Session L24: Energy: Wind Power Modeling

Optimal renewable microgrid energy dispatch in the presence of uncertainty wind speed forecasting

The stochastic nature of wind speed and other renewable energy resources (RES) requires robust approaches to the energy management of renewable energy microgrids (RES MG). In this study, enhancements to the control hierarchy of RES MGs are proposed by taking advantage of probabilistic forecasting of wind speed and solar radiation with machine and deep learning models as a way to introduce uncertainty to the tertiary operations of the system. Using uncertainty through prediction intervals provides improvements over the conventional deterministic approaches for long-term planning, resources allocation and risk management of electrical grids with RES MG participation. The formulated probabilistic economic dispatch problem also takes into consideration grid stability, reduced carbon emissions, operation-and-maintenance costs, and load uncertainty as part of the optimal cost solution. Uncertainty-based optimization techniques such as stochastic and chance-constrained programming are used to solve the optimization problem under different criteria. The results provide further motivation for the adoption of probabilistic forecasting of renewable energy sources in the long-term operations of RES MG for a more reliable and cost-effective clean energy grid.   

Analytical coupled Ekman and surface layer model for ABL flows and applications to wind farm modeling

We propose a physics-based analytical model for predicting wind velocities in conventionally neutral and stable atmospheric boundary layer (ABL) flows. Unlike traditional approaches like the Monin-Obukhov Similarity Theory, which focuses solely on modeling profiles of wind magnitude within the surface layer, our model offers improved predictions of both streamwise and wind veer velocity profiles across varying atmospheric conditions throughout the entire ABL height. The model also provides estimates of friction velocity and geostrophic wind velocities using a new self-consistent geostrophic drag law (GDL) that is derived by matching velocities in the inner and outer layers. To validate our model, we compare model results with Large Eddy Simulation (LES) data. The results demonstrate very good agreement over a wide range of stability conditions. Furthermore, we integrate the analytical velocity profiles into wind turbine wake models to effectively account for sheared wake structures resulting from wind veer in a thermally stably stratified ABL. We present comparisons of this enhanced analytical wake model with LES results.   

Effects of wind-farm blockage and gravity waves on large-scale wind farms operating in conventionally neutral boundary layers

Conventionally neutral boundary layers (CNBLs) often develop offshore as a result of land-sea transitions. These conditions have been studied in measurement campaigns and numerical simulations. However, the cost and complexity of wind-farm large-eddy simulations (LESs) in the presence of thermal stratification above the atmospheric boundary layer (ABL) have limited the number of studies to a handful of atmospheric states. To fulfil this gap, we perform 40 LESs of a large-scale wind farm which operates in various atmospheric states, including different capping inversion heights and strengths, as well as lapse rates in the free atmosphere. We use this suite of LESs to investigate wind-farm induced gravity-wave effects on farm efficiency and blockage, as well as the flow behaviour in and around the farm. The numerical domain length and width are determined by a sensitivity study which reveals that atmospheres with shallow ABLs necessitate a wider domain to limit artificial domain blockage. A tuned Rayleigh damping layer and a wave-free fringe region method are used to avoid spurious excitation of gravity waves. To distinguish between a case with hydrodynamic blockage only, a fully neutral reference case is also considered. These results are compared with cases that include hydrostatic blockage excited by gravity waves. We discuss in detail the dependence of gravity-wave excitation, flow fields, and wind-farm blockage on capping-inversion height, strength and free-atmosphere lapse rate. In all cases, an unfavourable and favourable pressure gradients are present in front and within the farm, respectively, with hydrostatic contributions arising from gravity waves at least an order of magnitude larger than hydrodynamic effects. We extend our analysis to the farm efficiencies, where we observe a strong negative correlation between unfavourable upstream pressure gradient and non-local efficiency, and a strong positive correlation between the favourable pressure drop in the farm and the wake efficiency. Using a simplified linear gravity-wave model, we formulate a scaling for the non-local to farm-efficiency ratio, which matches reasonably well with the LES results. Finally, we compare the LES results against several gravity-wave linear-theory models found in the literature, finding overall a very good agreement.   

Investigating aerodynamic coupling between wind shear and utility-scale wind turbines

Wind shear in the atmospheric boundary layer (ABL) affects wind turbine aerodynamics and power production. As utility-scale turbines extend farther into the ABL, they can be exposed to greater degrees of shear and more complex wind profiles, necessitating models that accurately capture the effect of shear on their aerodynamics. The commonly used rotor-equivalent wind speed model accounts for one-way coupling, where the effect of wind shear is captured through modifications to the rotor-averaged wind speed incident on the turbine. However, this model neglects aerodynamic feedback, where variations in wind shear affect rotor thrust and induction. In comparisons to utility-scale field observations of power production, we show that the rotor-equivalent model predictions are under-dispersive relative to the impacts of wind shear. Furthermore, the rotor-equivalent wind speed model does not account for the response of the turbine controller to the incident wind conditions, or the non-linear feedback produced by these interactions. Using field measurements, we demonstrate that utility-scale turbine control operation depends on the wind shear. To address this, we use a blade element model to study the aerodynamic effects of wind shear on airfoil performance and turbine power production. Among the questions we address are the effect of wind shear on azimuthal variations in axial induction and estimation of the rotor angular velocity in the absence of definitive supervisory control and data acquisition data.   

Effects of heterogeneous surface heat flux on wind turbine wakes in LES over a diurnal cycle

The atmospheric boundary layer (ABL) undergoes significant changes in its structure and dynamics throughout a typical diurnal cycle, which can consequently affect the behavior of wind turbine wakes in wind farms. These wakes can, in turn, alter convective heat transfer over the surface terrain, leading to two-way coupling and spatial heterogeneity in surface heat fluxes. In this work, we examine the effects of using different types of surface boundary conditions for the temperature field in Large Eddy Simulations (LES) of wind farms. The most commonly applied boundary conditions are imposed temperature or heat flux, but these approaches cannot account for the above-mentioned two-way coupling. To address this issue, we combine an LES wind farm model using a filtered actuator-line approach for the wind turbines with a simple local 1D soil heat conduction model. We compare the results related to the diurnal cycle and heterogenous thermal effects with those obtained using a prescribed heat flux or surface cooling rate instead of the 1D soil model.  In particular,  we investigate the influence of thermal effects on wake recovery behind the wind turbines. The generated datasets are planned to be shared publicly via an extension of the JHU turbulence database (JHTDB).   

Modeling the effects of swell and vertical wave motions in wall-modeled Large Eddy Simulation of offshore wind farms

Wind-driven waves provide feedback to the airflow by extracting energy and momentum from the wind. In addition, long-wavelength swell waves can be generated by remote storm events and impart energy and momentum to the wind. In Large Eddy Simulation, wave effects have been modeled through a single roughness length, or, more recently, through a form drag model that accounts for the wind-wave relative velocity and surface gradients. These models are implemented as strictly as drag and as surface parameterizations that do not include vertical motions of the wave surface. In this work, a recent wave drag model from Aiyer et al. (2023) is extended to model nonlocal waves, such as swell, through an empirical parameterization. The vertical wave motions are modeled using a transpiration-based wall model to incorporate surface kinematics in an equilibrium wall modeling framework. The effect of vertical wave motions on the phase-averaged wave fluctuation velocities and stresses is probed using LES. The model is applied to study flow past an offshore wind turbine, and the effects of the wave kinematics (transpiration effect) and the wave dynamics (wave drag) on the wind turbine wake are quantified.   

Examining the influence of upwide farm geometries on wind farm cluster wakes

The move towards large-scale deployment of offshore wind farms within the outer continental shelf of the U.S. Atlantic Coast raises many questions regarding how to maximally harness offshore wind energy resources. Two important elements in this regard are the arrangement of individual wind turbines in an array and the siting of entire wind farms with respect to each other. While much research has been devoted to understanding the former, relatively little has been learned about the latter. Understanding how wind farm wakes interact with each other in the realistic atmospheric conditions and geometric configurations of the Atlantic Coast is a crucial step towards optimizing our offshore wind energy systems. In this study, we employ high-fidelity, turbulence-resolving large eddy simulations to investigate factors that may impact the collective annual energy power production of offshore wind farms. Such factors include an individual array's geometry as well as intra-array parameters such as wind turbine spacing. Finally, we calculate the evolution of mechanical kinetic energy within and between farms to better understand the recovery mechanisms of long wakes as well as the potential impact on performance of downstream wind farms.   

Artificial Intelligence and High-Performance Computing in the Context of Particle-Laden Turbulent Flow and Wind Energy

Leading edge erosion on wind turbine blades is a significant issue that adversely affects the efficiency of wind turbines. Various airborne objects, such as ice, dust, insects, raindrops, and snowflakes, contribute to erosion, with the outer areas of the turbine blades being particularly susceptible due to higher speeds.

To investigate the behavior of particles causing erosion, this study investigates the fundamentals of turbulent particle-laden flow using experiments. Specifically, the stagnation area of the leading edge is measured, to observe particle behavior and erosion effects. A Lagrangian particle tracking technique is employed to gather data pertaining to the inertial particle dynamics and tracer particles, separately.

The study sheds light into the relationship between turbulence intensity, particle sizes, and deformation rates on leading edge erosion. An innovative approach involving deep learning-based models and high-performance computing to predict and model leading-edge erosion using the acquired dataset is proposed. The resulting predictive model can potentially be used to optimize blade surface materials and mitigate erosion effectively.    

Numerical Assessment of Dynamic Loads in Two-Blade and Three-Blade Wind Turbines under Varying Hurricane Categories and Power Generation.

Wind turbine blade configuration and number play a significant role in the dynamic loading and power generation. With the growing demand for wind energy all over the world, it’s important to analyze how wind turbines could withstand the hurricane prone areas specifically in the Caribbean and East coast of USA. This study aims to investigate the impact of wind turbine blade configuration and number on dynamic loading and power generation, particularly in hurricane-prone areas like the Caribbean and East coast of the USA. The research compares two-blade and three-blade wind turbine configurations under five distinct hurricane categories, maintaining a tip-speed ratio of 9. To analyze turbulent flow at the inlet, Fluid-Structure Interaction (FSI) and Unsteady Reynolds-Averaged Naiver-Stokes (URANS) simulations are employed, using a neutral atmospheric boundary layer (ABL) profile based on velocity measurements at a reference height of 61.5m. The main focus is to assess the risk of wind turbine failures caused by hurricanes' induced wind on the blades, comparing both configurations at normal and high-speed wind conditions.   

Quantifying Structural Reliability in Off-Shore Wind Turbines: Sampling, Instanton, and Sensitivity Analysis

This study focuses on assessing the structural reliability of off-shore wind turbines using the Betti model [1], a reduced-order approximation of the turbine’s dynamic response to stochastic wind and wave perturbations. Our primary objective is to use the computational efficiency of the reduced- order model to analyze the mechanical stress variability in different turbine components (ropes, tower, and blades) by investigating its probability distribution. We employ direct sampling of typical configurations and the instanton approach from statistical mechanics to identify extreme deviations. Additionally, we explore sensitivity of the bulk and tail regions of the stress probability distribution to the turbulence intensity and turbine control strategies. The findings from this research provide valuable insights into enhancing the structural reliability of off-shore wind turbines and optimizing their operational strategies.

[1] Giulio Betti, Marcello Farina, Giuseppe A. Guagliardi, Andrea Marzorati, and Riccardo Scattolini. Development of a Control-Oriented Model of Floating Wind Turbines. IEEE Transactions on Control Systems Technology, 22(1):69–82, January 2014.   

Capturing aerodynamic forces on wind turbine blades using a tip-corrected Actuator Line Method

The Actuator Line Method (ALM) is a technique used to simulate the aerodynamic performance of wind turbines in Large-Eddy Simulations (LES). It permits to capture the interaction and turbulence loading of wind turbine blades within turbulent atmospheric boundary layers. However, the approach has certain limitations, particularly when it comes to accurately capturing the radial loading distribution along the blades, especially towards the tip. This discrepancy arises because ALM represents the turbine blades as lines, and when coarse grid resolutions are used, the distance between the rotating line and the grid causes difficulties in realistically modelling the loads.

In large simulations of wind farms, this could constitute a serious limitation. To overcome this challenge, we have developed a simple tip-correction to the ALM approach that has a negligible computational cost and that consistently incorporates the Blade Element Momentum (BEM) tip-corrections within the ALM modelling framework. The validation of our tip-corrected ALM approach is made using BEM results for the NREL-5MW reference rotor under uniform inflow conditions. The comparison demonstrates that the results obtained from the tip-corrected ALM approach align with the BEM results that include tip corrections, as expected within the modelling framework. These findings indicate that our modified ALM approach effectively models tip-corrections even on coarser meshes, making it highly suitable for wind farm simulations.   

Effects of helical-shaped blades on wake characteristics and power production of vertical-axis wind turbines in finite-length wind farms

Recent experimental and numerical studies based on single-turbine tests have suggested that using helical-shaped blades on vertical-axis wind turbines (VAWTs) may induce beneficial effects to the wake flow characteristics, such as faster wind speed recovery and lower turbulent intensity. In this work, the large-eddy simulation (LES) method is utilized to further test the effects of helical-shaped blades on VAWTs in wind farm environments. In particular, two helical-bladed VAWTs (with opposite blade twist angles) and one straight-bladed VAWT are considered. Each VAWT type is tested in three different aligned finite-length array configurations with coarse, intermediate and tight turbine spacings. Statistical analyses of the flow field and VAWT performance are conducted based on the LES data of these nine simulation cases. The results indicate that using the helical-bladed VAWTs may help increase the power production rate in the fully developed flow region of the VAWTs array by about 3-7%. Compared with the straight-bladed VAWT, the helical-bladed VAWTs have smoother coverage of the azimuth rotation angle during the rotation, resulting in significant reductions of temporal fluctuations in power coefficient and structural bending moment.