Department of Mechanical Engineering and Material Science, Yokohama National University,
79-5 Tokiwadai, Hodogaya-ku, Yokohama 240-8501, Japan
1. INTRODUCTION so on are now actively studied.
Efficient use of wind turbine as the electric In this article the basic issues and some current power generator has been developed by application topics on aerodynamics of horizontal axis wind tur- of aerodynamics for rotor blade. “Airfoil” has made bines (HAWT) which are of majority of current op- rotor possible to rotate in high speed and load. Early erating turbines are surveyed. aerodynamics of wind turbine has based on theory of airplane and helicopter rotor. However, aerodynam- 2. AIRFOIL ics of wind turbine has been required different idea For HAWT blades the aviation airfoils such as from helicopter rotor. For example, the accuracy of NACA series have been widely used. But these air- rotor performance analysis depends mainly on the foils have been recognized to be insufficient for re- treatment of “wake effect”, because the wake of pro- quirements such as reduction of rapid stall charac- peller type wind turbine is induced a large velocity in teristics, insensitivity to wide Reynolds number in rotor plane. Progress of aerodynamics of wind turbi- the range of between 500,000 and 2 million, insensi- ne has ensured the new concept design turbines such tivity to leading edge roughness and structural effi- as a stall controlled machine. ciency for root region where is designed using a As the use of large scale commercial wind tur- thick airfoil. bines is expanded over the world, the wind turbine New airfoil families for HAWT are designing in industry is facing many difficulties in consulting Delft University of Technology [4], National Re- efficient wind turbine machines, which is caused by newable Energy Laboratory (NREL) [5] and Riso the inability to adequately predict structural loading National Laboratory [6], et al.. Figure 1 shows an and power output. As the wind turbines are operating example of airfoils for HAWT blades [6]. in the natural environment, turbines are always sub- jected to unsteady nature. However, recognizing on 3. ROTOR PERFORMANCE ANALYSIS aerodynamics of wind turbine associated with three- Rotor performance analysis of HAWT has been dimensional, unsteady and separated complex flow is performed using several methods as shown in table 1. not sufficient. The unsteady aerodynamics of wind In these methods a Blade Element Momentum turbines is one of the most incentive issues in wind (BEM) method is mainly employed as a tool of per- energy research and development [1,2,3]. The sub- formance analysis for HAWT because of their sim- jects on aerodynamics of wind turbines such as des- plicity and readily implementation. Vortex wake ign of new airfoil [4,5,6], three-dimensional effect of methods can adequately treat the effect of wake vor- rotating blade [7,8], dynamic inflow effect [9,10] dynamic stall associated with yawed inflow [11] and Table 1: Analysis methods of HAWT performance Method Model Researcher Actuator disk Betz [12] Momentum Blade element Wilson [13] Lifting line Afjeh [14] Vortex wake Lifting surface Simoes [15] Local circulation Elliptic blade Azuma [16] Acceleration poten- Pressure per- van Bussel [17] tial turbation Figure 1: Profiles of Riso A1- series airfoil [6] CFD RANS Duque [18] tices and have some advantages over BEM, but need computation burden. As computer power increases, use of vortex wake methods is increased, and CFD codes are also applied in the resent aspect. This section follows the features of each method on performance analysis of HAWT. 3.1 Blade Element Momentum Method Most wind turbine design codes are based on Blade Element Momentum (BEM) method [13]. The basic of BEM method assumes the blade can be Figure 2: Wake visualization of 10m HAWT by NREL analyzed as a number of independent element in (quotation by courtesy of NREL) spanwise direction. The induced velocity at each element is determined by performing the momentum balance for an annular control volume containing the blade element. The aerodynamic forces on the ele- ment are calculated using the lift and drag coefficient from the empirical two-dimensional wind tunnel test data at the geometric angle of attack (AOA) of the blade element relative to the local flow velocity. BEM methods have aspects by reasonable tool for designer, but are not suitable for accurate estima- tion of the effect of wake, the complex flow such as Figure 3: Tip vortex and pressure contours of a rotor flow using EllipSys2D/3D code by Risoe national laboratory three-dimensional flow or dynamic stall because of (quotation by courtesy of Risoe laboratory) their assumptions. 3.2 Vortex Wake Method The induced velocity in the rotor plane of treatment of the wake effect such as yawed inflow or HAWT is largely increased in a heavy loading con- dynamic inflow, but they was not common tool for dition and the wake vortices of HAWT develop to design of turbines because of their computation bur- the downstream constructing highly skewed vortex den. However, their use are increasing with increase sheet (see fig.2) in a largely decelerated axial flow of computer power. near the rotor plane. Thus, the determination of the An other method to the vortex wake methods is velocity induced by wake and wake geometry is one use of an asymptotic acceleration potential [17]. Ac- of the most important aspects in the rotor perfor- celeration potential method is basis on the Lapace mance analysis. equation of pressure perturbation. The rotor blades Vortex wake method directly calculates the in- are represented in the model as discrete surfaces on duced velocity from the bound vortices of blades and which a pressure discontinuity is present. The model the trailing vortex in wake which are represented by implies the presence of spanwise and chordwise lifting line [14] or lifting surface model [15]. pressure distributions, which are composed of ana- The treatment of wake geometry can be classi- lytical asymptotic solutions for Laplace equation. fied roughly into two types, as a prescribed wake The asymptotic acceleration method gives a model [14] and a free wake model [15]. In the possibility to simplify or elaborate the model in former model the wake represented by a line vortex specific areas of interests. In its simplest form the or spiral vortices with fixed pitch. In the latter one, a model is equivalent to the lifting line model with fractional step scheme is adopted and the configura- delinearized wake. More elaborated models makes it tions of the wake are calculated at every time step possible to calculate the dynamic loads caused by using a local velocity including the components in- dynamic inflow [10] and yawed inflow [11] situa- duced by wake and bound vortices. The free wake tions. model is generally tackled with vortex lattice method 3.3 CFD which can fit on arbitrary blade shape with camber, Recent development of the computational fluid taper and twist. Vortex wake methods are of accurate dynamics (CFD) allows us to simulate overall flow around HAWT including tower and nacelle. In 1999 Duque et al. [18] calculated aerodynamics of a HAWT using the RANS model and overset grids to facilitate the simulation of flows about complex con- figurations. Recently, some CFD codes actively are developed. Figure 3 shows an example of CFD ana- lysis of a rotor flow by three-dimensional Navier- Stokes code. Though the state of the art CFD is needed con- siderable computer power and validations for Navier-Stokes model, CFD has potential advantage for the detailed understanding of aerodynamics of Figure 4: Upper surface pressure distribution vs azimuth HAWT. angle; yaw angle of 30 degree and 30% span position [2]
4. UNSTEADY ROTOR AERODYNAMICS
Wind turbines in the open air field are always exposed to the operating conditions in a very insta- tionary environment. The aerodynamic loads on blades and structures consequently are of an insta- tionary nature. Therefore, the clarifications of the dynamic effects on the load and dynamic behavior of wind turbines in such conditions are quite important for the design of the wind turbine. 4.1 Yawed Inflow When incoming flow into the rotor is yawed to the rotor plane, due to the rapid and periodical change of the local AOA to the rotor blade, the dy- namic load to the rotor is excited by the dynamic stall effect. Figure 5: Normal force coefficient vs angle of attack at Figure 4 shows the experimental result accord- 30% and 63% span positions; yaw angle of 30 degree. ing to the azimuthal change on the upper surface Dashed line is 2D static wind tunnel test data [2]. pressure distribution at 30% span position in a yawed flow and low tip speed ratio condition for steady rotor aerodynamics, such as coherent wind down-wind type HAWT [2]. The blade passes behind gusts, collective blade pitch and rotor speed varia- the tower at 180 degree of azimuth angle. Because tions. the dynamic stall vortex is induced by rapid change Dynamic inflow accounts for the influence of of local AOA, the large and transient leading edge time varying trailing wake vortices on the inflow suction pressure peaks were formed. This causes the velocity in the rotor plane. The vorticity is formed at significant lift enhancement and the periodical load the blade and convected downstream with local total to the blades. Figure 5 shows the normal force coef- velocity, partly wake induced. Equally, the strength ficient for yaw angle of 30 degree at 30% and 63% of trailing and shed vortices depends on the wake span. The hysteresis loop shows significantly differ- through its effect on the angles of attack. Dynamic ent between inboard and outboard locations. inflow phenomena depend mainly on the trailing In natural environment, when inflow turbulence vortices, the shed near wake being accounted for is added, effecting both inflow magnitude and direc- through unsteady profile aerodynamics. The charac- tion, the flow becomes more three-dimensional and teristic time scale for this phenomenon is rotor di- the dynamic stall effect becomes much more com- ameter / inflow velocity. For large turbines, this will plex [3]. be of the order of 5 to 10s. Phenomena having a time 4.2 Dynamic Inflow scale large compared to this may be regarded as Dynamic inflow concerns with large scale un- quasi-steady. [2] Huyer, S.A., et al., Unsteady Aerodynamics Associated with a Horizontal-Axis Wind Turbine, AIAA Jour. Vol.34, No.7, 1996, pp.1410-1419. [3] Robinson, M.C., et al., Horizontal Axis Wind Turbine Aerodynamics: Three-Dimensional, Unsteady and Separat- ed Flow Influences, Proc. of 3rd ASME/JSME Joint Fluids Eng. Conf., FEDSM99-7811, 1999. [4] Timmer, W.A. and van Rooy, R.P.J.O.M., Wind Tunnel Results for 25% Thick Wind Turbine Blade Airfoil, Proc. of EWEC’93, 1993, pp.416-419. [5] Somers, D.M. and Tangler, J.L., Wind Tunnel Test of the S814 Thick Root Airfoil, Trans. of ASME, Jour. of Solar Energy Eng., Vol.118, 1996, pp.217-221 [6] Dahl, K.S., et al., Experimental Verifications of the New RISO-A1 Airfoil Family for Wind Turbines, Proc. of Figure 6: Spanwise load distribution for stepwise variation EWEC’99, 1999, pp.85-88. in wind speed [9] [7] Maeda, T. and Bruining, Cross-flow Effects of pressure distributions at the blade of the Delft Open Air Rotor Re- search Facility, Proc. of EWEC’96, 1996, pp.726-729. [8] Hansen, M.O.L., et al., Computed 3D Effects on a Rotat- ing Wind Turbine Blade, Proc. of EWEC’96, 1996, pp.761-763. [9] van Bussel, G.J.W., The Acceleration Potential Models PREDICHAT/PREDICDYN Applied for Calculation of Axisymmetric Dynamic Inflow Cases, IW-93071R, Insti- tute for Windenergy, Delft University of Technology, 1993. [10] Schepers, J.G., et al., Investigation and Modelling of Dy- namic Inflow Effects: Yaw Modelling and Validation, Proc. Figure 7: Measured and calculated flapping moment at of EWEC ’94, Volume I, 1994, pp.699-707. inboard position for step on the pitch angle [10]. [11] Hasegawa, Y., et al., Numerical Analysis of Yawed Inflow Effects on a HAWT Rotor, Proc. of 3rd ASME/JSME Joint Figure 6 shows an example of spanwise load Fluids Eng. Conf., FEDSM99-7820, 1999. [12] Betz, A., Windenergie und ihre Ausnutzung durch, Van- distribution when the wind speed is varied in step- denhoeck und Ruprecht, 1926. wise [9]. Figure 7 shows comparison of the meas- [13] Wilson, R.E. and Lissaman, P.B.S., Applied Aerodynamics ured and calculated flapping moment at inboard of Wind Power Machines, NTIS PB 238594, Oregon State blade position for step on the pitch angle [10]. Free University, 1974. and prescribed vortex wake model and acceleration [14] Afjeh, A. A. and Keith Jr. T. G., A Vortex Lifting Line Method for the Analysis of Horizontal Axis Wind Turbines, potential method were employed for simulation. The Transactions of the ASME, Journal of Solar Energy Engi- measurements appear to be in a good qualitative and neering, Vol.108, 1986, pp.303-309. quantitative agreement with the calculated results. [15] Simoes, F. J. and Graham, J. M. R., A Free Vortex Model of the Wake of a Horizontal Axis Wind Turbine, Proceed- 5. CONCLUSIONS ings of 12th British Wind Energy Association Conference, 1990, pp.161-165. This article presented the basic issues on [16] Azuma, A. and Kawachi, K., Local Momentum Theory aerodynamics of HAWT. Recognition of three di- and Its Application to the Rotary Wing, Journal of Aircaft, mensional, unsteady separated flow field around the Vol.16, No.1, 1979, pp.6-14. HAWT is indispensable to develop aerodynamics of [17] van Bussel, G.J.W., PREDICHAT, First Order Perfor- wind turbines. Cooperative research projects on the- mance Calculations of Windturbine Rotors Using the se subjects are now actively conducting by institutes Method of the Acceleration Potential, IW-93069R, Insti- tute for Windenergy, Delft University of Technology, in EU and USA. 1993. [18] Duque, E.P.N., et al., Navier-Stokes Analysis of Time- References Dependent Flows about Wind Turbines, Proc. of 3rd [1] Hansen, A.C. and Butterfield, C.P., Aerodynamics of Hori- ASME/JSME Joint Fluids Eng. Conf., FEDSM99-7814, zontal-Axis Wind Turbines, Annu. Rev. of Fluid Mech., 1999. 1993, pp.115-149.