AIRFOIL TYPES EFFECT ON GEOMETRY AND PERFORMANCE OF A SMALL-SCALE WIND TURBINE BLADE DESIGN

Many airfoils could be used to form a wind turbine blade. Different airfoil would result in an altered blade planform and its performance characteristics. Airfoil identifi cation that should be included in the design then becomes an important task. Utilizing Blade Element Momentum Theory computation procedures, eleven types of airfoils were applied in blade geometries of rotors of 2.4 m in diameters. The changes in the distribution of chord lengths, twists, solidity ratio, and Reynolds numbers of the blades were compared as geometry parameters. The designed and off designed powers and thrusts were calculated, and the characteristics of the performance of the rotor were represented by the coeffi cient of power and coeffi cient of thrust.The calculated geometries showed that distinct airfoils resulted in different segments sizes of the blades and different performance characteristics of the rotors. The use of an airfoil that has a high lift coeffi cient and a high glide ratio in the design of a blade will produce a narrow blade and a small solidity ratio. The design will also have a high power coeffi cient at the tip speed ratio design. However, the blade power coeffi cient may be sensitive to changes in rotational speed.


INTRODUCTION
Researchers have been striven to raise the aerodynamic performance of wind turbines. They have been modifi ed the used airfoil, the designed planform, and the designed system attached or around turbines. A new type of airfoil series has been developed and succeeded in increasing the power coeffi cient power, annual energy production, and lowering root load [1] [2]. The planform of a small wind turbine has been modifi ed from straight design to curved one [3] [4] and from simple tip geometry to a tip added with swept, winglet, dihedral or anhedral [5]- [7]. By adding a diffuser at a wind turbine rotor system, the performances were signifi cantly increased [4] [8]. A pitch control has been attached in the rotor systems to optimize harvesting power from a range of wind speeds [9]- [11]. A blade of a wind turbine rotor could consist of one or several types of airfoils. The designer might choose some of them to reach the optimum aerodynamics behavior and mechanical strength for rating operation. The choice of the airfoil for blade design is related to its aerodynamic characteristics. The lift and drag coeffi cient characteristics determine the glide airfoil ratio associated with the effi ciency of the profi le. The blade is designed at an airfoil angle of attack where the maximum glide ratio. The higher the glide ratio, the more effi cient the blade profi le. Meanwhile, the airfoil geometry in the blade design determines the centrifugal force and blade strength when the turbine is operated. Small wind turbine rotor blades usually consist of one type of airfoil. Thin airfoil can be used to save material, reduce weight and facilitate balancing. In the case of large-scale blade designs, several airfoils can be used simultaneously to meet the aerodynamic and structural strength requirements. Thin airfoil is used at the tip of the blade while thick airfoil is used at the root. Some researches show the optimum blade design [12] to [14], but the reason why the airfoils used in the works are selected from the others need extended explanation. Different types of airfoil might give different optimum blade planform and performance. The need for wind turbine blade design was relayed primarily on two-dimensional aerodynamic characteristics of a used airfoil. For low Reynolds number, the chosen airfoil should have a maximum thickness as small as possible, and the location of the relative thickness located close to trailing edge as it had good performance [15]. However, reviewing airfoil properties by looking for the angle of attack at maximum glide ratio might be more accessible. The design of the angle of attack would result in the end properties of the blade design [16]. By the fast and familiar method, BEMT, parameters needed from an airfoil for blade design are angle of attack of the design, coeffi cient of lift at an angle of attack of design, characteristic of the coeffi cient of lift as a function of angle of attack, and coeffi cient of drag as a function of angle of attack. The parameters were used to acquire the blade geometry and to determine the performance [17]- [19]. To know how airfoils affect the blades geometry and their performance, in the current paper, eleven blades that are designed by BEMT with different airfoils for each blade that usually used in researches and constructions will be discussed. BEM based computing method was selected for blades design in this paper since it was simple, fast calculating, had proven good accuracy [20]- [22], and not sensitive to the thickness of airfoil as blade design using this method only required information on the value of the lift and drag coeffi cient, not airfoil geometry.

Airfoils
The design process using BEM utilizes the two-dimensional aerodynamic characteristics of the airfoil. The polynomial equation is used to approach the lift and drag coeffi cient values as close as possible. The fi rst-order approach can be carried out in an area where changes in the airfoil angle of attack give a linear change to the lift and drag coeffi cients. Meanwhile, the second-order curve can be used in a change in the angle of attack which results in a parabolic change in the lift and drag coeffi cients. The choice of the order of the curve to approach the aerodynamic characteristics of an airfoil depends on the ability of the curve to approach the characteristics of the airfoil. Here, the 4th order polynomial has been chosen to approach the characteristics of an airfoil in all parts of the angle of attack. Blades cross cross-sec- tions of each HAWT rotor in this paper were built utilizing one of the airfoils mentioned in Table 1. All data in Table 1 were based on airfoil characteristics with Reynolds number 200 000. The sections' shapes were used commonly in wind turbine rotor developments. Coeffi cients of lift (C L ) and coeffi cients of drag (C D ) data were extracted from http://airfoiltools.com. The page uses Xfoil to design and analyze subsonic airfoils. The application uses the high order panel method combined with viscous/inviscid interactions. The data then were approximated by polynomial fi ts of order four as in Eq. 1. The equations arefunctions of the angle of attack of an airfoil (α) that have constants a1, a2, a3, a4, and a5. The lift coeffi cients for design (C L,D ) in the third column in the table were determined at a design angle of attack (α D ) where the ratio of lift to drag reached a maximum ((C L ⁄C D ) max ).

Blade Element Method (BEM)
BEM based computation procedures were the workhorse in forming blades geometries and analyzing rotors characteristics. The methods have been presented in some references, i.e., by Schaffarczyk [17], Gundtoft [18], and Burton et al. [19]. In this paper, we discussed rotors that consisted of three blades (B), radius (R) of 1.2 m and tip speed ratio of the design (λ D ) equalss 8. The tip speed ratio was the ratio between tangential speed of blade tip and wind speed on operated designs. The rotors had rated rotation speed of about 700 rpm which were matched with generator rated speed operation. Chords (c) and twists (β) distributions of the blades were calculated based on optimum Schmitz blade geometries as in Eq. 2 and 3. The coeffi cient of lift and drag characteristic of airfoils were approximated using Reynolds numbers (Eq. 4) of each blade segment when operated at wind speed 10 m.s -1 and air condition at 25 o C. The Reynolds numbers were calculated based on relative wind speed (V rel , Eq. 5 and 6) at an angle of φ with axial interference values (a) determined at optimum at 1/3.The angle of attacks (α D ) and coeffi cients of lift (C L,D ) of the designs then were defi ned at the maximum glide ratio ((C L /C D ) max ) of related airfoils characteristic. Small wind turbines are recommended to be placed in an area that is not affordable by the electricity network but has a large wind resource potential. The choice of wind speed of 10 m/s -1 here may be somewhat excessive because the nominal value is rarely encountered. Small wind turbines will receive energy from winds whose speeds vary due to their relatively low placement from the ground. This wind speed may be found in narrow valley areas that form the aisles or in coastal areas that have a certain landform. (2) Performances of the rotors were computed by iterating the axial (a) and tangential (a') interference. The values were started from zero then calculated by Eq. 7-13 iteratively until their error reaches under 0.001. Prandtl tip loss correction was used in rotor performance calculations. The tip loss factor F in Equation 11 was involved in iterations to correct the values of axial interference (Eq. 7) and tangential interference (Eq. 8). The axial force (T) of the rotors were was the sum of axial forces of blade segments (Eq. 14), and the powers (P) were results from the generated power of all blade segments (Eq. 15). The performances were presented in the coeffi cient of power (C P ) and thrust (C T ).C P was defi ned as the ratio of power produced by wind turbine rotors to the wind fl ow (Eq. 16) and C T was calculated as the ratio of the axial force of the rotors to wind thrust (Eq. 17). (11)

Segment test
The number of segments (Ne) for blade performance computation was chosen from segment tests. The tests were conducted for NACA 23012 cross-section blades, started from 4 to 20 segments. The blades were equally segmented. The fi rst segment ringswere not included in the computations. In the calculation of rotor performance, the hub or blade extension section at the root section was ignored. Based on the results of the test (Fig. 1), twelve-segmented blades were chosen for performance computing.

RESULTS AND DISCUSSION
The geometry of the blades Figure 2 presented the calculated geometries of rotor designs. Here, we chose normalized rotor geometry representations as chord distribution, twist angle distribution, solidity ratio (solidity factor) distribution, and Reynolds number of blade segments as a function of normalizedspanwise distances. Chord distributions describe the width of blade segments. The coeffi cient of lift of blade design (C L,D ) defi ne the length of the blade chord (Eq. 2). The lower the coeffi cient, the longer the chord. In Fig. 2.a, NACA 0012 cross-section had the widest blade as the results of its lowest C L,D . The blade has the highest solidity ratio dis-tribution ( Fig. 2. c, Eq. 10) and resulted in the highest Reynolds number distribution ( Fig. 2.d, Eq. 4). A higher solidity ratio means the blade needs more amount of materials for construction. Reynolds number is an important parameter that governs the rotor starting easiness. High Reynolds number would produce low viscose force and a low coeffi cient of drag on a body [19]. A blade that has a high Reynolds number would tend to rotate effort-lesser. In the case of a wind turbine blade, the Reynolds number is calculated using the relative speed and chord length in addition to the air kinematic viscosity parameters. Reynolds number is directly proportional to the relative wind speed and the length of the airfoil chord. The higher the relative wind speed or the greater the length of the chord, the greater the Reynolds number and the greater the torque produced by the blade. The greater the torque produced by the blade, the easier it will rotate.
The problem with small wind turbines is the small Reynolds number so that the torque produced is also small. To maximize the torque produced, the design of the angle of attack which produces the pitch angle of the blade must be precise so that maximum relative wind speed is obtained where the axial interference approaches 1/3. A pitch angle converter mechanism can be added to get the maximum Re when starting at the cut-in wind speed. Modifi cations to the blade planform can also be done by adding swept or winglet angles to suit the blade design. Figure 2.b showed the twist distribution of the designed , the designed angle depends on the intended angle of attack (α D ). However, the difference between root twist angle and tip twist angle will be the same for all airfoil. Therefore, the fl atness of the designed blade does not depend on α D .

Axial and tangential interferences
The maximum coeffi cient of power of a wind turbine is 16/27 as stated by Betz. The coeffi cient is reached when axial interference equals 1/3. A wind turbine rotor blade would have high performance if it can result in axial interference close to the value. Figure 3.a showed the axial interference distributions of the designed blade as a function of normalized spanwise distance.The blade which utilized NACA 63-418 as cross-section resulted in the closest axial interference to 1/3 and blade which applied S822 as cross-section produced the utmost of axial interference distribution. Schmitz has introduced a tangential interference factor as the rotation of wake included in torque analyzing. The power produced by low λ D (smaller than 4) will be much lower than predicted by Betz as a result of swirl (wake) loss [18], [30]. For λ D higher than 4, the Betz power co-(a) Axial interferences distribution Figure 3: Interferences values of rotor blades (b) Tangential interferences distribution effi cient will be approached [12]. Ideally, a three bladed wind turbine has λ D in the range of 5.24 -5.45 [30] and can be pushed to 8 since noise does not become an environmental issue. Figure 3.b described the tangential interference distribution of all designed blades. The fi gure showed that all the blades developed almost have the same distribution of tangential interference. Therefore, the used airfoil did not affect wake loss. shows the distribution of the effective angle of attack (α eff ) along spanwise. The angles were part of iterative computing results of a and a'. Then, the angels were used to calculate C L and C D distributions as depicted in Fig. 4.a and b. by employing Eq. 1. The computed α eff s distributions are close to α D s, hence the C L distributions come close to C L,D . The distribution of C L and C D , then, give effi ciency near to the maximum airfoil profi le effi ciency [30]. Figure 5 describes the performance computation results of the designed rotor blades.The generated power distributions of each segment were the designed power from a wind speed of 10 m•s -1 (Fig. 5. a).The blade that utilized NACA 4412 as cross-sections produced the highest power distribution. The airfoil succeeded in converting wind energy into torque better than others. It could build axial interference approaching the ideal value, i.e., 1/3. Airfoil NACA 4418 resulted in almost similar power distribution to NACA 4412 as it had the identical capability in producing axial interference.Airfoil S809 generated the lowest power distribution. The airfoil yielded higher axial interference than S822. However, S822 more successful in converting wind energy into torque because it had lower C D distribution than S809. Figure 5.c displays blade power-producing as functions of rotational speeds. At speed design, NACA 4412 generated the highest power. At the starting condition, all airfoil tended to have initial power value. However, with the increasing speed, the generating power grew differently. NACA 0012 made blade had the fastest power development although it did not deliver the highest power at designed speed condition. It was since NACA 0012 yielded the highest chord length distribution ( Fig. 2.a) and resulted in the highest Reynolds distribution (Fig. 2.d).

Power and axial force
Airfoil variation did not affect the designed thrust force of the blades signifi cantly (Fig. 5.b). The generated forces of each segment almost the same. However, the higher chord length distribution delivered the faster development of axial or thrust force (Fig. 5.d). As seen at the blade that used NACA 0012 as cross-sections, the blade had the highest chord distribution ( Fig. 2.a). Therefore, it had the fastest thrust development (Fig. 5.b and 5.d).  Figure 6 shows performance in non-dimensional parameters. A higher chord-distribution blade directed to better aerodynamic characteristics. The blade tends to have faster power or coeffi cient of power growth and more stable from rotational speed changes when it operated in rated condition( Fig. 6.a). However, the design did not underwrite a higher coeffi cient of performance. The blade also developed a higher thrust than the one that has a shorter chord (Fig. 6.b).

CONCLUSION
Eleven wind turbine rotors were designed by including eleven different airfoils for the cross-section of the blade of each rotor. The different airfoil of the blades resulted in different planforms and performance. The discussion results are (1) Airfoil that has low C L,D results in high chord distribution along blade span and high Reynolds number of the blade segments. If the airfoil yields low drag coeffi cient distribution (i.e. NACA 0012), the blade tends to start effortlessly; (2) employing an airfoil in a blade design that has higher lift-coeffi cient design tend to give smaller chords sizes and solidity ratio. The blade that used the airfoil also tends to have a higher coefficient of power at tip speed ratio design than that has a lower one. But, it could be more sensitive to the changes in its rotational speeds. (3) For a small wind turbine, the coeffi cient of power of the design (C P,D ) has a low correlation with the solidity ratio of the blade and glide ratio of the utilized airfoil. (4) For a small wind turbine, the coefficient of thrust design (C T ) has no correlation with the solidity ratio of the blade and glide ratio of the utilized airfoil.