Wind Tunnel Testing Of Scaled Wind Turbine Models Beyond .
Wind tunnel testing of scaled wind turbine models:Beyond aerodynamicsCarlo L. Bottasso a,b,n, Filippo Campagnolo b, Vlaho Petrović caWind Energy Institute, Technische Universität München, D-85748 Garching b. München, GermanyDepartment of Aerospace Science and Technology, Politecnico di Milano, Milano, Via La Masa 34, 20156, ItalycFaculty of Electrical Engineering and Computing, University of Zagreb, Unska 3, 10000 Zagreb, CroatiabArticle history:Received 8 July 2013Received in revised form14 January 2014Accepted 26 January 2014Available online 22 February 20141. Introduction and motivationThe understanding and simulation of the wind energy conversion process for single wind turbines and wind farms requires theability to model multiple complex interacting physical processestaking place at diverse spatial and temporal scales. Clearly, theability to effectively design wind energy systems ultimately relies,apart from an appropriate knowledge of the physics, on the fidelityto reality of the mathematical models used in simulations. Consequently, there is a need to validate such models and to calibratetheir parameters so as to maximize their accuracy.Validation and calibration can be performed with the help ofexperimental observations, conducted either on the full system oron its sub-components. When looking at the full wind turbine orwind farm system, testing and measurements conducted in the field,although invaluable, present some hurdles. First, it is usually difficultto have complete and accurate knowledge of the environmentaltesting conditions, which by the way cannot in general be controlled,and, secondly, costs and testing time are often quite relevant.nCorresponding author at: Wind Energy Institute, Technische Universität München,D-85748 Garching b. München, Germany. Tel.: þ 49 (0) 89 289 16680.E-mail address: carlo.bottasso@tum.de (C.L. Bottasso).To complement, support and, when possible, replace fieldtesting, one can resort to the use of scaled models. In such testingconditions it is usually impossible to exactly match all relevantphysics due to limitations of the scaling conditions, because of thefrequent impossibility of assuring the same full scale and scaledvalues for all non-dimensional parameters. On the other hand, onehas in general a better control and knowledge of the testingconditions, errors and disturbances. Furthermore, it may bepossible to perform measurements which might not be feasibleat full scale, and the testing typically incurs in much lower costs.Therefore, scaled testing does not replace simulation or fieldtesting, but works in synergy with both towards the goal ofdelivering validated and calibrated numerical simulation tools, aswell as an improved knowledge of the problem at hand.In the area of aerodynamics, wind tunnel testing of windturbine models has been reported by, among others, Oku et al.(1996), Hand et al. (2001), Vermeer et al. (2003), Snel et al. (2007)and Schepers and Snel (2007). These and similar studies haveproduced valuable information and measurements regarding theperformance of rotors and the behavior of airfoils, blades andwakes, helping not only with the understanding of the aerodynamical physical processes, but also with the validation and calibration of suitable mathematical models.
NomenclatureacginntrttfACDCFCLCPCTC L;αIJRTaTgVVrampβλμωspeed of soundblade chordgravitational accelerationcurrent intensityscale factortime ratioradial positiontimefault time instantrotor areadrag coefficientthrust coefficientlift coefficientpower coefficienttorque coefficientlift curve slopeblade flapping inertiaoptimization cost functionrotor radiusaerodynamic torquegenerator torquewind speedrated wind speedvector of model coefficientsvector of blade harmonicsvector of unknown parametersblade pitch angletip speed ratioair viscositypulsationNonetheless, aerodynamics is only one of the coupled phenomena that take place in the wind energy conversion process andwhose understanding is crucial for the most effective design andoperation of wind turbines. In fact, design loads on wind turbinesare dictated by transient phenomena, where the effects of inertialand elastic loads, as well as of the closed-loop control laws usedfor a variety of tasks onboard the machine, play a very major role.In this paper we propose to expand scaled wind tunnel testingbeyond the sole domain of aerodynamics. To this end, we describe anaeroelastically scaled model of a multi-MW wind turbine, featuringactive individual blade pitch and torque control. The model wasdesigned so as to deliver realistic aerodynamic performance, and canbe used for aerodynamic investigations, for example regardingwakes, their characteristics and their modeling. However, the modelwas also conceived for conducting experimental investigations onthe aeroservoelasticity of wind turbines in the controlled environment of a wind tunnel. As such, it can be used for studying themachine response in extreme operating conditions (e.g., emergencyshutdowns, operation at high yaw angles, response following failuresof onboard sub-systems, etc.), something that is difficult to do in thefield. The model can also support research on advanced pitch-torquecontrol laws, on load and wind observers, as well as a variety of otheraeroelastic investigations such as the study of the effects of loadsinduced within a wind farm by wake impingement caused byupstream wind turbines.The paper is organized according to the following plan. Section 2describes the scaled wind turbine model characteristics. At first,Section 2.1 describes the wind tunnel of the Politecnico di Milanowhere the model is typically operated. Next, Section 2.2 states thedesign requirements that stem from the diverse non-standardapplications that need to be supported by the model, whileϕρτΩFrLoMaReð ÞTð ÞIPð ÞOPð ÞMð ÞPð Þ1cð Þ1sð Þfa ð Þ ð Þð ÞADCBEMCANCFDDLCEOGFBGFEMIPCLESPIDRANSTSRwind misalignment angleair densitynon-dimensional timerotor angular speedFroude numberLock numberMach numberReynolds numbertransposerotor in-plane componentrotor out-of-plane componentquantity pertaining to the scaled modelquantity pertaining to the physical full scale systemfirst cosine harmonicfirst sine harmonicfore-aft componentexperimentally measured quantityquantity obtained by simulationderivative w.r.t. time, d dtactuator duty cycleblade element momentumcontroller area networkcomputational fluid dynamicsdynamic load caseextreme operating gustfiber Bragg gratingfinite element methodindividual pitch controllarge eddy simulationproportional integral derivativeReynolds-averaged Navier–Stokestip speed ratioSection 2.3 formulates the scaling laws. The general configurationof the model is given in Section 2.4, followed by Section 2.5 thatdescribes the aerodynamic design, while Sections 2.6–2.8 describesensors and the pitch and torque systems. The description of themodel is complemented by Section 2.9 that discusses the real-timecontrol and model management system, Section 2.10 that definessupport tools that were designed for the testing, calibration andmaintenance of the models and of their principal sub-components,and finally Section 2.11 that describes a comprehensive aeroservoelastic simulation environment of the experimental facility. Next, anumber of non-aerodynamic and non-standard applications arepresented in Section 3. A wind misalignment estimator, used insupport of active yaw control, is validated in Section 3.1. Next,Section 3.2 describes the optimization of emergency shutdownmaneuvers, including the calibration of a suitable mathematicalmodel. Finally, Section 3.3 describes active control applications,focusing on regulation in wake interference conditions, as well ashigher harmonic individual blade pitch control. Conclusions and anoutlook to future work end the paper at Section 4.2. Scaled wind turbine model2.1. The Politecnico di Milano wind tunnelThe wind tunnel of the Politenico di Milano, which was usedfor all tests carried out during this project, is a closed-returnconfiguration facility arranged in a vertical layout with two testrooms in the loop, as depicted in Fig. 1.The wind tunnel features 14 driving fans for a total installedpower of 1.4 MW, and has two test sections.
Civil t est sect ion:Dim ensions: 13.8m x 3.8mMax im um wind speed:14m/ secTurbulence intensity 2%Turbulence with t urbulence generators 25%13m t urntableAeronautical t est section:Dim ensions: 4m x 3.8mMax im um wind speed: 55m / secTurbulence intensity 0.1%Open- closed t est sectionFig. 1. Configuration of the wind tunnel of the Politecnico di Milano.The first one — mostly used for aeronautical engineering andsports-related applications — is a low turbulence ð o 0:1%Þ testsection, which is located at the lower level of the facility buildingand placed between the contraction cone and the diffuser. The lowturbulence test chamber has a maximum flow velocity of 55 m/s, across-sectional area of 4 m 3.84 m and a length of 6 m. Tests canbe conducted in both closed and open jet conditions.The second test section — primarily used for civil, environmental and wind power engineering applications — is a boundarylayer one, situated in the return duct at the upper level of thefacility building. The test chamber has a maximum flow velocity of14 m/s and a cross-sectional area of 13.84 m 3.84 m with alength of 36 m, which enables tests on relatively large modelswith low blockage effects. Atmospheric boundary layer conditionscan be simulated by the use of turbulence generators placed at thechamber inlet and of roughness elements located on the floor. Aturntable of 13 m of diameter allows for the model to be yawedwith respect to the incoming wind.The wind tunnel is equipped with a complete array of devicesfor measuring mechanical and fluid dynamics quantities, asneeded for conducting the various experiments.2.2. Design requirementsFor the present project, models were designed with the goal ofsupporting experimental observations not only in the field ofaerodynamics but also in the areas of aeroelasticity and control,for single and interacting wind turbines. The need to support thesediverse applications dictated a number of specific design requirements, which include A realistic energy conversion process enabled by good aero- dynamic performance at the airfoil and blade level, translatinginto reasonable aerodynamic loads and damping, as well aswakes of realistic geometry, velocity deficit and turbulenceintensity.Aeroelastic scaling, i.e. the ability to represent the mutualinteractions of aerodynamic, elastic and inertial forces, whichimplies the realization of a flexible machine with given specificstiffness and mass properties. For a wind turbine, this alsoimplies the ability to match the relative placement of theprincipal natural frequencies (of rotor, drive-train and tower)with respect to the harmonic per rev excitations, so that both the reference full scale and scaled machines have the sameCampbell diagram in the lowest frequency range.Individual blade pitch and torque control, so as to enable thetesting of modern control strategies, with appropriate bandwidthand a reasonable rendering of the principal dynamic effects ofservo actuators, mainly due to time delays and maximumattainable rates.A comprehensive onboard sensorization of the machine,giving the ability to measure loads, accelerations and positions (e.g., blade pitch and rotor azimuth), with sufficientaccuracy and bandwidth. The onboard sensorization is complemented by off-board automated traversing systems usedfor characterizing the flow field upstream and downstream ofthe machines.Dimensions of the models as large as possible, to avoid too lowa Reynolds number, but also to avoid excessive miniaturization,which would complicate the realization of active controlcapabilities and the need for a comprehensive sensorization.On the other hand, the rotor dimension should not causeexcessive blockage effects due to interference with the windtunnel walls, and should allow for the testing of at least twowind turbines in wake-interference conditions, capable ofrepresenting the typical couplings taking place within a windfarm between and among wind turbines.The scaled models were loosely based on the Vestas V90 windturbine, a 3 MW machine with a 90 m diameter rotor and a hubheight of about 80 m. Based on the minimum cross-sectionaldimensions of the wind tunnel test chambers, the scaled modelrotor diameter was chosen to be equal to 2 m, leading to ageometric scaling factor of 1/45. This dimension avoids excessiveblockage, and still allows for the testing in the boundary layer testsection of two wind turbines one in front of the other at distancesof about 4–5 rotor diameters.The models were complemented by a number of customdesigned devices used for their characterization, calibration andoff-line pre-testing outside of the wind tunnel. Furthermore, acomprehensive simulation environment of the experimental facility was developed, including aeroservoelastic multibody models ofthe wind turbine as well as CFD models of the coupled machine–wind tunnel environment. The simulation models supported thedesign and testing phases, and were validated and calibrated withthe use of experimental data gathered in the wind tunnel, asdescribed later on in this work.
2.3. Scaling lawsThe non-dimensional parameters governing the dynamicsof wind turbines can be derived with the help of Buckingham ΠTheorem (Buckingham, 1914; Barenblatt, 1996), and are represented by the tip-speed-ratio (TSR) λ ¼ ΩR V, where Ω is theangular speed of the rotor of radius R and V is the wind speed, theReynolds number Re ¼ ρVc μ, where ρ is the density of air, c is acharacteristic length and μ is the fluid viscosity, the Froudenumber Fr ¼ V 2 gR, where g is the acceleration of gravity, theMach number Ma ¼ V a, where a is the speed of sound, and theLock number Lo ¼ C L;α ρcR4 I, where C L;α is the slope of the liftcurve and I is the blade flapping inertia, as well as the none i ¼ ωi Ω, and non-dimensionaldimensional natural frequencies ωtime τ ¼ Ωt, with t indicating time.Scaling laws were here derived based on two criteria.The first requires the exact enforcement between the full scaleand scaled models of the same values of TSR (which amounts tohaving the same aerodynamic kinematics), of the same Lock number(which amounts to having the same ratio of aerodynamic to inertialforces), and of the same non-dimensional natural frequencies(which amounts to having the same relative placement of naturalfrequencies and harmonic excitations), or, in symbolsλM ¼ λP ;LoM ¼ LoP ;e iM ¼ ωe iP ;ωproblem:! 2tMk n22 ReM¼ minþþ nt ;min kntntReP t Pntð2Þwhere nt ¼ t M t P is the time ratio, n ¼ RM RP ¼ 1 45 is the lengthscale factor, so that the Reynolds ratio is ReM ReP ¼ n2 nt ; finally, kis a weight factor in the objective function, chosen in the presentcase to be equal to the value of 2. The solution to problem (2) isreadily found to be nt ¼ kn.By this definition of the scaling, one finds a mismatch in therepresentation of the Reynolds number equal to ReM ReP ¼ n k,which in the present case is equal to 1/90; the effects of such areduced Reynolds number on the model can be partially compensated, as discussed later on, by the use of special low-Reynoldsairfoils augmented by transition strips. Similarly, the Froude2mismatch is FrM FrP ¼ 1 ðnk Þ, i.e. 11.25 in this case; the effectsof a mismatched Froude are important only for very large windturbines, where gravity plays a relevant role. Finally, the Machmismatch is MaM MaP ¼ 1 k, i.e. 0.5 in this case; although compressibility effects do not play a role at the tip velocities of boththe full scale and scaled machines, it is interesting to notice thatMach scaling corresponds to the optimization-based one definedby Eq. (2) for the case k¼ 1.ð1Þwhere ð ÞM indicates quantities referred to the scaled model, and ð ÞPindicates quantities referred to the physical full scale one.The second criterion is to look for the best compromisebetween the contrasting requirements of limiting the Reynoldsmismatch ReM ReP , which is related to the quality of the aerodynamics of the scaled model, and the need to limit the speed-upof scaled time t M t P , in order to avoid an excessive increase in thecontrol bandwidth. In fact very high control frequencies, madenecessary by an excessively fast scaled time, would make itdifficult to test advanced control laws, which is one of the goalsof the project, since such laws might possibly imply a nonnegligible number of operations but would still need to beoperated in real-time on the model.The design of best compromise between these two requirements can then be expressed as the following minimization2.4. General configurationThe general arrangement of the scaled wind turbine model isshown in Fig. 2, while Fig. 3 shows a view of the rotor–nacelleassembly.The blades are mounted on the hub with two bearings, to avoidany flapwise or edgewise free-play, and house in their hollowroots zero-backlash pitch motors with built-in relative encoders.Aeroelastically scaled blades are realized using a machined Rohacell core, which ensures the right shape to the variable chord andtwist blade, and two spar caps made of unidirectional carbon fiber,whose width and thickness were optimized to achieve the rightstiffness. The blade surface is covered with a polymeric layer thatcloses the pores of the Rohacell core and ensures a smooth finish.Multiple load measurement points along the span are provided byNacelle andspinner covers6 dof balanceElectronic boardof blade strain gagesFig. 2. General arrangement of the scaled wind turbine model.
Up-tilt 6 degPitch actuatorwith backlashrecovering springPitch actuator control units,with position controlShaft strain gages andtheir conditioning board36 channel slipringConical spiralgearsCone 4 degTorque actuator housed in towertop, with planetary gearhead,and torque/speed controlTower topaccelerometerFig. 3. View of the rotor–nacelle assembly.21.5L1CFiber Bragg Grating (FBG) sensors (Hill and Meltz, 1997), integrated in the spar caps. Further details on the design, manufacturing and characterization of the aeroelastically scaled blades aregiven in Campagnolo (2013).Non-aeroelastic, i.e. rigid, blades were also manufactured usingcarbon fiber, and are shown in the figures. To measure blade rootbending, a machined steel component, inserted at the blade rootduring the curing process, is composed of four small bridges, sizedso as to exhibit sufficiently high load-induced strains to achievethe necessary level of accuracy of the strain gages, which aremounted on the same bridges.The shaft was also machined for a similar reason, and hostsstrain gages that measure the torsional and bending loads.Additionally, the shaft was sized so as to match the placement ofthe first torsional frequency of the reference machine drive-train.Two electronic boards, one in front (cf. Fig. 2) and one behind (cf.Fig. 3) the hub, provide for the power supply, conditioning and A/Dconversion of, respectively, the blade root and shaft strain gages.The shaft is mounted on two bearings, held by a rectangularcarrying box that constitutes the main structural member of thenacelle. Here an optical relative encoder measures the azimuthalposition of the shaft, while a triaxial accelerometer provides formeasurements of the acceleration in the nacelle, used for triggering emergency shutdown procedures and optionally for controlpurposes. A pair of conical spiral gears, with a 2/1 reduction ratio,connects the shaft with a motor that provides for the torque (andoptionally speed) control of the rotor. The torque motor is housedin the top of the tower; compressed air is blown in at the towerfoot and, traveling along the hollow tower, cools the torque motor,before escaping from a small hole in the back part of the tower top.Behind the nacelle carrying box, the three electronic controlboards of the pitch actuators are mounted on the shaft
Nonetheless, aerodynamics is only one of the coupled phenom-ena that take place in the wind energy conversion process and whose understanding is crucial for the most effective design and operation of wind turbines. In fact, design loads on wind turbines are dictated by transient phenomena, where the effects of inertial
Wind Tunnel Testing - Research Paper.doc Page 5 of 5 The Douglas DC-3, one of the most successful commercial aircraft ever built, required about 100 hours of wind tunnel testing. Wind tunnel time has been steadily increasing since: the Boeing 747 required over 1000 tunnel hours; the NASA Space Shuttle
Wind Tunnel Testing For Teachers: Teacher Resource Lesson Goal The Wind Tunnel Testing lesson focuses on wind tunnel tests that engineers in many industries use to when developing products such as airplanes, cars, and buildings. Teams of students build their own mode
Tunable compression of wind tunnel data 507 2. Wind-tunnel testing Fig. 1 shows a typical wind tunnel model instrumented with pressure taps. For each wind direction (or, equivalently, the model being placed on a rotating table, for each orientation of the model with respect to the mean flow
Wind tunnel testing is not cheap, however. In addition to the cost of the wind tunnel itself, there is the requirement for the infrastructure to house the wind tunnel and the cost of electricity to run the tests. While a crude wind tunnel c
red wind/red wind xlr h50 t-15m l 35 mm red wind/red wind xlr h80 t-16m l 65 mm red wind/red wind xlr h105 t-17m l 90 mm racing speed xlr h80 t-19m l 74 mm profile rim female valve adapter (option) red wind/red wind xlr h50 t-15f l 37 mm red wind/red wind xlr h80 t-16f l 67 mm red wind/red wind xlr h105 t-17f l 92 mm racing speed .
WIND TUNNEL TESTING: - 2 - Alan G. Davenport Wind Engineering Group A GENERAL OUTLINE 2 THE MODELLING OF THE SITE AND THE WIND 2.1 General The basic tool used is the Laboratory's Boundary Layer Wind Tunnel. This wind tunnel is designed with a very long test section, which allows exten
covers wind tunnel testing, rather than the wind tunnel procedure, of which wind tunnel testing is only a part. Improvements in the methods for determining wind loads on buildings and other struc-tures can eliminate the underestimation of wind effects on some types of buildings or re-duce unnecessary cos
simulation (LES) and validation with wind tunnel data obtained from RWDI USA LLC, Miramar FL (Dagnew et al. 2009). 2 GENERAL OUTLINE OF WIND TUNNEL TESTING . The wind tunnel testing facility has a testing section of 8.5ft wide by 7ft tall. The floor has me-chanical systems which monitor the d
