Session H15: Energy: Wind Power II

A Graph Theory Based Approach to Modeling Dynamic Wind Turbine Yaw

Yaw control has shown great potential as a method for increasing the power output of wind farms but there have been relatively few efforts to model the dynamic behavior of an array of dynamically yawing turbines. When a turbine is dynamically yawed, the wake created by that turbine is deflected and its shape changes, which also affects the dynamics of turbines that encounter the changed wake downstream.  To capture these dynamics, we model the wind farm as a graph in which each turbine is a node and the dynamic changes in wake interactions between turbines within the farm are represented through time-varying edge weights.  These edge weights are represented using a normalized velocity deficit coefficient describing the individual velocity deficits between each pair of turbines.  The deficits are then superposed linearly to find the local velocity at each turbine, according to the relationships defined by the graph.  This enables us to estimate the time-varying effect of dynamic yawing at each turbine and to determine the impacts on the overall farm power output. As a first step, the model is validated through comparisons to statistically steady state results from large eddy simulations.    

The Impact of Atmospheric Stability and Wake Turbulence on the Wind Turbine Blade Aerodynamics

Significant challenges remain in understanding wind turbine performance within the turbulent atmospheric boundary layer (ABL) and the wakes of upstream turbines. To help address these points, large-eddy simulations of the convective ABL under a range of stability states and inflow velocities are computed with two IEA-15 megawatt turbines separated by seven rotor diameters in a 5 km by 5 km by 1 km domain using the AMR-Wind solver from NREL. A comparison is made between the length- and time-scales encountered by the turbine in the global reference frame and the relative or local scales encountered by the rotating blade as a function of the ABL state and the rotor span. Since the ABL turbulence contains length scales much larger than the blade chord length; it has often been treated as a quasi-steady change in the local blade angle of attack and relative velocity. However, recent research from Vita et al. (2020) suggests that turbulence effects are non-negligible on blade aerodynamic performance even at integral length scales up to 3 times the local chord length. This study aims to elucidate these differences as a function of the ABL state, to better isolate the quasi-steady, unsteady, and turbulent flow conditions imposed on a wind turbine blade.   

Mass Consistent, Analytical Near Wake Models for Wind Turbines

A new three-dimensional, analytical wake model is presented which includes improved wake description near the rotor plane when compared to existing models. Wake structures, including the momentum deficit, hub jet, wake rotation, and flow acceleration around the rotor, are economically described with Ricker wavelets and Gaussian functions. These physics are included through the addition of free parameters, balancing the total degrees of freedom with model fidelity. Large-Eddy Simulations (LES) are used as reference data to assemble the model. Implementation into the FLOw Redirection and Induction in Steady State (FLORIS) wake modeling framework provides opportunity the perform model validation within a wind farm setting. Analytical modeling of each component of flow allows for mass conservation to be assessed.   

Parameter Exploration of a Two-Turbine Array Under Advanced Control Strategies

Cross-flow turbines, also known as vertical-axis turbines, use blades that rotate about an axis perpendicular to the incoming flow to convert the kinetic energy in a moving fluid into mechanical energy. Arrays of cross-flow turbines under advanced control strategies have been shown to outperform equivalent turbines in isolation by up to 30%. The array performance is dependent on a high-dimensional parameters space. This parametric dependence is challenging to characterize, as traditional uniform sampling quickly becomes prohibitively expensive. In this work, we implement a real-time active learning strategy, based on Gaussian Process Regression, to accurately and efficiently sample and model the high-dimensional parameter space of turbines operating in a recirculating water channel.  This model-based, hardware-in-the loop experimental approach results in a surrogate model that may be used for future optimization and control efforts.    

On the effects of blade aeroelasticity in control co-design of large offshore wind turbines

To increase the competitiveness of wind energy against conventional energy sources, proponents have sought to reduce the Levelized cost of energy. An inclination toward larger rotors to increase the power production may halt this progress due to an increase in maintenance costs that it may require. Advanced control systems such as the individual pitch control (IPC) seek to reduce fatigue loads on the rotor, reduce maintenance costs, and extend their life span. However, as blades reach lengths over 100 meters, deflection and inertial forces cannot be ignored. Traditionally, high-fidelity simulations have been considering turbine rotors to be rigid. To address this, we have performed Large-Eddy simulations of a wind turbine. The blades are modeled using the actuator surface model coupled with an aeroelastic model. Additionally, simulations with the IPC are compared against that of a rigid and an aeroelastic blade. Results have shown that the aeroelastic rotor reduces the fluctuating aerodynamic loads corresponding to the rotational frequency of the rotor.   

Validation of the ExaWind hybrid solver framework using field measurements of the NM-80 turbine under turbulent inflow

We present a hybrid solver framework for high-fidelity blade-resolved simulations of wind turbines in realistic atmospheric conditions. The challenge in this approach is to create a solver framework that can efficiently simulate both the flow around the complex blade geometry as well capture the wake formation and interaction with the atmosphere. We coupled the CFD solver Nalu-wind capable of resolving the turbine geometry with the block-structured background solver AMR-wind using overset technology from TIOGA to achieve this capability. The hybrid solver framework will allow us to simulate wind energy flows across a range of length scales that are eight orders of magnitude apart from 10 microns in the blade boundary layer to 1 km in the atmospheric boundary layer (ABL). The new hybrid solver was used to perform a comparison of the aerodynamic and structural loads under turbulent inflow conditions for the blade-resolved simulations of the NM80 rotor (IEA Wind Task 29 benchmark case), for which turbulence is generated in two different ways: 1. Synthetic turbulence from a Mann model introduced as source terms within the computational domain; 2. A full precursor Large-Eddy simulation of the ABL.   

Numerical study of turbulent wake flows behind a helical-blade vertical axis wind turbine

Vertical axis wind turbine (VAWT) has been widely used as a renewable energy harvesting device. The turbulent wake flow characteristics behind VAWTs with straight blades have been well studied over the past decades, which have provided valuable insights to help improve their performance in energy harvesting when used individually or in an array configuration. In comparison, there have been much fewer studies on the wake flow dynamics of VAWTs with helical-shaped blades. In this talk, the characteristics of turbulent wake flow behind the helical-blade and straight-blade VAWTs are studied using a large-eddy simulation (LES) model of wind turbulence combined with an actuator-line model of VAWTs. In particular, small helical and straight 5-blade VAWTs rotating at a low tip-speed ratio (~0.3) are considered, which are suitable for small-scale residential applications. The simulation results show that the wake flow behind the helical-blade VAWT exhibits considerable three-dimensional structures than that behind the straight-blade VAWT, resulting in noticeable differences in the turbulent flow characteristics and the wake speed recovery.   

Multi-fidelity deep learning for wake modeling of wind turbines

Wake prediction of wind turbines is one of the challenging problems in wind farms due to due to the complex unsteady nature of interactions of turbine wake with other wakes as well as atmospheric turbulence. The engineering wake model should be sufficiently accurate and computationally cheaper to be employed for tasks like wind farm layout optimization and wind farm controls. In this talk, we explore an application of a composite neural network framework to learn the wake model from large samples of low-fidelity data along with very few samples of high-fidelity data. The composite framework consists of a neural network to learn the low-fidelity data coupled with two neural networks to learn the linear and nonlinear correlation between low and high fidelity data. In particular, we train a composite neural network using data generated from hierarchies of physical models to predict the three-dimensional velocity field in turbine wakes. The prediction from the composite neural network matches well with the high-fidelity data compared to a neural network trained solely using the high-fidelity data. This works opens up possibilities for data-efficient construction of surrogate models for wake prediction that can be utilized to study the influence of wind speed, yaw angles, and layout configuration on wind farm power production.

           

On the effect of atmospheric stability on the efficacy of wind turbine blade pitch control strategies

Upscaling of wind turbines has been the general tendency to increase their power production. However, increasing the dimensions of turbines poses new engineering challenges to maintain their structural integrity.  Cyclic loads due to turbulence, gravity, and wind shear are detrimental for the blades. Atmospheric flow stratification aggravates the magnitude of these loads as it causes an increase in wind shear, often accompanied by veering in wind direction with height. In this work, we delve into using the individual pitch control (IPC) of the blades for load reduction due to flow heterogeneity at the rotor. Large-eddy simulations of a turbine in stable and neutral atmospheric flow have been performed. The turbine blades and nacelle have been modeled with the actuator surface model. LES results using the baseline controller and IPC were compared to investigate their effects on the blades' bending moment fluctuations. Our study shows a larger magnitude of cyclic loads due to high flow heterogeneity impinging the rotor for stable atmospheric conditions than in neutral. In addition, the IPC reduces the amplitude of these cyclic loads.   

Airfoil selection for a small-scale vertical axis wind turbine; a numerical study

Airfoil selection is one of the most important criteria in the overall performance of Vertical-Axis Wind Turbines. In this paper, we present methods for the selection of an airfoil for a small-scale vertical axis wind turbine to operate at relatively low tip-speed ratios. With the application of Computational Fluid Dynamics, the aerodynamic performance of the selected airfoil is then compared to the commonly used NACA 0018 airfoil. The selection is based on the effect of the tangential force coefficient over a specific range of angles of attack. A code was developed using MATLAB software to use NACA four-digit algorithm to generate the geometry of the airfoils. Using this algorithm, over 2000 airfoil geometries were generated and analyzed. The tangential force coefficient is a lift and drag force coefficient product at a defined angle of attack. Xfoil V6.99 was implemented in the code to predict drag and force coefficients. These values are then compared to the NACA 0018 airfoil. The airfoil that yields the highest tangential force coefficient at a defined range of angles of attack is chosen. The CFD analysis shows that the power coefficient of the turbine equipped with the selected airfoil exceeds that of the one equipped with NACA 0018 for a lower rotational rate.