NUMERICAL PREDICTION OF HORIZONTAL AXIS WIND TURBINE FLOW

Natalino Mandas Francesco Cambuli Carlo Enrico Carcangiu mandas@dimeca.unica.it cambuli@dimeca.unica.it carcangiu@dimeca.unica.it Professor, DIMeCa Research Assistant, DIMeCa PhD Student, DIMeCa

University of Cagliari, Department of Mechanical Engineering (DIMeCa) Piazza dArmi, 1, 09123, Cagliari (Italy) Tel. (+39) 070 675 5712 Fax. (+39) 070 675 5717- http://dimeca.unica.it

SUMMARY The aerodynamics of HAWT are investigated using a commercial CFD code. RANS equations are used to solve the 3-D turbulent-steady incompressible flow, using the Spalart-Allmaras and the k- SST turbulence models for closure. Starting from the specifications of an existing middle-sized turbine, the classical Blade Element Momentum (BEM) method is adopted for the design of the rotor. The active part of the blade is extended to the hub, following the design tendencies of modern wind turbines. The computational domain is discretized with a structured grid of near 1.5 million of volumes, that is locally refined in order to resolve the wall boundary layer on blade surface and coarsened in other parts of the dominion, e.g. the wake region, taking care of the different geometric scale involved by the problem. All the computations are carried out for a turbine with both a classical and an innovative high performance blade geometry. The predicted values of the power generated are found to be in good agreement with those calculated with BEM method and the results also show the considerable increment of power obtained by the innovative design. Furthermore the study yields detailed information about basic aerodynamics, like axial interference factor distribution. Finally, the near and far wake are evaluated and some features of the root and tip vortices are shown. Keywords: CFD, Fluent, Blade Element Momentum (BEM), aerodynamics, blade design, wind power, wake, vortex.

1. INTRODUCTION In the energy sector Italy and many others European countries are strongly dependent on fuel imported by foreign countries and, consequently, on their inevitable depletion. Moreover, in June 2002, the Italian parliament ratified the Kyoto protocol on Climate Changes, formally taking account of environmental impact of burning fossils fuels. Thus, energy policy has confirmed the improvement of the environmental sustainability of energy as a primary objective, also through increasing use of renewable sources [1]. Wind energy is a low density source of power. To make wind power economically feasible, it is important to maximize the efficiency of converting wind energy into mechanical energy. Of all the different aspects involved, rotor aerodynamics is a key determinant for achieving this goal. In addition, the ability to predict the downstream wake from a wind turbine is a significant factor for determining the interactions between turbines. Research work conducted in this area has brought to a substantial improvement in the overall efficiency of the conversion process, with the result that the capital costs of installing wind power can now compete effectively with other energy sources. Three approaches are available to analyze the flow around and downstream of a wind turbine: field testing, which provides accurate results but is highly complex and expensive; analytical and semi-empirical models, which adopt simplifying assumptions and are thus not universally reliable; and CFD, which offers the best alternative to direct measurements [2]. Today, industrial rotor design codes are still based on BEM [2, 3, 4]. Nevertheless, in the last decade the opinion has been advanced that aerodynamic modelling of HAWT rotors by means of the conventional engineering methods has reached a point where no further improvement can be expected without a full understanding of the flow physics [5]. Thus the last years have seen the rise of numerical studies on all HAWT aerodynamics features, performed on many different levels, ranging from BEM methods integrated by CFD calculations to full 3D Navier-Stokes models.

Many authors use the generalized Actuator Disk Method, that represents roughly an extension of BEM method, integrated in a Euler or N-S frame [6, 7]. To overcome this main limitation, i.e. the forces are distributed evenly along the actuator disk in the azimuthal direction, 3D N-S solver has been combined with the so-called Actuator Line Technique, in which the loading is distributed along lines representing the blade forces [3, 4, 7]. In the few past years, Sankar and co-workers [8, 9, 10] developed a hybrid Navier-Stokes/Full-Potential/Free Wake Method, mainly for predicting 3D viscous flow over helicopter rotors and then extended it to deal with the HAWT flow field. The computational domain is divided in different region, each one solved by the proper approach: N-S solution near the blades, potential flow representation on outer field and a collection of vortex methods for modelling the vorticity field. Full three-dimensional computations employing the Reynolds-averaged Navier-Stokes (RANS) equations have been carried out by Kang and Hirsh [5], focused on 3-D effects on a HAWT, using the EURANUS/Turbo solver with either algebraic and k- turbulence models for closure; Ris and DTU carried out several numerical investigations on HAWT using their Navier-Stokes solver EllipSys2D/3D, dealing with overall performances and design of rotors and blade sections [3, 11, 12], extreme operation conditions [13], tip shape [14]. At the Department of Mechanical Engineering of the University of Cagliari (DIMeCa) the commercial CFD code Fluent 6.2 is used to perform a detailed analysis of HAWT flow [15, 16]. The steady flow field around an isolated rotor of a middle-sized HAWT is predicted in a non-inertial reference frame, using both the Spalart-Allmaras and the k- SST (Menter, 1993) turbulence models for closure, and specifying a constant axial wind velocity at the inlet. The classical Blade Element Momentum (BEM) method is adopted for the design of the turbine rotor, extending the active part of the blade close to the hub. This blade region is aerodynamically redesigned following the tendencies of modern wind turbines [17]. Several 2D and 3D simulations were carried out to yield information on the different aspects involved by this problem, ranging from performance calculations to wake development. The paper is organized as follows. In the next paragraph the mathematical and the numerical model are presented. The third paragraph deals with the geometry of the rotor, putting the focus on the hub-region. In the fourth paragraph the features of the computational discretization are illustrated, together with the mesh generation process. The fifth paragraph includes all the results, arranged in four sections: overall performances predicted by CFD and calculated by BEM method; new and classical rotor comparison; rotor aerodynamics investigations; near and far wake analysis. 2. MATHEMATICAL MODEL The mathematical model includes Continuity and Momentum equations. These equations are solved in a single Moving Reference Frame (MRF) attached to the rotor blades, with the assumption of incompressible and turbulent-steady flow. Basically, they are the incompressible steady-RANS classical equations [18]

Continuity 0ii

ux

= (1)

Momentum ( )( ) ' 'i ji j ijj i j

pu u u ux x x

= + (2)

The one-equation Spalart-Allmaras model with standard wall functions (y+ 30) is chosen for turbulence closure, because it represents a good compromise between accuracy and computing costs [19]

( ) ( ) ( )2

21

i v b v vi v j j j

v vv v u G v C Y St x x x x

+ = + + + +

(3)

Equation (3) concerns the transport variable , related to the eddy viscosity as follows t vv f = (4) The viscous damping factor fv is expressed by

3

3 31

fC

= + (5)

Where relates the transport variable and the molecular viscosity

=

(6)

For some calculations the k- SST eddy viscosity model of Menter is used, yielding nearly analogous results. The system of non-linear equations is solved by a segregated approach in all simulations. As the code solves the incompressible flow equations, no equation of state exist for the pressure, and the SIMPLE algorithm is used to enforce the pressure-velocity coupling. Second order and QUICK spatial discretization schemes are used for continuity and momentum equations. 3. GEOMETRY OF THE TURBINE The classical Blade Element Momentum (BEM) method is adopted for the design of the turbine rotor, using the specifications for an existent HAWT, found in literature [20]. This is the Nordtank 41/500 turbine, equipped with three LM 19.1 blades, which has a rotor diameter of 41 m and 500 kW of rated power. The turbine is of fixed pitch type and, thus, stall regulated. The real blade sections consist of NACA 63-4xx and FFA-W3-xxx series airfoils, but in the BEM method computations, based on a work previously done at DIMeCa [21], only NACA 63-4xx profiles data are used [22]. In the last years, improved performances have been declared for some commercial turbines with an innovative blade design, in which the aerodynamic active part of the blade is extended to the hub [17]. Hence, we perform the computations on a HAWT with this innovative aerodynamic design tendencies (fig. 1.a, b) and on a classical shaped turbine, in which the inner part of the rotor is configured only on structural issues (fig. 1.c, d).

a) b)

c) d)

Figure 1. Innovative rotor design (a, b); traditional rotor design (c, d).

4. COMPUTATIONAL GRID AND DOMAIN The wind turbine tower and the ground are not included in the flow model and a uniform wind speed profile is assumed at the entrance of the domain. To generate the volume mesh for the three bladed rotor, the 120 degrees periodicity of the rotor is exploited by only meshing the volume around one blade. The remaining two blades are included in the computations through the use of periodic boundary conditions. The pre-processor GAMBIT is used to build an hexahedral mesh of approximately 1.5 million volumes. Around the blade the grid is H-shaped (Figure 2) and it is optimized by means of 2-D simulations, to correctly resolve the boundary layer.

Figure 2. H-mesh around a blade section. Figure 3. Modular mesh generation. The computational domain used is conical shaped, extending in the axial direction roughly 5 diameters upstream and 10 diameters downstream of the rotor (Figure 4). In the plane of the rotor, the domain diameter is five times that of the rotor. In figure 4 are also shown the boundary conditions imposed: at the inlet face and at the lateral boundary undisturbed uniform wind velocity and turbulence are fixed; static pressure is set at the outlet; no-slip condition is selected for blade and nacelle surface. Mesh generation presents many difficulties, owing to the different range of geometric scales represented: length of the domain (600 m), size of the rotor (41 m), typical chord lengths (0.5 3 m), and boundary layer (10 mm). A representative image of the mesh generation process is shown in figure 3. In far wake predictions, a slightly different configuration of the computational domain is adopted, as will be illustrated in the results section.

Figure 4. Computational domain and boundary conditions. 5. RESULTS Performance: CFD vs BEM The commercial code FLUENT 6.2 is used in all the calculation presented. To validate the numerical model, the overall performances for the new design turbine are computed and compared to BEM calculations. Figure 5 shows the shaft mechanical power as a function of the wind velocity, and the corresponding power coefficient as a function of tip speed ratio. The CFD results are found to be in good agreement with those obtained using the BEM method for the undisturbed wind velocities tested, ranging from 6.8 to 12 m/s, i.e. for generally attached flow conditions. At higher wind speeds, however, there are significant discrepancies, probably related to a lack of the model in predicting stall effects. Different turbulence models and a grid refinement are required for investigating turbine performances in near and complete stall conditions.

Shaft Mechanical Power

0

100

200

300

400

500

0 4 8 12 16 20 24

V0 [m/s]

P [k

W]

Power Coefficient

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0 0,1 0,2 0,3 0,4

1 = (R/V0)-1

CP

CFD BEM

Figure 5. Mechanical Power P and Power Coefficient CP . BEM method and CFD calculations, for different wind velocities, V0, and tip speed ratios, .

New and Classic Rotor Design The classic rotor blade design is characterized by a cylindrical connection to the hub. This solution while overcoming structural limits, causes a lack of yielded power, as theoretically noticed several decades ago [23]. Only in recent years significant progress in new materials and structural design, with the contribution of accurate CFD studies, has led to a new rotor design, where the active part of the blade is extended to the hub, which has normally a greater diameter [17]. In order to assess the advantage of the new rotor configuration, a comparison is made between classic and innovative rotor design. Mechanical power found with CFD simulations are compared in Table 1, showing that innovative blade allows a 5 percent increment of the power coefficient.

Table 1. Comparison of performances. Traditional and innovative rotor design. V0 = 6.8 m/s. Rotor Design Mechanical Power [kW] Power Coefficient

Traditional (Fig. 1.a, b) 117.0 0.465 Innovative (Fig. 1.c, d) 122.7 0.487

Rotor Aerodynamics In the following some representative numerical results that characterize the aerodynamics of the rotor are shown. All the computation outputs are obtained for an undisturbed wind velocity of 6.8 m/s. Figure 6 shows the distribution of the axial interference factor, a = 1 - Va/V0 , in the rotor plane. A large change in a takes place across the blades, due to the effects from the bound vorticity located on the blade. CFD simulations give useful information to improve the BEM model. The tip geometry is obtained by simply truncating the blade with a cylindrical surface. Hence, the simulations reveals the formation of strong tip vortices (Figure 7), which in figure 6 appear as a localized region with negative a-values: here the axial velocity is higher than the inflow wind speed. Future work will be addressed to simulate different tip geometries, accurately shaped, that allow a reduction of fluid dynamic and acoustic tip effects. Finally, pathlines of figure 8 show the relative flow field near the blade root, where strong 3-D effects can also be found.

Figure 6. Axial interference factor, a. Figure 7. Pathlines at blade tip. Figure 8. Pathlines at blade root.

Near and Far Wake Wind turbine wakes have been a topic of research from the early start of renewed interest in wind energy utilisation, which dates to late 1970s. A distinct division can be made into near and far wake region. The near wake is taken as the region just behind the rotor, where the properties of the rotor flow can be noticeably discriminated, roughly up to one rotor diameter downstream. In this region the presence of the rotor is visible by the presence of blades and the related aerodynamic features, including stalled flow, 3-D effects and tip vortices. The near wake survey is focused on the physical process of energy conversion. Otherwise, the far wake is the region beyond the near wake, where the focus is put on the mutual influence of wind turbines in farm situations. The main research interest is to study how the far wake decays downstream, in order to estimate the effects produced in downstream turbines. These are lower velocity and higher turbulence intensity, that make the power production decrease and increase the unsteady loads. In the following some representative numerical results that characterize the aerodynamics of HAWT wakes are presented. In the near wake the shed vortices appear first as distinct vortex tubes and then merge into a continuous vortex sheet at a short distance from the rotor plane (Figure 9.a). Basically, the vorticity field structure presents the classical signature, derived from merely theoretical considerations (Figure 9.b) [24].

11V s = =G

a) b)

Figure 9. Iso-surface of computed vorticity, V0 = 6.8 m/s (a); theoretical scheme of vortex structure (b) [24]. Experimental results show that the spiral geometry is maintained at a distance from the rotor longer than that found by CFD [3]. Future works will be addressed to show the importance of Reynolds number and grid resolution, in order to achieve a better resolution of this feature. Tip vortices are also visible in the axial velocity contour plot (Figure 10). Even in Figure 10, the velocity distributions (solid lines), averaged along the pitchwise direction, are depicted for different axial positions. Moving from the internal to the external region the figure shows strong gradients in the interface between the wake and the undisturbed region. Axial velocity contours are used to identify the transition from the near wake to the far wake, according to the definition given above (Figure 11).

V0=6.8m/s

y/D=0.25 y/D=0.1 y/D=0 y/D=0.1

y/D=0.25 y/D=0.5 y/D=1 y/D=2.5

ContourofAxialVelocity

Figure 10. Near wake axial velocity distributions, V0 = 6.8 m/s. Figure 11. Contour of axial velocity, V0 = 6.8 m/s.

The velocity distributions in the far wake are predicted using a different configuration of the computational domain, enclosed between a small inner cylinder and an outer cylinder with diameter equal to 10 times the rotor diameter, both axial centred (Figure 12). In this way the axial region, difficult to mesh, is totally removed. The downstream axial length is now 15 times the rotor diameter and the boundary conditions used are symmetry for the external surface, Euler-slip for the internal surface and velocity inlet at inlet face (Figure 12). In the contour plot of Figure 13 velocity distributions (solid lines) are shown, for different axial positions, as well as for the near wake.

1-

- 3/4

- 1/2

- 1/4

0

1/4

1/2

3/4

1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

y/Dr/D

Vy/V1,00,0

Figure 12. Far wake computations: domain and BC. Figure 13. Far wake axial velocity distributions, V0 = 6.8 m/s.

6. CONCLUSIONS In this paper, a broad numerical analysis on HAWT aerodynamics is presented. A commercial CFD code is used to predict the global flow field about rotors of wind turbines. From the outside, aerodynamics of wind turbines seem quite simple. Despite the apparent simplicity, the aerodynamics description of wind turbines is complicated by the presence of several 3-D effects, separated flow, wakes interaction, vortices, etc. The aim of this work is to get a better understanding of HAWT aerodynamic mechanisms governing the energy conversion process. Moreover, this is needed since todays industrial design codes for wind power applications are still based on the BEM method. Different aspects of HAWT flow field are resolved with good accuracy, despite the different relevant scales involved. A complete overview of the flow field is given, while almost all past CFD works on this subject are focused on specific flow features. In addiction, special focus is put on recent rotor design tendencies, in order to verify the performances improvement declared. The overall performances are computed under various inflow conditions and are validated with the BEM method. By predicting the power output for different rotor configuration, the validity of the recent design in blade rotors are confirmed. The numerical simulations allow also to predict the basic features of both near and far wakes, including velocity deficit distribution and vortex structures. This study confirms that CFD simulations can be nowadays considered the most important tool for predicting the aerodynamics of modern wind turbines. NOTATION HAWT Horizontal Axis Wind Turbine 2-D, 3-D Two-dimensional, Three-dimensional BEM Blade Element Momentum CFD Computational Fluid Dynamics MRF Moving Reference Frame QUICK Quadratic Upwind Interpolation for Convective Kinetics RANS Reynolds Averaged Navier-Stokes S-A Spalart-Allmaras SIMPLE Semi-Implicit Method for Pressure Lined Equations SST Shear Stress Transport a Axial interference factor CP Power coefficient D, R Rotor diameter, rotor radius

f Viscous damping factor G ,Y , S Production, destruction and source of M Shaft mechanical moment P Shaft mechanical power t Time ui Velocity components V0 Undisturbed wind velocity y+ Non-dimensional distance of the first grid point from the wall , y, r Cylindrical coordinates (y, rotational axis) Density Ratio between and Global tip speed ratio ij Shear stress tensor , Molecular kinetic and dynamic viscosity t, t Turbulent kinetic and dynamic viscosity Transport variable of the S-A model Vorticity

2, bC Constants of the S-A model REFERENCES 1. IEA, International Energy Agency. Wind Energy Annual Report 2004; April 2005. 2. Vermeer LJ, Srensen JN, Crespo A. Wind Turbine Wake Aerodynamics; Progress in Aerospace Science, 2003; Vol.

39; 467-510. 3. Srensen JN, Shen WZ. Numerical Modeling of Wind turbine Wakes; J. Fluid Engineering 2002; Vol.124; 393-399. 4. Ivanell SSA. Numerical computations of wind turbine wakes; Technical reports from KTH Mechanics, Royal Institute

of Technology; Stockholm, Sweden; 2005. 5. Kang S, Hirsch C. Features of the 3D flow around wind turbine blades based on numerical solutions; Proceedings

from EWEC 2001, Copenhagen. 6. Alinot C, Masson C. Aerodynamic simulation of wind turbines operating in atmospheric boundary layer with various

thermal stratifications; AIAA Paper 2002-0042. 7. Mikkelsen R. Actuator disc methods applied to wind turbines; Dissertation submitted to the Technical University of

Denmark in partial fulfillment of the requirements for the degree of PhD in Mechanical Engineering; Lyngby, 2003. 8. Xu G, Sankar LN. Computational study of HAWT; AIAA Paper 1999-0042. 9. Xu G, Sankar LN. Effects of transition, turbulence and yaw on the performance of HAWT; AIAA Paper 2000-0048. 10. Benjanirat S, Sankar LN, Xu G. Evaluation of turbulence models for the prediction of wind turbine aerodynamics;

AIAA Paper 2003-0517. 11. Michelsen JA, Srensen NN. Current developments in Navier-Stokes modelling of wind turbine rotor flow;

Proceedings from EWEC 2001, Copenhagen. 12. Srensen NN. 3-D Background aerodynamics using CFD; Ris National Laboratory, Roskilde, Denmark, 2002. 13. Srensen NN, Johansen J, Conway S. KNOW-BLADE Task-3.1 report. Computations of wind turbines blade loads

during standstill operation. Ris National Laboratory, Roskilde, Denmark, 2004. 14. Srensen NN, Johansen J, Conway S, Voutsinas S, Hansen MOL, Strmer A. KNOW-BLADE Task-3.2 report. Tip

Shape Study. Ris National Laboratory, Roskilde, Denmark, 2005. 15. Mandas N, Carcangiu CE, Cambuli F. The economy of large scale wind turbines; Fluent News Summer 2005; Vol.

XV; 5-7. 16. Cambuli F, Carcangiu CE, Mandas N. Studio numerico del flusso su rotori eolici ad asse orizzontale; Proceedings

from the 60th ATI meeting, Rome, 13-15 September 2005. 17. Rohden R. Revolutionary Rotor Blade Design; Wind Blatt, the Enercon Magasine 2004; Issue 03. 18. Anderson DA, Tannehill JC, Pletcher RH. Computational Mechanics and Heat Transfer; Taylor & Francis Inc., New

York, 1997; 249-285. 19. FLUENT Inc. Fluent 6.1 Documentation, User's Guide; 2003. 20. Hansen, MOL. Aerodynamics of Wind Turbines. Rotors, Load and Structures; James&James, London, 2000. 21. Mandas G. Progetto fluidodinamico di un rotore di turbina ad asse orizzontale; B.Sc.Thesis, University of Cagliari,

2002. 22. Abbott JH, Von Dohenoff AE. Theory of Wind Sections; Dover Publications Inc.; New York, 1959; 522-545. 23. De Vries O. Fluid Dynamics of Wind Energy Conversion; AGARDograph N 243, 1979. 24. Wilson RE, Lissaman PBS. Applied Aerodynamics of Wind Power Machines; Oregon State University, 1974.

UNIVERSITDEGLISTUDIDICAGLIARI

PiazzadArmi09123CAGLIARIITALYTel.+390706755951Fax+390706755717

NatalinoMandas FrancescoCambuliCarloEnricoCarcangiu ProfessorD.I.Me.Ca ResearchAssistantD.I.Me.CaPh.D.StudentD.I.Me.Ca mandas@dimeca.unica.it cambuli@dimeca.unica.itcarcangiu@dimeca.unica.it

SUMMARYTheaerodynamicsofaHorizontalAxisWindTurbine(HAWT)areinvestigatedusingacommercialCFDcode.RANSequationsareusedtosolvethe3Dturbulentsteadyincompressibleflow.Allthecomputationsarecarriedoutforamiddlesizedturbinewithbothaclassicalandaninnovativehighperformancebladegeometry.ThefeaturesoftheflowarepredictedandtheglobalperformancesarecomparedwithBEMmethodresults.Theresultsshowtheconsiderableincrementofpowerobtainedbytheinnovativedesign.Thestudyyieldsinformationaboutbasicaerodynamicfeaturesoftherotor,nearandfarwakeflow.

1.INTRODUCTIONWindenergy isa lowdensity sourceofpower.Tomakewindpowereconomically feasible, it is important tomaximize theefficiencyofconvertingwindenergyintomechanicalenergy.Ofallthedifferentaspectsinvolved,rotoraerodynamicsisakeydeterminantforachievingthisgoal.Inaddition,theabilitytopredictthedownstreamwakefromawindturbineisasignificantfactor fordetermining the interactionsbetween turbines.Researchwork conducted in thisareahasbrought toa substantialimprovementintheoverallefficiencyoftheconversionprocess,withtheresultthatthecapitalcostsofinstallingwindpowercannowcompeteeffectivelywithotherenergysources.Threeapproachesareavailable toanalyze the flowaroundanddownstreamofawind turbine: field testing,whichprovidesaccurate results but is highly complex and expensive; analytical and semiempirical models, which adopt simplifyingassumptionsandarethusnotuniversallyreliable;andCFD,whichoffersthebestalternativetodirectmeasurements.

2.MATHEMATICAL MODELThemathematicalmodelincludes

x Continuityequation

x Momentumequation

TheseequationsaresolvedinasingleMovingReferenceFrame(MRF),withthehypothesisofincompressible and turbulentsteady flow(incompressiblesteadyRANSequations).

The oneequation SpalartAllmaras modelwith standard wall functions (y+ 30) ischosenforturbulenceclosure.

For some calculations the NZ SST eddyviscosity model of Menter is used, givinghoweveranalogousresults.

3.GEOMETRYOFTHETURBINE

The classical BladeElementMomentum (BEM)method is adopted for thedesign of the turbine rotor, using the specifications for the threebladedhorizontalaxisNordtank41/500turbine(diameter=41m,ratedpower500kW,stallregulated)andNACA634xxprofiles.Theactivepartofthebladewasextended to thehub (Figure1a), inkeepingwith thedesignused inmodernwindturbines(Figure1b).

4.COMPUTATIONALGRIDANDDOMAINTogeneratethevolumemeshforthethreebladedrotor,the120degreesperiodicityoftherotorisexploited by only meshing one blade. The remaining two blades are included in thecomputations through the use of periodic boundary conditions. Thewind turbine tower andground are not included in themodel and a uniformwind speed profile is assumed at theentranceofthedomain.The preprocessorGAMBIT is used to build an exahedralmesh of approximately 1.5millionvolumes (Figure 3).Around the blade the grid isHshaped (Figure 2) and it is optimized (bymeansof2Dsimulations),toresolvetheboundarylayer.The computationaldomainused is conical shaped, extending in the axialdirection roughly 5diametersupstreamand10diametersdownstreamof the rotor (Figure4). In theplaneof the

rotor,thedomaindiameterisfivetimesthatoftherotor.Meshgenerationspresentsmanydifficultiesowingtothedifferentrangeofgeometricscalesrepresented:lengthofthedomain(600m),sizeoftherotor(41m),typicalchordlengths(0.53m),andboundarylayer(10mm).

5.RESULTSPerformance:CFDvsBEM

The commercial code FLUENT 6.2 isused in all calculation presented in thefollowing.To validate the model the overallperformances for thenewdesign turbine,evaluated with CFD and BEM method,are compared. Figure 5 shows the shaftmechanical power as a function of thewind velocity and the correspondingpower coefficient as a function of tipspeedratio.TheCFD resultsare found tobe ingoodagreementwith thoseobtainedusing theBEMmethod.

6.CONCLUSIONSDifferentaspectsofHAWTflowfieldareresolvedwithgoodaccuracy,despitethedifferentrelevantscalesinvolved.AcompleteoverviewoftheflowfieldisgivenwhilealmostallpastCFDworksonthissubjectarefocusedonspecificflowfeatures.TheoverallperformancesiscomputedundervariousinflowconditionsandarevalidatedwithBEMmethod.Bypredictingthepoweroutput fordifferentrotorconfiguration, thevalidityof therecentdesign inbladerotorsareconfirmed.Thenumericalsimulationsallowalsotopredictthebasicfeaturesofbothnearandfarwakes,includingvelocitydeficitdistributionandvortexstructures.Atlast,thisstudyconfirmsthatCFDsimulationscanbenowadaysconsideredthemostimportanttoolforpredictingtheaerodynamicsofmodernwindturbines.

V0=6.8m/s

y/D=0.25 y/D=0.1 y/D=0 y/D=0.1

y/D=0.25 y/D=0.5 y/D=1 y/D=2.5

ContourofAxialVelocity

NewandClassicRotorDesign

In order to assess the advantage of the newconfiguration a comparison ismadewith the classicrotor,characterizedbyacylindricalconnectionclosetothehub.Thissolution,whileovercomingstructurallimits,causesalackofyieldedpower.Mechanical power foundwith CFD simulations forthedifferentconfigurationarecompared inFigure6,showing that innovative blade allows a 5 percentincrementofthepowercoefficient.

RotorAerodynamics

Figure7shows thedistributionof theaxial interference factor,a=1Va/V0, in therotorplane.A largechange in the inducedvelocitytakesplaceacrosstheblades.CFDsimulationsgiveusefulinformationtoimprovetheBEMmodel,stillthemostusedasadesignmethod.Thetipgeometryisobtainedbysimplytruncatingthebladewithacylindricalsurface.Thesimulationsrevealstheformationofstrongtipvortices(Figure8).Futureworkwillbeaddressedtosimulatedifferenttipgeometriesthatallowareductionoffluiddynamicandacoustictipeffects.PathlinesofFigure9showtherelativeflowfieldnearthebladeroot.

FarWake

Thevelocitydistributionsinthefarwakearepredictedusingadifferentconfigurationofthecomputationaldomain,comprisedbetweenasmallinnercylinderandanoutercylinderwithdiameterequalto10timestherotordiameter.Thetotalaxiallengthis15 times the rotordiameter and theboundary conditionsused are symmetry for the external surface andEuler/slip for theinternalsurface (Figure14). In thecontourplotofFigure15velocitydistributions (solid lines)are shown, fordifferentaxialpositions.

NearWake

In the nearwake the shed vortices appear first as distinct vortex tubes and theymergeintoacontinuousvortexsheetatashortdistancefromtherotorplane(Figure10). Tip vortices are also visible in a axial velocity contour plot (Figure 11).Experimental results show that thespiral geometry is maintained at alongerdistancefromtherotor.Futureworkswillbeaddressed to show theimportance ofReynolds number andgridresolution.InFigure12thevelocitydistributions(solid lines), averaged along thepitchwise direction, are depicted for

different axial positions.Moving from the internal to the external region thefigure shows strong gradients in the interface between the wake and theundisturbed region. Finally axial velocity contours are used to identify thetransitionfromthenearwaketothefarwake(Figure13).

[ u G

11V s

0.487122.7New

0.465117.0Classic

PowerCoefficient CPMechanical Power[kW]V0 =6.8m/s

0.487122.7New

0.465117.0Classic

PowerCoefficient CPMechanical Power[kW]V0 =6.8m/s

New ClassicNew Classic

(a) (b)Fig.1Modernwindturbinebladeinboarddesign.

Gambitmodel(a)andcorrespondingrealturbine(b,Enercon).

Fig.3Modularmeshgeneration Fig.4Computationaldomainandboundaryconditions

Fig.2Hmesharoundabladerootsection

Fig.5MechanicalPowerPandPowerCoefficientCP,BEMmethodandCFDcalculations,fordifferentwindvelocities,V0,andtipspeedratios,

Fig.7Axialinterferencefactor,a=1Va/V0 Fig.8Pathlinesatbladetip Fig.9Pathlinesatbladeroot

Fig.6Computedmechanicalpower.Newandclassicbladedesign

Fig.10Isosurfaceofvorticity,V0=6.8m/s

Fig.11Axialvelocity

Fig.13ContourofAxialVelocityFig.12Nearwakeaxialvelocitydistributions

Shaft Mechanical Power

0

100

200

300

400

500

0 4 8 12 16 20 24

V0 [m/s]

P [k

W]

Power Coefficient

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0 0,1 0,2 0,3 0,4

O = (ZR/V0)-1

CP

CFD BEM

1 -

- 3/4

- 1/2

- 1/4

0

1/4

1/2

3/4

1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

y/D

r/D

V y/V 01,00,0 8.0

7.5

7.0

6.5

6.0

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Vy m/s

1 -

- 3/4

- 1/2

- 1/4

0

1/4

1/2

3/4

1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

y/D

r/D

V y/V 01,00,0 8.0

7.5

7.0

6.5

6.0

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Vy m/s

Fig.15FarwakeaxialvelocitydistributionsFig.14Farwakecomputations:domainandbc

-2,0

-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

2,0

-0,5 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 5,5

y/R

r/R

Vy/V0

1,0 - 0,0 1,0 -0,0 1,0 - 0,0 1,00,0

MCC_EWEC06_Paper.pdfMCC_EWEC06_PosterA4.pdf