research and Forecasting (WRF) Model and the Model for Prediction Across Scales (MPAS). In addition, the internal variability in wind speed and direction within a wind farm, the wind direction uncertainty, and the superposition of multiple wakes are taken into account with an innovative approach. The proposed approach and parameterization are tested against observational data at two offshore wind farms: Lillgrund (small in size and tightly spaced) and Anholt (large and widely spaced). Results indicate that power production is predicted more accurately with the Jensen than with the Fitch wind farm parameterization, which is the only one available in WRF. Power predictions with the Jensen parameterization are similar in WRF and MPAS. The sensitivity to grid resolution is small and the bias is generally low and negative. In conclusion, we recommend that the Jensen wind farm parameterization be used in the WRF and MPAS models, especially for coarse resolution, high turbine density, and wind directions aligned with the turbine columns.