Dynamic stall is a complex fluid dynamics problem that occurs on an airfoil during rapid, transient motion in which the angle of attack goes beyond the static stall angle. Since the instantaneous sectional aerodynamic loads may surpass the static values, dynamic stall events often dictate the operational load range in several systems, including wind energy machines. Typically the phenomenon of dynamic stall is modelled using semi-empirical or the so-called engineering approaches, derived from 2D wind tunnel tests. However, extrapolation to wind energy machines’ behaviour must be done carefully as real conditions of DS occurrence arise as a combination of complex, interacting phenomena, including 3D aerodynamic features (e.g. due to yaw misalignment and rotational augmentation) and also blade structural vibrations. ... Read more

This work presents an airfoil probabilistic design methodology for wind energy applications, using an analytical approach to estimate the angle-of-attack fluctuations produced by non-uniformities in the incoming wind field. The contemplated wind speed perturbations include wind shear, yaw misalignment and atmospheric turbulence, and several combinations of the perturbation sources are considered. A probabilistic design space mapping is carried out to evaluate which wind conditions occur more often in practice and how likely each specific combination of perturbation sources is to occur, for both an onshore and offshore scenario. The proposed probabilistic method and specifically the level of angle-of-attack fluctuations is verified by employing the aero-elastic simulation tool FAST for each case. Finally, the probabilistic approach is used to design airfoils sections employing the genetic multi-objective airfoil optimization tool Optiflow, where the probability of angle of attack fluctuations for a given scenario is used to prescribe the operational angle of attack range over which the airfoil performance is relevant. Results of the airfoil optimization for a 24% thick section are presented, illustrating the trends in foil geometry and aerodynamic performance for three different possible optimization objective functions. ... Read more

Offshore wind energy is emerging as a large contributor to installed renewable energy capacity. In order to continue the momentum of its development, the offshore wind industry is looking to continually lower the levelized cost of electricity (LCOE). One area being explored in an effort to lower the LCOE of offshore wind generation is the optimization of the wind farm layout. Many of the offshore wind farm layout designs that exist today are structured in a rectilinear form where turbines are spaced evenly along columns and rows. This research explores the economic advantages of removing rectilinear constraints and optimizing the positions of the individual turbines within an offshore wind farm. At the core of achieving the research objective was the development of a model that is capable of simulating an existing offshore wind farm by converting representative wind farm data into an LCOE. The positions of the turbines within the wind farm can be modified using an optimization framework with the intent to minimize the LCOE. The model comprised of the Jensen Wake Model, a hybrid cable layout heuristic and a cost scaling model. The wind farm layout was optimized using a genetic algorithm. The cost estimation model and optimization framework were applied into two case studies to analyze the results of the wind farm layout optimization of two wind farms, Horns Rev and Borssele. In both case studies the optimized layouts provided higher AEP, shorter intra-array collection cable lengths and ultimately a lower LCOE than the baseline rectilinear layouts. ... Read more

The accuracy of airfoil polar predictions is limited by the usage of imperfect turbulence models. Can machine-learning improve this situation? Will airfoil polars teach the effect of turbulence on skin-friction? We try to answer these questions by refining turbulence treatment in the Rfoil code: boundary layer closure relations are learned from airfoil polar data. Two turbulent closure relations, for skin friction and energy shape factor, are parametrized with a class-shape transformation. An experimental database is then used to define code inaccuracy measures that are minimized with an interior point gradient algorithm. Results show that airfoil polars contain exploitable information about turbulent phenomena. Inferred closures agree with direct numerical simulation results of skin friction and the new code predicts drag more accurately. Maximum lift remains under-predicted but Rfoil maintains its robustness and suitability for optimization of wind energy airfoils. ... Read more

We describe a probabilistic approach to design airfoils for wind energy applications. An analytical expression is derived for the probability of perturbations to the operational blade-section angle of attack. It includes the combined influence of wind shear, yaw-misalignment, and turbulence intensity. The theoretical fluctuations in angle of attack are validated against an aero-structural simulation of a 10 MW horizontal axis wind turbine, operating under different inflow conditions. Finally we incorporate the probabilistic approach into a multi-objective airfoil optimization problem, which is solved with a genetic algorithm. The results illustrate the compromise between airfoil performance for a specific angle of attack and robustness of airfoil performance over a large range of angle of attack fluctuations ... Read more

Unsteady loads are a major limiting factor for further upscaling of HAWTs considering the high costs associated to strict structural requirements. Alleviation of these unsteady loads on HAWT blades, e.g. using active flow control (AFC), is of high importance. In order to devise effective AFC methods, the unsteady loading sources need to be identified and their relative contribution to the load fluctuations experienced by blades needs to be quantified. The current study investigates the effects of various atmospheric and operational parameters on the fluctuations of α and C_L for a large HAWT. The investigated parameters include turbulence, wind shear, yawed inflow, tower shadow, gravity and rotational imbalances. The study uses the DTU's aeroelastic software HAWC2. The study identifies the individual and the aggregate effect of each source on the aforementioned fluctuations in order to distinguish the major contributing factors to unsteady loading. The quantification of contribution of each source on the total fatigue loads reveals >65% of flapwise fatigue loads is a result of turbulence while gravity results in >80% of edgewise fatigue loads. The extensive parametric study shows that the standard deviation of C_L is 0.25. The results support to design active load control systems by highlighting the magnitude of C_L and α variations experienced by HAWTs, and thus the dC_L that needs to be delivered by an AFC system. ... Read more

... Read more

This study describes a methodology for designing airfoils suitable to employ actuation in a wind energy environment. The novel airfoil sections are baptized wind energy actuated profiles (WAP). A genetic algorithm-based multi-objective airfoil optimizer is formulated by setting two cost functions: one cost function for wind energy performance and the other representing actuation suitability. The wind energy cost function compares the candidate airfoils' performance with 'reference' wind energy airfoils, considering a probabilistic approach to include the effects of turbulence and wind shear. The actuation suitability cost function is developed considering horizontal axis wind turbines active stall control, including two different control strategies designated by 'enhanced' and 'decreased' performance. Two different actuation types are considered, namely, boundary layer transpiration and dielectric barrier discharge plasma. Results show that using WAP airfoils provides much higher control efficiency than adding actuation on reference wind energy airfoils, without detrimental effects in non-actuated operation. The WAP sections yield an actuator employment efficiency that is two to four times larger than those obtained with reference wind energy airfoils, at equivalent wind energy performance. Regarding geometry, and compared with typical wind energy airfoils, WAP sections for decreased performance display an upper surface concave aft region, while for increased performance, a convex upper surface aft region is obtained. The present study emphasizes that there is much to gain in designing airfoils from the beginning to include actuation effects, especially compared with employing actuation on already existing airfoils. The results demonstrate the potential of including actuation effects in the airfoil design process, thus enabling novel horizontal axis wind turbines control strategies. ... Read more

Pumping-mode airborne wind energy systems (PM-AWE) consist of an airborne drone (UAV) that flies tethered to a ground station. The tethered UAV is expected to generate high lifts when reeling the tether out, descend with low force and damp gusts during take-off or landing events. These missions take place in unpredictable environmental conditions. Winds can be more or less turbulent, ground proximity prompts leading edge soiling and aerostructural effects induce shape deformations. Conventional aviation airfoils were not designed for these conditions but wing designers miss specialized alternatives. The need for specialized airborne wind energy airfoils was first identified by Venturato [1]. He redesigned the Clark Y airfoil with a genetic optimization approach based on a simple performance goal and a RANS CFD solver. This type of CFD model cannot capture important physical phenomena like laminar bubbles or turbulent transition but Venturato’s work had the groundbreaking merrit of raising awareness about specialized airfoils. Here we present a collection of airfoils designed for the needs of the airborne wind energy industry. The design exercise builds upon an established airfoil optimization framework with a proven track record in the conventional wind turbine industry [2,3,4,5]. The method is known to provide a broad coverage of the design space thanks to the use of a CST parametrization, a tuned version of the RFOIL viscous-inviscid solver and multi-objective genetic optimization algorithms. Candidate airfoils are assessed in terms of conflicting structural and aerodynamic goals. Pareto fronts quantify compromises between glide ratio, maximum lift, building height, stall harshness and resilience to leading edge soiling. Finally, desirable pitching moment characteristics are framed within the broader question of planform design, a question on which we hope to engage the audience in a lively discussion. ... Read more

Flow augmentation consists in modifying mass flow across the actuation plane of a rotor to enhance energy extraction or propulsive efficiency. The talk sketches the distinction between passive and active rotor augmentation strategies. Power coefficient trends are compared analytically while numerical results illustrate differences in flow topology. Rotors are stylized as actuator disks that exert homogeneous normal forces on the steady flow of inviscid fluids to highlight the distinctive features of each augmentation principle. Passive augmentation principles have been well documented because they guide the design of ducted, shrouded and diffuser-augmented wind turbines1-6. These axisymmetric bodies decrease average static pressures on the rotor plane to increase mass flux and power coefficient. Rotor-body interactions are dominated by conservative forces5,7: the bodies don’t exchange energy with the fluid but act as augmenting devices and affect global energy balance by changing rotor state. Virtual work arguments show that bodies exert streamwise forces4,6 that can be related with the power coefficient through the law of de Vries1,6. Active flow augmentation is a rather recent theoretical concept8. Its simplest energy extraction embodiment consists of an upstream actuator that accelerates flow onto a downstream actuator. This augmentation strategy is coined as active because the upstream actuator injects (spends) energy into the flow for the downstream actuator to extract (produce) energy from a greater mass flux than if it were alone. The interaction mechanism depends on the action of non-conservative forces and actuators interact exclusively through changes in total flow enthalpy when they are sufficiently far apart. No pressure interactions occur in this asymptotic case and a closed solution exists together with an analytical power coefficient law. Parallels can be drawn with wake ingestion propeller setups9 but no practical energy extraction realizations have been attempted yet. Passive and active flow augmentation concepts are different but we hope that parallels between them shed further light on the physics of energy extraction from ideal fluid flows. The communication concludes with a few reflections meant to trigger an open discussion about the implications and applicability of the discussed theories. ... Read more

The contribution of sustainable Wind Energy (WE) to the global energy scenario has been steadily increasing over the past decades. In the process, Horizontal Axis Wind Turbines (HAWT) became the most widespread and largest WE harvesting machines. Nevertheless, significant challenges still lie ahead of further expansion of HAWT, namely concerning systemrobustness and cost-of-energy(COE) competitiveness. This dissertation studies aHAWT design concept termed modern Active Stall Control (ASC). With this concept HAWT power regulation is achieved using flowcontrol actuators to trim the aerodynamic loads across the operational envelope. The underpinning idea is that as the aerodynamic loads are trimmed by flowcontrol actuatorswithout pitching the blades, the pitch system may be mitigated. In turn, this might lead to decreased failure-rates and down-time, and thus eventually present a more cost-effective solution than state-of-the art HAWTs. Going specifically into ASC, if aerodynamic load trimming is performed it is necessary to employ a flow control actuator. From different flow control actuator types, since the aim is to reduce the maintenance and operational costs of ASC machines, actuators with few mechanical parts become more interesting. As such the present research also focuses on the Alternating Current Dielectric Barrier Discharge (AC-DBD) plasma actuator, owing among other things to its absence of moving parts, negligible mass and virtually unlimited bandwidth of actuation. A preliminary study on the feasibility of active stall control to regulate HAWT power production in replacement of the pitch system is conducted. By taking the National Renewable Energy Laboratory 5MWturbine as reference, a simple blade element momentum code is used to assess the required actuation authority. Considering half of the blade span is equipped with actuators, the required change in the lift coefficient to regulate power is estimated in ¢Cl Æ 0.7. Concerning actuation technologies, three flow control devices are investigated, namely Boundary Layer Transpiration, Trailing Edge Jets and Dielectric Barrier Discharge plasma actuators. Results indicate the authority of the actuators considered is not sufficient to regulate power, since the change in the lift coefficient is not large enough. Especially if a pitch-controlledmachine is used as baseline case. Active stall control of Horizontal AxisWind Turbines appears feasible only if the rotor is re-designed from the start to incorporate active-stall devices. Regarding AC-DBD plasma actuators, three specific topics are investigated. The different studies aim at DBD performance characterization, namely at the influence of external flow on DBD plasma momentum transfer and on the frequency response of actuator flow region characteristic of DBD pulse operation. Both these topics are important to bridge the gap between academic-laboratory employment of DBD and large-size industrial applications. Finally regarding DBD plasma actuator modeling, a method is developed to describe plasma actuation effects in integral boundary layer formulation, and coupled to a viscous-inviscid panel code (similar to XFOIL), while an experimental campaign is carried to validate the predictions. The three DBD plasma studies are further described below. Addressing cross-talk effects between DBD plasma actuators and external flow, a study is carried out in which an actuator is positioned in a boundary layer operated in a range of free stream velocities from 0 to 60m/s, and tested both in counter-flow and co-flow forcing configurations. Electrical measurements and a CCD camera are used to characterize the DBD performance at different external flow speeds, while the actuator thrust is measured using a sensitive load cell. Results show the power consumption is constant for different flow velocities and actuator configurations, while the plasma light emission is constant for co-flow forcing but increases with counter-flow forcing for increasing free stream velocities. The measured force is constant for free stream velocities larger than 20m/s, with same magnitude and opposite direction for the counter-flow and co-flow configurations. In quiescent conditions the measured force is smaller due to the change in wall shear force by the induced wall-jet. In addition to the experimental study, an analytical model is presented to estimate the influence of external flow on the actuator force. It is based on conservation of momentum through the ion-neutral collisional process while including the contribution of the wall shear force. Model results compare well with experimental data at different external flow velocities, while extrapolation to larger velocities shows variation in actuator thrust of at least 10% for external speedU Æ 200m/s. Concerning the response of DBD actuator region flow to pulsed operation, a methodology is provided to derive the local frequency response of flow under actuation, in terms of the magnitude of actuator induced velocity perturbations. The method is applied to an AC- DBD plasma actuator but can be extended to other kinds of pulsed actuation. For the derivation, the actuator body force termis introduced in the Navier-Stokes equations, from which the flow is locally approximated with a linear-time-invariant (LTI) system. The proposed semi-phenomenologicalmodel includes the effect of both viscosity and external flow velocity, while providing a system response in the frequency domain. Experimental data is compared with analytical results for a typical DBD plasma actuator operating in quiescent flow and in a laminar boundary layer. Good agreement is obtained between analytical and experimental results for cases below the model validity threshold frequency. These results demonstrate an efficient yet simple approach towards prediction of the response of a convective flow to pulsed actuation. Future application of the methodology might include actuation scheduling design and optimization for different flow control scenarios. The third study specifically addressing DBD plasma actuators presents a methodology to model the effect of DBD plasma actuators on airfoil performance within the framework of a viscous-inviscid airfoil analysis code. The approach is valid for incompressible, turbulent flow applications. The effect of (plasma) body forces in the boundary layer is analyzed with a generalized form of the von Kármán integral boundary layer equations. The additional terms appearing in the von Kármán equations give rise to a new closure relation. The model is implemented in a viscous-inviscid airfoil analysis code and validated by carrying out an experimental study. PIV measurements are performed on an airfoil equipped with DBD plasma actuators over a range of Reynolds number and angle of attack combinations. Balance measurements are also collected to evaluate the lift and drag coefficients. Results show the proposed model captures the magnitude of the variation in IBL parameters from DBD actuation. Additionally the magnitude of the lift coefficient variations (¢Cl ) induced by plasma actuation is reasonably estimated. As such, this approach enables the design of airfoils specifically tailored for DBD plasma flow control. Transitioning into ASCrotor design, and building on the previously presented, a methodology is introduced for designing airfoils suitable to employ actuation in a wind energy environment. The novel airfoil sections are baptized WAP (Wind Energy Actuated Profiles). A genetic algorithm based multiobjective airfoil optimizer is formulated by setting two cost functions, one for wind energy performance and the other representing actuation suitability. The wind energy cost function considers ’reference’ wind energy airfoils while using a probabilistic approach to include the effects of turbulence and wind shear. The actuation suitability cost function is developed for HAWT active stall control, including two different control strategies designated by ’enhanced’ and ’decreased’ performance. Two different actuation types are considered, namely boundary layer transpiration and DBD plasma. Results show that using WAP airfoils provides much higher control efficiency than adding actuation on reference wind energy airfoils, without detrimental effects in non-actuated operation. The WAP sections yield an actuator employment efficiency that is 2 to 4 times larger than obtained with reference wind energy airfoils. Regarding geometry, WAP sections for decreased performance display an upper surface concave aft-region compared to typical wind energy ’reference’ airfoils,while retaining common sharp nose and S-tail (characteristic aft-loading) features. Results show there is much to gain in designing airfoils from the beginning to include actuation effects, especially compared to employing actuation on already existing airfoils, which ultimately might pave theway for novelHAWT control strategies. Finally addressing the complete rotor planform design, an optimization study tailors a HAWT rotor to ASC operation, in a aero-structural-servo formulation. The study considers the aerodynamic and structural loads are in static equilibrium, and as such no unsteady effects are taken into account. The optimization includes planformgeometry design but also actuation scheduling, rated rotational speed and spanwise laminate skin thickness. Results show that, compared to variable-pitch turbines, ASC planform displays increased chord at inboard stations with decreased twist angle towards the tip, resulting in increased AOA. Actuation is employed to trim the (static) loads across the operational wind speed envelope and hence reduce load overshoots and associated costs. Comparing with state-of-the-art pitch machines, the expected COE of the ASC rotor does not indicate a significant decrease, but appears to be at least competitive with pitch-controlled HAWTs if the pitch system is effectively mitigated. Additionally, and though not explicitly considered in the present work, it is foreseen ASC might become interesting if the actuation system allows for further OM cost reduction via fatigue load-alleviation, since the actuation trimming load system is anyhow included in an ASC machine. ... Read more

The article seeks to unify the treatment of conservative force interactions between axi-symmetric bodies and actuators in inviscid ow. Applications include the study of hub interference, di_user augmented wind turbines and boundary layer ingestion propeller con_gurations. The conservation equations are integrated over in_nitesimal streamtubes to obtain an exact momentum model contemplating the interaction between an actuator and a nearby body. No assumptions on the shape or topology of the body are made besides (axi)symmetry. Laws are derived for the thrust coe_cient, power coe_cient and propulsive e_ciency. The proposed methodology is articulated with previous e_orts and validated against the numerical predictions of a planar vorticity equation solver. Very good agreement is obtained between the analytical and numerical methods ... Read more

There are many ways to learn from data. Our first experiment consisted in reproducing the way aerodynamicists work [2] with a genetic optimizer. The data pool was too narrow and asymptotic tendencies were unreliable. Our 2nd Experiment, a simple version of [4], had a virtually unlimited data pool and used neural networks. Results were better, but computationally expensive. Data assimilation approaches used in EO [ 7] could yield better results.. ... Read more