- Email: [email protected]
Developement and verification of a performance based optimal design software for wind turbine blades
Renewable Energy 54 (2013) 166e172
Contents lists available at SciVerse ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
Developement and verification of a performance based optimal design software for wind turbine blades Bumsuk Kim a, Woojune Kim a, Sanglae Lee a, Sungyoul Bae a, Youngho Lee b, * a b
Green and Industrial Technology Center, Korean Register, 23-7 Jang-dong Yuseong-gu Daejeon, Republic of Korea Division of Mechanical & Energy System Engineering, Korea Maritime University, 1Dongsam-dong Youngdo-gu, Busan, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 26 January 2012 Accepted 13 August 2012 Available online 7 September 2012
In this research, we developed software for designing the optimum shape of multi-MW wind turbine blades and analyzing the performance, and it features aerodynamic shape design, performance analysis, pitchetorque analysis and shape optimization for wind turbine blades. In order to verify the accuracy of the performance analysis results of the software developed in this research, we chose the 5 MW blade, designed by NREL, as the comparison model and compared with the analysis results of well known commercial software (GH-Bladed). The calculated performance analysis results of GH-Bladed and our software were consistent in all values of CP in all l ranges. Also, to confirm applicability of the optimum design module, the optimum design of the new 5 MW blade was performed using the initial design data of the comparison model and found that solidity was smaller in our design even though it produced the same output and efficiency. Through optimization of blade design, efficiency increased by 1% while the thrust coefficient decreased by 7.5%. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Wind turbine blade Aerodynamic design Performance analysis Optimal design
1. Introduction Most of the leading wind turbine manufacturers consider blades as their key components of wind turbine system and have concentrated their efforts on developing their own blade design and increasing the supply through in-house production facility. As world wind energy market grows continuously, a number of independent blade manufacturers are emerged recently, especially in China. However, wind blade market is still dominated by the leading wind turbine system manufacturers and a giant independent blade supplier, LM Wind-power [1], and their overall share in world blade market is estimated around 80 percent. Even though the total amount of blade supply has increased due to the new blade manufacturers, only a few of them have ability to produce large blades bigger than 3 MW class. Therefore, supply shortage of large blades is predicted as global offshore wind-power market expands, and blade could be a bottle neck component for large wind turbine system. In Korea, some blade manufacturers have succeeded in producing 2e3 MW class prototype blades by collaborating with foreign engineering companies. So as to enter the market successfully, securing the fundamental technology on design and production of multi-MW class blades is indispensable. * Corresponding author. E-mail address: [email protected] (Y. Lee). 0960-1481/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.renene.2012.08.029
It is required to take into account the external environmental conditions for a wind blade design that can be occurred during the design life time. Therefore, design evaluations by aero-elastic simulation, CFD analysis, and FEM analysis should be conducted interactively to secure safety and reliability of blades and to satisfy the requirements of international standards. So as to optimize blade shape for the purpose of enhancing aerodynamic performance and reducing thrust force, both of aerodynamic and structural designs should be considered. From the perspective of aerodynamic design, thrust force, power performance, aerodynamic efficiency, and AEP (Annual Energy Production) are the important, and from the perspective of structural design, composite material lay-up, mass, stiffness, buckling stability, and ultimate and fatigue loads are concerned. As of now, software that is capable of creating optimum aerodynamic design and analysis for Multi-MW class wind blades at a time has not been developed. Excel-based BOT (Blade Optimization Tool) developed by ECN in Netherlands is the only design tool that is comparable to such purpose [2]. Because BOT is not capable of designing initial shapes from a blank state and can only optimize aerodynamic shapes of existing blades, it cannot be considered as a complete aerodynamic design tool. Therefore, the goal of this research is to develop integrated userfriendly software which can make blade initial shape, power and load analysis, aerodynamic optimum design, and pitch and torque control strategy.
B. Kim et al. / Renewable Energy 54 (2013) 166e172
167
T am a0m aT cm fm
thrust force local axial flow induction factor local tangential flow induction factor critical axial induction factor local chord length local tip loss factor atipfoil angle of attack at maximum lift to drag ratio of tip airfoil h mechanical efficiency of drive train qlimit the upper limit of twist angle qm local twist angle ldesign design tip speed ratio ɸm local inflow angle r air density Ublade,min the lowest blade rotational speed Ublade,max the upper limit of blade rotational speed Ublade,rated rated blade rotating speed
Nomenclature Cl,tipfoil lift coefficient of a tip airfoil drag coefficient CD lift coefficient CL CP,expected expected aerodynamic power coefficient thrust coefficient CT empirical thrust coefficient when a ¼ 1 CT1 hub diameter Dhub rotor diameter Drotor gear ratio Gratio Irange,max the upper limit of inverter range Irange,min the lower limit of inverter range N number of blade rated power Prated Q torque radius Rblade local Reynolds number Rem
2. Developement of an optimum BLADE design tool
Ublade;min ¼
Urated;gen Irange;min
At the baseline design stage, the design parameters presented in Table 1 such as desired power output, rated wind speed, assumed efficiency, maximum twist angle limitation, tip speed restriction, inverter range, rated speed of generator, and gear ratio need to be set. Next, based on input parameters from Table 1, diameter of the blade Drotor, rated speed Ublade,rated, design tip speed ratio ldesign, min/max values of rotor speed for producing power Ublade,min and Ublade,max can be calculated according to Eqs. (1)e(5), but designating a specific diameter is also possible if blade length is predetermined.
Drotor ¼
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8Prated 3 hCP;estimated rpVrated
Ublade;rated ¼
ldesign ¼
Vtip;rated Drotor orDuser =2
h atipfoil
CL,tipfoil
r
Dhub Drotor
n qlim
Gratio
(5)
After confirming initial design parameters, baseline blade shape is generated by calculating chord length and twist angle at every calculation point set by the designer. These points can be defined either on equal intervals or on unequal intervals as in Fig. 1. The calculation is performed according to Eqs. (6)e(10), and the detailed process is like below [3]. 1) Calculate tip loss factor, fm,tip
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 1 2 ðldesign mÞ 1 þ ð1aÞ2 A C B @ N2 ð1mÞ=m B C 2 B C ftip;m ¼ cos1 Be C B C p @ A 0
(6)
(2) 2) Calculate axial flow induction factor, am
! (3)
am ¼
1 1 1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ fm 1 fm þ fm2 3 3 3
(7)
3) Check final convergence value after iterative calculations for fm,tip and am 4) Calculate tangential flow induction factor, a0m
Table 1 Initial design parameters.
Shape parameters N Prated Vdesign Vrated CP,estimated
Urated;gen Irange;max
0
(1)
ðDrotor orDuser Þ=2 Ublade;rated Vdesign
Parameter
Ublade;min ¼
(4)
Gratio
2.1. Blade Design Module(BDM)
Unit e kW m/s m/s e e e kg/m3 m m m2/s
Parameter Operation parameters Vtip,rated R
Urated,gen Irange,min Irange,max Gratio Calculated parameters R
Ublade,rated ldesign Ublade,min Ublade,max
Unit m/s m rpm % % e m rpm e rpm rpm
am 1 a0m ¼
am
!
ftip;m
(8)
l2design m2
5) Calculate chord length, cm
2
cm ¼
4ldesign m2 a0m R 2p rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ffi Nldesign CL;tipfoil 2 1 am þ ldesign m 1 þ a0m
(9)
168
B. Kim et al. / Renewable Energy 54 (2013) 166e172
Fig. 1. Unequally spaced calculation section arrangement.
6) Calculate twist angle, qm
2.2. Performance analysis module(PAM)
qm ¼ fm atipfoil
(10)
After this procedure, chord length and twist angle distribution along blade spanwise direction can be obtained, and it is easily found that the values of the chord length cm are unrealistically high in the inboard region. In the aerodynamic aspect, the root region does not contribute to aerodynamic torque generation at all due to its circular shape, therefore, the chord length around the blade root must be reduced to practical ranges for the manufacturing and economical purpose. However, blade root is exposed to high bending moments which could cause aero-structural problem. Thus, the result of additional structural analysis must be considered to determine proper chord length around blade root to secure structural stability. Most of blade torque is generated at the spanwise points within 65%e95% ranges from the root, thus chord lengths in those regions should be maintained during chord linearizing process. Generally, five to seven types of airfoils are used in blade design, and the designer can create additional elements between existing airfoils(Table 2). The chord lengths and twist angles for the newly-created elements are updated by interpolating nearby values. Later in the performance analysis stage, aerodynamic properties (CL, CD) of the airfoils are required for all elements. Securing aerodynamic properties that reflect the exact Reynolds number at each calculation elements is important to predict power performance and thrust force precisely. Therefore, the software calculates Reynolds numbers at all elements by using the Eqs. (11)e (13) and the designer can check them through the GUI. The final shape design variables cm and qm can be plotted shown in Fig. 2, and can also be exported to Excel files. Finally, all numerical data and graphs obtained from Blade Design Module can be easily verified automatically through the report-printing function.
Vrot ¼ r Ublade;rated Vlocal ¼
(11)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 þ V2 Vrot design
Rem ¼
cm Vlocal
Performance analysis is needed in order to find out whether the aerodynamic design of blade obtained from BDM (Blade Design Module) meets the design objectives. Generally, the analysis is performed for torque, power output, efficiency, and thrust force using the BEM (Blade Element Momentum) theory. The results of the analysis, which depend on how accurately the flow induction factor is calculated, are obtained by the following iterative calculation process. Flow induction factors am and a0m are assumed to be zero as initial guess at step 1, and calculated results after step 2e6 are checked whether they are converged or not. If so, it proceeds to step 8, and otherwise it goes back to step 2 and flow induction factors are calculated again with updated initial values. Performance analysis should be conducted independently at all calculation points in blade spanwise direction that were determined in the blade design stage. Local torque and thrust force at each element are calculated by using local aerodynamic properties and the overall torque and thrust force can be yielded by integrating local values along blade spanwise direction. Hence, securing reliable aerodynamic data at each calculation point is indispensable to improve the accuracy of performance analysis. If the rotor is assumed to a simple actuator disk as in Fig. 3, the force due to the pressure difference is equal to the force generated by change of momentum across the disk. Then wake velocity Uwake can be induced as (12a)UN. However, if the axial flow induction factor of the blade is predicted to be over 0.5, Uwake obtains a value of 0 or less at step 8, and then the momentum theory is not valid any more. In that condition, the calculated value of the thrust coefficient significantly differs from the experimental results as
(12)
(13)
n
Table 2 Arrangement of 2-D airfoils. Section no.
Position [m]
Airfoil
1 2 3 4 5 6 7 8 9
0.0272 0.0928 0.1574 0.2326 0.3177 0.4153 0.5305 0.6585 1.0000
Cylinder Cylinder DU-401 DU-350 DU-300 DU-250 DU-212 NACA64618 NACA64615
Fig. 2. Example of calculated chord length and twist angle.
B. Kim et al. / Renewable Energy 54 (2013) 166e172
169
6) Calculate the flow induction factors (Eqs. (19)e(21))
am ¼
g1 1 þ g1
(19)
a0m ¼
g2 1 g2
(20)
Ncm CL cosfm þ CD sinfm H g1 ¼ 2pr 4floss;m sin2 fm Ncm CL sinfm CD cosfm g2 ¼ 2pr 4floss;m sinfm cosfm
Fig. 3. Flow model of momentum theory.
shown in Fig. 4 and it is called ‘Momentum Theory Breakdown’. To compensate it, the value of am must be corrected by empirical correlation such as Eq. (22) if the calculated axial flow induction factor at a local element exceeds a specific value, am aT [4,5]. 1) Initialize am, a0m ¼ 0 2) Compute the inflow angle (Eq. (14))
1 am
tanfm ¼
(14)
lm 1 þ a0m
(15)
4) Compute the tip and hub loss coefficients (Eqs. (16) and (17))
ftip;m ¼
fhub;m ¼
2
N ðRrÞ acos e 2 rjsinfm j
p
2
NðrRhub Þ acos e 2 rjsinfm j
p
for a aT, H ¼ 1.0;
for a>aT ; H ¼
(16)
1 pffiffiffiffiffiffiffiffi CT1 2
(22)
9) Compute local and overall thrust force (Eq. (23))
dT ¼
1 rNW 2 CL cosfm þ CD sinfm cm dr 2
(23)
10) Compute local and overall torque (Eq. (24))
dQ ¼ (17)
4að1 aÞ pffiffiffiffiffiffiffiffi CT1 4 CT1 1 ð1 aÞ
7) Check the convergences of flow induction factor. If the changes exceed certain tolerance level, go to step 2. 8) Update flow induction factors considering the effects of momentum theory breakdown (Eq. (22))
aT ¼ 1
3) Compute the angle of the attack (Eq. (15))
am ¼ fm b þ qset þ qm
(21)
1 rNW 2 r CL sinfm CD cosfm cm dr 2
(24)
11) Compute overall power output (Eq. (25)) 5) Compute the solidity (Eq. (18))
sm ¼
Ncm 2pr
dPaero ¼ Ublade;rated dQ (18)
(25)
Thorough the above procedure, all values can be plotted and saved as an Excel file. Also, all numeric data and graphs that are generated at performance analysis stage can easily be verified automatically through report-printing function.
2.3. Pitch and torque control module(PTM)
Fig. 4. Example of momentum theory breakdown.
Torque control map provides the information to determine rotor speed of maximum efficiency for a specific wind speed at below rated condition. The calculation points for torque control map are determined by a combination of the blade rotating speed and the wind speed. The number of the calculation points for rotating speed is determined by DU, which divides the initial Ublade,mine Ublade,max in Table 1, while the number of calculation point for wind speed is depending on DV, which divides the interval Vcut-ine Vcut-out. Finally, all calculation points needed for calculating the torque control map are shown in Fig. 5. When the calculation point
170
B. Kim et al. / Renewable Energy 54 (2013) 166e172
user-friendly environment as in Fig. 7. After thrust clipping is applied, performance analysis must be carried out again to check the effects of thrust force restriction and to obtain new pitch control map. 2.4. Blade optimization module(BOM)
is determined after combining the wind speed and the rotating speed, values for the torque, power output, efficiency, and the thrust force along actual operating line can be calculated according to PAM in Section 2.2, and the torque control map as shown in Fig. 6 is obtained. At above the rated wind speed range, the optimum pitch angle of the blade to regulate power production is calculated by a combination of the changes in the wind speed and the pitch angle while rotating speed is kept constant. After the wind speed range of Velec, ratedeVcut-out is divided by, DV, the blade pitch angle for each given wind speed that produces rated power output by changing the blade pitch angle from 0 to 90 in 1 increments. Subsequently, PAM can calculate all aerodynamic properties including torque, power output, efficiency, thrust force, and the pitch angle at all wind speeds. During the analysis of PAM, the change in the pitch angle of the blade can be realized by changing the qm in Eq. (15). The blade designer must consider the inhibition of unnecessary instantaneous peak thrust force that occurs in a narrow wind speed interval near the rated wind speed and it can be reduced by controlling pitch angle of the blade [6]. This manipulation is called thrust clipping or peak shaving and the software developed in this research can limit the maximum thrust force as desired level in
The initial blade shape that had completed by BDMePAMe PTM, the aerodynamic optimum design process can be conducted in order to maximize power and efficiency. Several restricting options such as maximum chord length and twist angle limitation, and axial thrust force ratio comparing to initial blade can be chosen by designer. Research on optimization of the aerodynamic design of the blade is usually done by BEM, which uses the flow induction factor as suggested by Glauert [7]. To optimize aerodynamic shape at a designated TSR, the objective function is the local torque at each calculation points, and chord length and twist angle are defined as independent variables. The maximum overall aerodynamic efficiency can be obtained by collecting the best combination of chord length and twist angle which shows the highest local torque at each calculation points. An increment of chord length and twist angle for optimizing process can be adjusted by designer, and the smaller increment could lead better results. Maalawi optimized the 100 kW class blade rotors with code length and twist angle variation and Lanzafame predicted the performance of the NREL Phase VI based on the results of the experiment on S809 airfoil [8,9]. However, the blade models used for optimization in the two researches were applied a single airfoil, which is different from the contemporary large blades, which applies five to seven types of airfoil in order. The BEM theory calculates the overall performance by integrating local torques along spanwise direction of the blade. During this process, mass and momentum interactions in blade spanwise direction among the elements are ignored. Therefore, it can be thought that the power performance can be maximized when every local torque has its maximum value. Thus, the final goal of blade aerodynamic shape optimization is to find optimal combinations of chord lengths and twist angles that secure the maximum performance at every local element. Minimum and maximum values for search range of the chord length and the twist angle are defined by the designer, as shown in Fig. 8. Optimum aerodynamic design is carried out at a certain TSR (tip speed ratio) and the calculating range and cases should be
Fig. 6. Example of torque control map.
Fig. 7. Definition of thrust clipping rate.
Fig. 5. Definition of calculation points.
B. Kim et al. / Renewable Energy 54 (2013) 166e172
171
Fig. 9. Aerodynamic coefficients of the airfoils.
3. Optimal design and verification of the calculated results 3.1. Blade model for verification and validation
Fig. 8. Examples on the final optimum blade geometry.
chosen properly to reduce calculation time. Additionally, the designer can choose the objectives of optimization such as securing high performance, minimizing thrust force, and limiting the range for chord length and twist angles by comparing with the initial blade models. Since various types of airfoils are applied to large blades and optimizing calculation is conducted respectively for local elements, the distributions of chord lengths and the twist angles do not show continuous tendency after the initial optimization, especially for airfoil transition region [10]. To solve this problem, an option to limit the change rate of the chord length and the twist angle of the neighboring elements is available and more than five orders of curve fitting for chord length and twist angle distribution is indispensable to make blade shape smooth globally.
In order to assess the accuracy of the performance analysis module of the software developed in this research, we chose the 5 MW blade, developed by NREL, as the reference model and compared the performance analysis results to those of GH-Bladed, conventional system load analysis software. The specification for reference 5 MW model is listed in Table 3 [11]. Because the results of the performance analysis performed by NREL were obtained through aero-elastic analysis after comprehensive modeling of the 5 MW wind turbine system, they cannot be compared directly to results from performance analysis module of this software, which only analyzes blades. Hence, in order to obtain static power curve of the blade for comparing the results under same conditions, shape information of 5 MW blade and aerodynamic data were used as proposed by NREL in Table 3 and Fig. 9. With this information, aerodynamic performance analysis under normal conditions in GHBladed is carried out. Also, in order to verify the optimum shape design module of the software developed in this research, the specifications for the reference model Drotor ¼ 126 m, l ¼ 7.8,
Table 3 Aerodynamic properties of verification model. Section
Position [m]
Chord [m]
Twist [ ]
Airfoil
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
2.8667 5.6000 8.3333 11.7500 15.8500 19.9500 24.0500 28.1500 32.2500 36.3500 40.4500 44.5500 48.6500 52.7500 56.1667 58.9000 61.6333
3.542 3.854 4.167 4.557 4.652 4.458 4.249 4.007 3.748 3.502 3.256 3.010 2.764 2.518 2.313 2.086 1.419
13.308 13.308 13.308 13.308 11.48 10.162 9.011 7.795 6.544 5.361 4.188 3.125 2.319 1.526 0.863 0.370 0.106
Cylinder.1 Cylinder.1 Cylinder.2 DU40_A17 DU35_A17 DU35_A17 DU30_A17 DU25_A17 DU21_A17 DU21_A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17
Fig. 10. Comparison of aerodynamic power coefficients.
172
B. Kim et al. / Renewable Energy 54 (2013) 166e172 Table 4 Performance analysis results.
Baseline Optimized Smoothed
CP
CT
l ¼ 7.8
l ¼ 7.8
CP,max e
0.4798 0.4900 0.4889
0.8635 0.7976 0.7989
0.4822 at l ¼ 7.5 0.4904 at l ¼ 7.95 0.4903 at l ¼ 8
5 MW blade. Baseline efficiency of the blade increased approximately by 1%, while the thrust coefficient decreased by about 7.5%.
4. Conclusion
Fig. 11. Comparison of chord and twist angle distribution.
Vrated ¼ 11.4 m/s, Gratio ¼ 1:97 were introduced as input parameters for a new 5 MW blade design, which were used to compare the distribution of chord lengths and the twist angles with the reference model. 3.2. Comparison of design and analysis results The performance analysis results obtained from the software in this research are shown in Fig. 10. The results were identical to the results predicted by GH-Bladed and was shown to be CP,max ¼ 0.49 at l ¼ 7.8. Fig. 11 shows the results of comparison between the chord length change in the reference model and the newly-designed 5 MW blade, for the purpose of verifying the validity of design application of BDM and BOM. Even though it has the same diameter with the reference model, it generated 5 MW of power at same rated wind speed and achieved the maximum CP of 0.49 at l ¼ 7.8 even though it has smaller chord length distribution globally. Also, even though the new model can produce the same power with same efficiency like the reference model, it has lower solidity and is thus considered to be a better model in terms of mass and aerodynamic loads. Table 4 lists the final results of the performance analysis of
In this research, a unique software for designing optimum shape and analyzing aerodynamic performance of multi-MW wind turbine blades has been developed, and the verification of its performance analysis module and the aerodynamic shape design module is conducted. The performance analysis results between GH-Bladed and our software were identical in the entire region. The maximum CP of 0.49 is obtained at l ¼ 7.8. To check the validity of the blade optimizing module, the optimum design of the new 5 MW blade was conducted using the same specification for reference model as input parameters. It is found that a model with smaller solidity than reference model can be designed and it produced the same output and efficiency. Through optimization of blade design, efficiency increased by 1% while the thrust coefficient decreased by 7.5%. If the optimum aerodynamic design method is applied to a shape of well-designed blade, significant improvements in power and efficiency cannot be expected. However, it can achieve outstanding results if it is used when design conditions have changed or used to decrease the thrust force and mass of blade under the same conditions.
References [1] BTM Consult. International wind energy developmentesupply chain assessment 2010e2013. BTM Consult ApS; Jan. 2010. [2] Bot ETG, Ceyhan O. Blade optimization tool user manual; 2009. ECN-E-09-092. [3] Kim BS, Kim WJ, Bae SY, Han SG, and Kim ME. “A study on the aerodynamic design of a 3MW wind turbine blade: part I”, Proc. KWEA 2009; Nov. 2009. [4] Hansen Martin, OL. Aerodynamics of wind turbines. 2nd ed. London: Earthscan; 2008. [5] Tony B, David S, Nick J, Ervin B. Wind energy handbook. WILEY; 2006. [6] Kooijman HJT, Lindenberg C, Winkelaar D, van der Hooft EL. Aero-elastic modeling of the DOWEC 6 MW pre-design in PHATAS; 2003. DOWEC-F1W2HJK-01-046/9. [7] Glauert H. The elements of aerofoil theory and airscrew theory. Cambridge University Press; 1959. [8] Maalawi K, Badawy M. A direct method for evaluating performance of horizontal axis wind turbines. Renewable and Sustainable Energy Reviews 2001: 175e95. [9] Lanzafame R, Messina M. Fluid dynamics wind turbine design: critical analysis, optimization and application of BEM theory. Renewable Energy 2007;32: 2291e305. [10] Kim BS, Kim WJ, Bae SY, Han SG and Kim ME. “A study on the aerodynamic design of a 3MW wind turbine blade: part III”, Proc. KWEA 2009; Nov. 2009. [11] Jonkman J, Butterfield S, Musial W, Scott G. Definition of a 5-MW reference wind turbine for offshore system development; 2009. NREL/TP-500-38060.
Contents lists available at SciVerse ScienceDirect
Renewable Energy journal homepage: www.elsevier.com/locate/renene
Developement and verification of a performance based optimal design software for wind turbine blades Bumsuk Kim a, Woojune Kim a, Sanglae Lee a, Sungyoul Bae a, Youngho Lee b, * a b
Green and Industrial Technology Center, Korean Register, 23-7 Jang-dong Yuseong-gu Daejeon, Republic of Korea Division of Mechanical & Energy System Engineering, Korea Maritime University, 1Dongsam-dong Youngdo-gu, Busan, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 26 January 2012 Accepted 13 August 2012 Available online 7 September 2012
In this research, we developed software for designing the optimum shape of multi-MW wind turbine blades and analyzing the performance, and it features aerodynamic shape design, performance analysis, pitchetorque analysis and shape optimization for wind turbine blades. In order to verify the accuracy of the performance analysis results of the software developed in this research, we chose the 5 MW blade, designed by NREL, as the comparison model and compared with the analysis results of well known commercial software (GH-Bladed). The calculated performance analysis results of GH-Bladed and our software were consistent in all values of CP in all l ranges. Also, to confirm applicability of the optimum design module, the optimum design of the new 5 MW blade was performed using the initial design data of the comparison model and found that solidity was smaller in our design even though it produced the same output and efficiency. Through optimization of blade design, efficiency increased by 1% while the thrust coefficient decreased by 7.5%. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Wind turbine blade Aerodynamic design Performance analysis Optimal design
1. Introduction Most of the leading wind turbine manufacturers consider blades as their key components of wind turbine system and have concentrated their efforts on developing their own blade design and increasing the supply through in-house production facility. As world wind energy market grows continuously, a number of independent blade manufacturers are emerged recently, especially in China. However, wind blade market is still dominated by the leading wind turbine system manufacturers and a giant independent blade supplier, LM Wind-power [1], and their overall share in world blade market is estimated around 80 percent. Even though the total amount of blade supply has increased due to the new blade manufacturers, only a few of them have ability to produce large blades bigger than 3 MW class. Therefore, supply shortage of large blades is predicted as global offshore wind-power market expands, and blade could be a bottle neck component for large wind turbine system. In Korea, some blade manufacturers have succeeded in producing 2e3 MW class prototype blades by collaborating with foreign engineering companies. So as to enter the market successfully, securing the fundamental technology on design and production of multi-MW class blades is indispensable. * Corresponding author. E-mail address: [email protected] (Y. Lee). 0960-1481/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.renene.2012.08.029
It is required to take into account the external environmental conditions for a wind blade design that can be occurred during the design life time. Therefore, design evaluations by aero-elastic simulation, CFD analysis, and FEM analysis should be conducted interactively to secure safety and reliability of blades and to satisfy the requirements of international standards. So as to optimize blade shape for the purpose of enhancing aerodynamic performance and reducing thrust force, both of aerodynamic and structural designs should be considered. From the perspective of aerodynamic design, thrust force, power performance, aerodynamic efficiency, and AEP (Annual Energy Production) are the important, and from the perspective of structural design, composite material lay-up, mass, stiffness, buckling stability, and ultimate and fatigue loads are concerned. As of now, software that is capable of creating optimum aerodynamic design and analysis for Multi-MW class wind blades at a time has not been developed. Excel-based BOT (Blade Optimization Tool) developed by ECN in Netherlands is the only design tool that is comparable to such purpose [2]. Because BOT is not capable of designing initial shapes from a blank state and can only optimize aerodynamic shapes of existing blades, it cannot be considered as a complete aerodynamic design tool. Therefore, the goal of this research is to develop integrated userfriendly software which can make blade initial shape, power and load analysis, aerodynamic optimum design, and pitch and torque control strategy.
B. Kim et al. / Renewable Energy 54 (2013) 166e172
167
T am a0m aT cm fm
thrust force local axial flow induction factor local tangential flow induction factor critical axial induction factor local chord length local tip loss factor atipfoil angle of attack at maximum lift to drag ratio of tip airfoil h mechanical efficiency of drive train qlimit the upper limit of twist angle qm local twist angle ldesign design tip speed ratio ɸm local inflow angle r air density Ublade,min the lowest blade rotational speed Ublade,max the upper limit of blade rotational speed Ublade,rated rated blade rotating speed
Nomenclature Cl,tipfoil lift coefficient of a tip airfoil drag coefficient CD lift coefficient CL CP,expected expected aerodynamic power coefficient thrust coefficient CT empirical thrust coefficient when a ¼ 1 CT1 hub diameter Dhub rotor diameter Drotor gear ratio Gratio Irange,max the upper limit of inverter range Irange,min the lower limit of inverter range N number of blade rated power Prated Q torque radius Rblade local Reynolds number Rem
2. Developement of an optimum BLADE design tool
Ublade;min ¼
Urated;gen Irange;min
At the baseline design stage, the design parameters presented in Table 1 such as desired power output, rated wind speed, assumed efficiency, maximum twist angle limitation, tip speed restriction, inverter range, rated speed of generator, and gear ratio need to be set. Next, based on input parameters from Table 1, diameter of the blade Drotor, rated speed Ublade,rated, design tip speed ratio ldesign, min/max values of rotor speed for producing power Ublade,min and Ublade,max can be calculated according to Eqs. (1)e(5), but designating a specific diameter is also possible if blade length is predetermined.
Drotor ¼
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8Prated 3 hCP;estimated rpVrated
Ublade;rated ¼
ldesign ¼
Vtip;rated Drotor orDuser =2
h atipfoil
CL,tipfoil
r
Dhub Drotor
n qlim
Gratio
(5)
After confirming initial design parameters, baseline blade shape is generated by calculating chord length and twist angle at every calculation point set by the designer. These points can be defined either on equal intervals or on unequal intervals as in Fig. 1. The calculation is performed according to Eqs. (6)e(10), and the detailed process is like below [3]. 1) Calculate tip loss factor, fm,tip
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 1 2 ðldesign mÞ 1 þ ð1aÞ2 A C B @ N2 ð1mÞ=m B C 2 B C ftip;m ¼ cos1 Be C B C p @ A 0
(6)
(2) 2) Calculate axial flow induction factor, am
! (3)
am ¼
1 1 1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ fm 1 fm þ fm2 3 3 3
(7)
3) Check final convergence value after iterative calculations for fm,tip and am 4) Calculate tangential flow induction factor, a0m
Table 1 Initial design parameters.
Shape parameters N Prated Vdesign Vrated CP,estimated
Urated;gen Irange;max
0
(1)
ðDrotor orDuser Þ=2 Ublade;rated Vdesign
Parameter
Ublade;min ¼
(4)
Gratio
2.1. Blade Design Module(BDM)
Unit e kW m/s m/s e e e kg/m3 m m m2/s
Parameter Operation parameters Vtip,rated R
Urated,gen Irange,min Irange,max Gratio Calculated parameters R
Ublade,rated ldesign Ublade,min Ublade,max
Unit m/s m rpm % % e m rpm e rpm rpm
am 1 a0m ¼
am
!
ftip;m
(8)
l2design m2
5) Calculate chord length, cm
2
cm ¼
4ldesign m2 a0m R 2p rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2ffi Nldesign CL;tipfoil 2 1 am þ ldesign m 1 þ a0m
(9)
168
B. Kim et al. / Renewable Energy 54 (2013) 166e172
Fig. 1. Unequally spaced calculation section arrangement.
6) Calculate twist angle, qm
2.2. Performance analysis module(PAM)
qm ¼ fm atipfoil
(10)
After this procedure, chord length and twist angle distribution along blade spanwise direction can be obtained, and it is easily found that the values of the chord length cm are unrealistically high in the inboard region. In the aerodynamic aspect, the root region does not contribute to aerodynamic torque generation at all due to its circular shape, therefore, the chord length around the blade root must be reduced to practical ranges for the manufacturing and economical purpose. However, blade root is exposed to high bending moments which could cause aero-structural problem. Thus, the result of additional structural analysis must be considered to determine proper chord length around blade root to secure structural stability. Most of blade torque is generated at the spanwise points within 65%e95% ranges from the root, thus chord lengths in those regions should be maintained during chord linearizing process. Generally, five to seven types of airfoils are used in blade design, and the designer can create additional elements between existing airfoils(Table 2). The chord lengths and twist angles for the newly-created elements are updated by interpolating nearby values. Later in the performance analysis stage, aerodynamic properties (CL, CD) of the airfoils are required for all elements. Securing aerodynamic properties that reflect the exact Reynolds number at each calculation elements is important to predict power performance and thrust force precisely. Therefore, the software calculates Reynolds numbers at all elements by using the Eqs. (11)e (13) and the designer can check them through the GUI. The final shape design variables cm and qm can be plotted shown in Fig. 2, and can also be exported to Excel files. Finally, all numerical data and graphs obtained from Blade Design Module can be easily verified automatically through the report-printing function.
Vrot ¼ r Ublade;rated Vlocal ¼
(11)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 þ V2 Vrot design
Rem ¼
cm Vlocal
Performance analysis is needed in order to find out whether the aerodynamic design of blade obtained from BDM (Blade Design Module) meets the design objectives. Generally, the analysis is performed for torque, power output, efficiency, and thrust force using the BEM (Blade Element Momentum) theory. The results of the analysis, which depend on how accurately the flow induction factor is calculated, are obtained by the following iterative calculation process. Flow induction factors am and a0m are assumed to be zero as initial guess at step 1, and calculated results after step 2e6 are checked whether they are converged or not. If so, it proceeds to step 8, and otherwise it goes back to step 2 and flow induction factors are calculated again with updated initial values. Performance analysis should be conducted independently at all calculation points in blade spanwise direction that were determined in the blade design stage. Local torque and thrust force at each element are calculated by using local aerodynamic properties and the overall torque and thrust force can be yielded by integrating local values along blade spanwise direction. Hence, securing reliable aerodynamic data at each calculation point is indispensable to improve the accuracy of performance analysis. If the rotor is assumed to a simple actuator disk as in Fig. 3, the force due to the pressure difference is equal to the force generated by change of momentum across the disk. Then wake velocity Uwake can be induced as (12a)UN. However, if the axial flow induction factor of the blade is predicted to be over 0.5, Uwake obtains a value of 0 or less at step 8, and then the momentum theory is not valid any more. In that condition, the calculated value of the thrust coefficient significantly differs from the experimental results as
(12)
(13)
n
Table 2 Arrangement of 2-D airfoils. Section no.
Position [m]
Airfoil
1 2 3 4 5 6 7 8 9
0.0272 0.0928 0.1574 0.2326 0.3177 0.4153 0.5305 0.6585 1.0000
Cylinder Cylinder DU-401 DU-350 DU-300 DU-250 DU-212 NACA64618 NACA64615
Fig. 2. Example of calculated chord length and twist angle.
B. Kim et al. / Renewable Energy 54 (2013) 166e172
169
6) Calculate the flow induction factors (Eqs. (19)e(21))
am ¼
g1 1 þ g1
(19)
a0m ¼
g2 1 g2
(20)
Ncm CL cosfm þ CD sinfm H g1 ¼ 2pr 4floss;m sin2 fm Ncm CL sinfm CD cosfm g2 ¼ 2pr 4floss;m sinfm cosfm
Fig. 3. Flow model of momentum theory.
shown in Fig. 4 and it is called ‘Momentum Theory Breakdown’. To compensate it, the value of am must be corrected by empirical correlation such as Eq. (22) if the calculated axial flow induction factor at a local element exceeds a specific value, am aT [4,5]. 1) Initialize am, a0m ¼ 0 2) Compute the inflow angle (Eq. (14))
1 am
tanfm ¼
(14)
lm 1 þ a0m
(15)
4) Compute the tip and hub loss coefficients (Eqs. (16) and (17))
ftip;m ¼
fhub;m ¼
2
N ðRrÞ acos e 2 rjsinfm j
p
2
NðrRhub Þ acos e 2 rjsinfm j
p
for a aT, H ¼ 1.0;
for a>aT ; H ¼
(16)
1 pffiffiffiffiffiffiffiffi CT1 2
(22)
9) Compute local and overall thrust force (Eq. (23))
dT ¼
1 rNW 2 CL cosfm þ CD sinfm cm dr 2
(23)
10) Compute local and overall torque (Eq. (24))
dQ ¼ (17)
4að1 aÞ pffiffiffiffiffiffiffiffi CT1 4 CT1 1 ð1 aÞ
7) Check the convergences of flow induction factor. If the changes exceed certain tolerance level, go to step 2. 8) Update flow induction factors considering the effects of momentum theory breakdown (Eq. (22))
aT ¼ 1
3) Compute the angle of the attack (Eq. (15))
am ¼ fm b þ qset þ qm
(21)
1 rNW 2 r CL sinfm CD cosfm cm dr 2
(24)
11) Compute overall power output (Eq. (25)) 5) Compute the solidity (Eq. (18))
sm ¼
Ncm 2pr
dPaero ¼ Ublade;rated dQ (18)
(25)
Thorough the above procedure, all values can be plotted and saved as an Excel file. Also, all numeric data and graphs that are generated at performance analysis stage can easily be verified automatically through report-printing function.
2.3. Pitch and torque control module(PTM)
Fig. 4. Example of momentum theory breakdown.
Torque control map provides the information to determine rotor speed of maximum efficiency for a specific wind speed at below rated condition. The calculation points for torque control map are determined by a combination of the blade rotating speed and the wind speed. The number of the calculation points for rotating speed is determined by DU, which divides the initial Ublade,mine Ublade,max in Table 1, while the number of calculation point for wind speed is depending on DV, which divides the interval Vcut-ine Vcut-out. Finally, all calculation points needed for calculating the torque control map are shown in Fig. 5. When the calculation point
170
B. Kim et al. / Renewable Energy 54 (2013) 166e172
user-friendly environment as in Fig. 7. After thrust clipping is applied, performance analysis must be carried out again to check the effects of thrust force restriction and to obtain new pitch control map. 2.4. Blade optimization module(BOM)
is determined after combining the wind speed and the rotating speed, values for the torque, power output, efficiency, and the thrust force along actual operating line can be calculated according to PAM in Section 2.2, and the torque control map as shown in Fig. 6 is obtained. At above the rated wind speed range, the optimum pitch angle of the blade to regulate power production is calculated by a combination of the changes in the wind speed and the pitch angle while rotating speed is kept constant. After the wind speed range of Velec, ratedeVcut-out is divided by, DV, the blade pitch angle for each given wind speed that produces rated power output by changing the blade pitch angle from 0 to 90 in 1 increments. Subsequently, PAM can calculate all aerodynamic properties including torque, power output, efficiency, thrust force, and the pitch angle at all wind speeds. During the analysis of PAM, the change in the pitch angle of the blade can be realized by changing the qm in Eq. (15). The blade designer must consider the inhibition of unnecessary instantaneous peak thrust force that occurs in a narrow wind speed interval near the rated wind speed and it can be reduced by controlling pitch angle of the blade [6]. This manipulation is called thrust clipping or peak shaving and the software developed in this research can limit the maximum thrust force as desired level in
The initial blade shape that had completed by BDMePAMe PTM, the aerodynamic optimum design process can be conducted in order to maximize power and efficiency. Several restricting options such as maximum chord length and twist angle limitation, and axial thrust force ratio comparing to initial blade can be chosen by designer. Research on optimization of the aerodynamic design of the blade is usually done by BEM, which uses the flow induction factor as suggested by Glauert [7]. To optimize aerodynamic shape at a designated TSR, the objective function is the local torque at each calculation points, and chord length and twist angle are defined as independent variables. The maximum overall aerodynamic efficiency can be obtained by collecting the best combination of chord length and twist angle which shows the highest local torque at each calculation points. An increment of chord length and twist angle for optimizing process can be adjusted by designer, and the smaller increment could lead better results. Maalawi optimized the 100 kW class blade rotors with code length and twist angle variation and Lanzafame predicted the performance of the NREL Phase VI based on the results of the experiment on S809 airfoil [8,9]. However, the blade models used for optimization in the two researches were applied a single airfoil, which is different from the contemporary large blades, which applies five to seven types of airfoil in order. The BEM theory calculates the overall performance by integrating local torques along spanwise direction of the blade. During this process, mass and momentum interactions in blade spanwise direction among the elements are ignored. Therefore, it can be thought that the power performance can be maximized when every local torque has its maximum value. Thus, the final goal of blade aerodynamic shape optimization is to find optimal combinations of chord lengths and twist angles that secure the maximum performance at every local element. Minimum and maximum values for search range of the chord length and the twist angle are defined by the designer, as shown in Fig. 8. Optimum aerodynamic design is carried out at a certain TSR (tip speed ratio) and the calculating range and cases should be
Fig. 6. Example of torque control map.
Fig. 7. Definition of thrust clipping rate.
Fig. 5. Definition of calculation points.
B. Kim et al. / Renewable Energy 54 (2013) 166e172
171
Fig. 9. Aerodynamic coefficients of the airfoils.
3. Optimal design and verification of the calculated results 3.1. Blade model for verification and validation
Fig. 8. Examples on the final optimum blade geometry.
chosen properly to reduce calculation time. Additionally, the designer can choose the objectives of optimization such as securing high performance, minimizing thrust force, and limiting the range for chord length and twist angles by comparing with the initial blade models. Since various types of airfoils are applied to large blades and optimizing calculation is conducted respectively for local elements, the distributions of chord lengths and the twist angles do not show continuous tendency after the initial optimization, especially for airfoil transition region [10]. To solve this problem, an option to limit the change rate of the chord length and the twist angle of the neighboring elements is available and more than five orders of curve fitting for chord length and twist angle distribution is indispensable to make blade shape smooth globally.
In order to assess the accuracy of the performance analysis module of the software developed in this research, we chose the 5 MW blade, developed by NREL, as the reference model and compared the performance analysis results to those of GH-Bladed, conventional system load analysis software. The specification for reference 5 MW model is listed in Table 3 [11]. Because the results of the performance analysis performed by NREL were obtained through aero-elastic analysis after comprehensive modeling of the 5 MW wind turbine system, they cannot be compared directly to results from performance analysis module of this software, which only analyzes blades. Hence, in order to obtain static power curve of the blade for comparing the results under same conditions, shape information of 5 MW blade and aerodynamic data were used as proposed by NREL in Table 3 and Fig. 9. With this information, aerodynamic performance analysis under normal conditions in GHBladed is carried out. Also, in order to verify the optimum shape design module of the software developed in this research, the specifications for the reference model Drotor ¼ 126 m, l ¼ 7.8,
Table 3 Aerodynamic properties of verification model. Section
Position [m]
Chord [m]
Twist [ ]
Airfoil
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
2.8667 5.6000 8.3333 11.7500 15.8500 19.9500 24.0500 28.1500 32.2500 36.3500 40.4500 44.5500 48.6500 52.7500 56.1667 58.9000 61.6333
3.542 3.854 4.167 4.557 4.652 4.458 4.249 4.007 3.748 3.502 3.256 3.010 2.764 2.518 2.313 2.086 1.419
13.308 13.308 13.308 13.308 11.48 10.162 9.011 7.795 6.544 5.361 4.188 3.125 2.319 1.526 0.863 0.370 0.106
Cylinder.1 Cylinder.1 Cylinder.2 DU40_A17 DU35_A17 DU35_A17 DU30_A17 DU25_A17 DU21_A17 DU21_A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17 NACA64A17
Fig. 10. Comparison of aerodynamic power coefficients.
172
B. Kim et al. / Renewable Energy 54 (2013) 166e172 Table 4 Performance analysis results.
Baseline Optimized Smoothed
CP
CT
l ¼ 7.8
l ¼ 7.8
CP,max e
0.4798 0.4900 0.4889
0.8635 0.7976 0.7989
0.4822 at l ¼ 7.5 0.4904 at l ¼ 7.95 0.4903 at l ¼ 8
5 MW blade. Baseline efficiency of the blade increased approximately by 1%, while the thrust coefficient decreased by about 7.5%.
4. Conclusion
Fig. 11. Comparison of chord and twist angle distribution.
Vrated ¼ 11.4 m/s, Gratio ¼ 1:97 were introduced as input parameters for a new 5 MW blade design, which were used to compare the distribution of chord lengths and the twist angles with the reference model. 3.2. Comparison of design and analysis results The performance analysis results obtained from the software in this research are shown in Fig. 10. The results were identical to the results predicted by GH-Bladed and was shown to be CP,max ¼ 0.49 at l ¼ 7.8. Fig. 11 shows the results of comparison between the chord length change in the reference model and the newly-designed 5 MW blade, for the purpose of verifying the validity of design application of BDM and BOM. Even though it has the same diameter with the reference model, it generated 5 MW of power at same rated wind speed and achieved the maximum CP of 0.49 at l ¼ 7.8 even though it has smaller chord length distribution globally. Also, even though the new model can produce the same power with same efficiency like the reference model, it has lower solidity and is thus considered to be a better model in terms of mass and aerodynamic loads. Table 4 lists the final results of the performance analysis of
In this research, a unique software for designing optimum shape and analyzing aerodynamic performance of multi-MW wind turbine blades has been developed, and the verification of its performance analysis module and the aerodynamic shape design module is conducted. The performance analysis results between GH-Bladed and our software were identical in the entire region. The maximum CP of 0.49 is obtained at l ¼ 7.8. To check the validity of the blade optimizing module, the optimum design of the new 5 MW blade was conducted using the same specification for reference model as input parameters. It is found that a model with smaller solidity than reference model can be designed and it produced the same output and efficiency. Through optimization of blade design, efficiency increased by 1% while the thrust coefficient decreased by 7.5%. If the optimum aerodynamic design method is applied to a shape of well-designed blade, significant improvements in power and efficiency cannot be expected. However, it can achieve outstanding results if it is used when design conditions have changed or used to decrease the thrust force and mass of blade under the same conditions.
References [1] BTM Consult. International wind energy developmentesupply chain assessment 2010e2013. BTM Consult ApS; Jan. 2010. [2] Bot ETG, Ceyhan O. Blade optimization tool user manual; 2009. ECN-E-09-092. [3] Kim BS, Kim WJ, Bae SY, Han SG, and Kim ME. “A study on the aerodynamic design of a 3MW wind turbine blade: part I”, Proc. KWEA 2009; Nov. 2009. [4] Hansen Martin, OL. Aerodynamics of wind turbines. 2nd ed. London: Earthscan; 2008. [5] Tony B, David S, Nick J, Ervin B. Wind energy handbook. WILEY; 2006. [6] Kooijman HJT, Lindenberg C, Winkelaar D, van der Hooft EL. Aero-elastic modeling of the DOWEC 6 MW pre-design in PHATAS; 2003. DOWEC-F1W2HJK-01-046/9. [7] Glauert H. The elements of aerofoil theory and airscrew theory. Cambridge University Press; 1959. [8] Maalawi K, Badawy M. A direct method for evaluating performance of horizontal axis wind turbines. Renewable and Sustainable Energy Reviews 2001: 175e95. [9] Lanzafame R, Messina M. Fluid dynamics wind turbine design: critical analysis, optimization and application of BEM theory. Renewable Energy 2007;32: 2291e305. [10] Kim BS, Kim WJ, Bae SY, Han SG and Kim ME. “A study on the aerodynamic design of a 3MW wind turbine blade: part III”, Proc. KWEA 2009; Nov. 2009. [11] Jonkman J, Butterfield S, Musial W, Scott G. Definition of a 5-MW reference wind turbine for offshore system development; 2009. NREL/TP-500-38060.