Professional Documents
Culture Documents
been developed in some countries. Even though the and structural design procedure, shown in Figure 1, is
large-scale wind turbine can effectively produce electri- proposed to design a vertical wind turbine using the skin-
cal power, the small-scale wind turbine has been contin- spar-foam core sandwich structure having glass/epoxy
uously developed due to some advantages; for instance, skin and spar and a polyurethane foam core.
it can be easily built at a low cost without any limitation
of location, i.e., even in the city. In case of small-scale
wind turbines, the vertical axis wind turbine (VAWT) is
used in cities having frequent wind direction change,
2 Aerodynamic design
even though it has a lower efficiency than the horizontal
The aerodynamic design of the Darrieus-type rotor includ-
axis wind turbine. The VAWT can be classified as a differ-
ing the H-type vertical wind turbine follows Templin’s
ential drag machine like the panemone, Lafond turbine,
theory [8–10]. Its maximal power is approximated by the
Savonious, screened machines, machines with flapping
following expression:
blades, machines with turning blades and machines with
fixed, movable or cyclic pitch blades such as Darrieus [1]. P = 0.25V2(1)
Among them, the Darrieus-type VAWT, first proposed by
where P, S and V are power (W), swept area (m2) and wind
Georges Jean Marie Darrieus, uses aerodynamic lift force.
speed (m/s), respectively.
The aerodynamics, light structure, fatigue and genera-
The swept area is given as a function of geometrical
tion system of the high-performance Darrieus-type wind
parameters by the expression:
turbines have been studied in the USA and Canada [2].
For instance, NCR developed the 4 MW class Darrieus- 8 RH
S= (2)
type VAWT in 1988 [3]. Recently, many studies have been 3
performed on the improvement of bad starting behaviors
where R and H are the maximum radius and half of the
at low wind speed and lower efficiency relative to the hor-
rotor height, respectively.
izontal axis wind turbine system [4]. Furthermore, most
The tip speed ratio λ0 is approximately related to the
small-scale wind turbine systems have been designed at
ratio R/l by the expression
the rated wind speed of approximately 12 m/s, but they
have a great reduction of aerodynamic efficiency in low- 5R
λ0 = (3)
wind-speed regions [5–7]. b1
The objective of this work is to design a high-efficiency
500W class composite VAWT blade that is applicable to where b and l are the number of blades and the chord
relatively low speed regions. In this work, an aerodynamic length, respectively.
Design
requirement
Structure design
Aerodynamic
design Analysis by FEM
Design Design
modification Aerodynamic modification
analysis Blade
manufacture
CFD Aerodynamic
analysis Strutural test
Performance test
Finalization of
design
The power coefficient Cp is defined as In order to validate the proposed aerodynamic design
H 2π program, test results are compared with those of the
2p b1 W 2
Cp = = ∫ ∫ Ct
ρSV1 2 πs = H 0 V
3
ωrdzdθ u
2 (4) Uppsala University H-type VAWT, which can produce a
1
power of 12 kW with three blades. It has an NACA0021
airfoil, blade length of 5 m, turbine radius of 3 m and
where Ct, Wu, V1, ω, r, z and θ are the parallel Lilienthal
chord length of 0.25 m at the rated wind speed of 12 m/s
aerodynamic coefficient, the tangential component of the
[10]. Figure 3 shows the comparison between the pro-
relative wind velocity, the upstream wind velocity, the
posed program results and the Uppsala University VAWT
rotor angular velocity, the vertical axis and the horizontal
test results. Through this comparison, it is confirmed that
location angle of the blade element.
the proposed program agrees well with the test results.
The aerodynamic power P and the electrical power Pe
The optimal aerodynamic design of the 500W class
become respectively by the following expression:
H-type vertical wind turbine rotor is carried out by the
2π
b1 proposed in-house program through a parametric study
2 πs ∫0
P= Ct q ωdθdz (5)
to find an optimal aerodynamic configuration having high
Pe = ηmηgP(6)
1 Data points
where q= , W is the relative wind velocity, ρ is the 0.30 4th order fitting
2 6th order fitting
Program result
air density, ηg is the generator efficiency and ηm is the Estimated Cp 0.25
mechanical efficiency. 0.20
An in-house aerodynamic design program based on
0.15
the theory above is newly coded using Visual Basic. If
the wind speed and design parameters are given as input 0.10
data, the program can calculate the tip speed ratio, power 0.05
coefficient and power considering Reynold’s number.
0
Here, the look-up table of the local Reynold’s number of 1 2 3 4 5 6 7 8
TSR
each airfoil is obtained by computational fluid dynamics
(CFD) analysis. Figure 2 shows the front page of the aero- Figure 2 Front page of the proposed user friendly aerodynamic
dynamic design program. design program of VAWT.
Solidity 0.45
0.40 Airfoil
0.50 0.40 NACA0015
0.35
0.75 NACA0018
0.30 1.00 0.35
NACA0021
0.25 1.25 0.30
0.20 1.50 0.25
Cp
2.00
Cp
0.15 0.20
2.50
0.10 0.15
0.05 0.10 H/R:1.5
H/R:1.5 0.05 Blade number:5
0
Blade number:5
-0.05 0
0 1 2 3 4 5 0 1 2 3 4 5
TSR TSR
0.20 0.20
0.15 0.15
0.10 H/R:1.5 0.10
Solidity:1.5 Blade number:5
0.05 0.05
Solidity:1.5
0 0
0 1 2 3 4 5 0 1 2 3 4
TSR TSR
Figure 3 Comparison between the aerodynamic performance calculation results by the proposed and the test results of Uppsala
University VAWT.
386 C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type
efficiency in both low and high wind speed regions with Table 1 Aerodynamic design results of VAWT.
the following design parameters: number of blades, solid-
ity, airfoil, height-to-radius ratio, etc. Figure 4 shows the Rated aerodynamic power (electric power) 600 W (500 W)
Rated wind speed (m/s) 8
parametric case study results with solidity, airfoil, number
Rated rpm 168
of blades and height-to-radius ratio. Table 1 and Figure Number of blades 5
5 show the optimal aerodynamic design results and the Radius (m) 0.9
designed configuration using the parametric case study. Blade length (m) 2.56
Figure 6 shows the performance estimation of the power Blade chord length (m) 0.27
Airfoil NACA0018
coefficient vs. tip speed ratio and the power vs. wind speed.
Prior to the performance test, the designed wind
turbine is numerically analyzed using a commercial CFD
code, ANSYS CFX. Figure 7 shows the flow pattern obtained
3 L oad case analysis and structural
by CFD analysis using this code. In the analysis, input data design
as an operating condition are the rotational velocity of
167 rpm and the rated wind speed of 8 m/s at sea level. According to IEC 61400-2 design requirements for small
In order to consider the rotating position of the blade, wind turbines [12], a total of eight load cases are defined.
the dynamic moving mesh method is applied. Separate However, this work considers three extreme load cases
disconnecting zones are used for expressing the rotating at normal operation, cutoff condition and standstill and
and stationary regions. The shear stress transport (SST) storm conditions among eight load cases. The main loads
turbulence model based on the κ-ω model is used [11]. acting on the blade are the aerodynamic load and the cen-
The analysis ω shows that the calculation power of 663 W trifugal force. The centrifugal force can be simply calcu-
meets closely the design target rated output of 600 W. lated from rotational speed, and the aerodynamic loads
Figure 4 Parametric case study results with solidity, airfoil, blade number and height to radius ratio.
C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type 387
0.20 Gust condition ± (20 m/s, 40°) Without gust With gust Storm
0.15 Rotational speed 167 rpm 353 rpm Stop
0.10
0.05
0
0 0.5 1.0 1.5 2.0 2.5 3.0 where N, T and cn/ct are the normal aerodynamic force, the
TSR
chordwise aerodynamic force, and the normal tangential
B 1200 aerodynamic coefficient, respectively.
1000 The shear forces and the bending moments can be
defined by integrating the normal and chordwise aero-
800
dynamic force component distributions acting on each
Power (w)
600 section of the blade depending on the wind speed and the
incidence angle in various operating conditions. Table 2
400
shows the definition of various aerodynamic load cases
200 having gust conditions and centrifugal body forces con-
- sidered in the structural design. According to load case
0 2 4 6 8 10
analysis, load case 2 is found to be the most severe condi-
Wind speed (m/s)
tion. Therefore, the structural design is performed in con-
Figure 6 Performance calculation results of the designed wind sideration of load case 2.
turbine. In the structural design, the blade adopts the skin-
spar-foam core sandwich structure concept. The glass
fabric/epoxy composite material, which is supplied by a
defined as the following expressions can be calculated at domestic company, is used for both skin and spar. The
three load cases mentioned in Table 2. bending force is endured by the spar flange layered with
The components of the aerodynamic force in the the ply angle of 0°/90°, and the torsion is endured by the
chordwise direction and normal to the blade element are upper and lower skins layered with the ply angle of ± 45°.
Figure 8 shows the blade structural design concept with
dN = Cnqldz(7)
skin-spar-foam core sandwich. The material properties for
dT = Ctqldz(8) skin, spar and core are presented in Table 3.
388 C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type
The initial design of the composite blade is carried out PCOMP four-node shell elements, which can have good
using the netting rule and the rule of mixture, which were composite structural behaviors. In the FEM analysis, the
used in the previous study [5–7], and then the designed aerodynamic loads calculated by the developed in-house
feature is repeatedly modified by structural analysis results code for the aerodynamic design are directly applied to
using a commercial finite element method (FEM) code, the FEM model, and the centrifugal body force is applied
NASTRAN. The blade section profile shape is constant to the model by the centrifugal body force application
along the spanwise direction because it has the same rela- method of MSC NASTRAN using rotational speed, the cen-
tive wind velocity and angle of attack during operation. The tripetal center and blade mass. The aerodynamic loads
thicknesses including lamination sequences for skin, spar are applied on the aerodynamic centers of the spanwise
and core are the same for easy manufacturing and design. distributed blade airfoils, and the boundary condition is
Therefore, its weight may increase a bit. The lamination assumed as a fixed condition at the joint part between the
sequences of skin, spar and core are shown in Table 4. connection support and the blade root. Figure 9 shows
In the FEM analysis, stresses, strains, tip deflections, the FEM mesh generation results using PCOMP four-node
buckling loads and natural frequencies are found. Skin, shell elements.
spar and core of the blade structure are meshed by 3584 The results show that the highest stresses occur at the
1st ply of the skin and the 13th ply of the spar, respectively.
Skin: Figure 10 shows the spanwise (y axis) stress contour on
Glass/Epoxy fabric
±45° 7ply
the blade illustrated by FEM analysis. Table 5 shows struc-
tural analysis results of the fiber direction at the 1st ply of
the skin and the 13th ply of the spar. In order to check the
resonance possibility, eigenvalue analysis is performed
Spar: Polyurethane foam using FEM code MSC NASTRAN. Figure 11 shows the esti-
Glass/Epoxy fabric mated natural vibration mode shapes and frequencies. To
0/90° 13ply
check the possibility of resonance, the Campbell diagram
Figure 8 Structural design concept of blade with skin-spar-foam is drawn. Because all the five low natural frequencies do
core sandwich. not coincide with the operating rotational speed of 167
rpm, resonance may not occur during operation shown as
Table 3 Mechanical properties of materials used for blade.
Figure 12.
Wind turbine blades should withstand the repeated
Materials
fatigue loads during the required fatigue life operation. In
Glass/epoxy Polyurethane this work, the load spectrum is obtained using the aero-
Property fabric foam
dynamic design program and the wind speed data were
E1, Young’s Modulus 0° (N/mm2) 10,500 60.86 measured at Ganyeoam, Yeosu, Jeollanam-do in Korea.
E2, Young’s Modulus 90° (N/mm2) 10,500 59.86 The measured wind speed data are divided into 13 inter-
G12, In-plane Shear Modulus (N/mm2) 1450 19.18
vals. In the load spectrum analysis, it is assumed that the
ν12, Major Poisson’s Ratio 0.27 0.2
Xt, Tensile Strength 0° (N/mm2) 283.9 2.63 rotor blade load is not produced at a wind speed of < 2 m/s
Xc, Comp. Strength 0° (N/mm2) 184.6 1.41 as well as more than 20 m/s. The rotational speed and
Yt, Tensile Strength 90° (N/mm2) 283.9 2.49 the blade load at various wind speeds are calculated with
Yc, Comp. Strength 90° (N/mm2) 184.6 1.41 the proposed aerodynamic design program. The 20-year
S, In-plane Shear Strength (N/mm2) 15.0 0.71
cycles at each wind speed region are calculated from the
Density (g/cc)
rotational speed and the required fatigue life. The normal-
S, In-plane Shear Strength (N/mm2) 1.705 0.1197
Density (g/cc) ized cycles at each wind speed region are obtained from
Ply thickness (mm) 0.25 12.5 the cycles during 20 years divided by the cycles during
maximum wind speed region of 20–40 m/s. Table 6 shows
the load spectrum obtained by the aerodynamic design
Table 4 Designed lamination sequence and thickness. program and the wind data measured in a local Korean
province, and Figure 13 shows the load spectrum calcu-
Lay-up sequence Thickness (mm)
lated by the proposed load spectrum estimation program
Skin ± 457 (Glass/epoxy fabric) 1.75 with the previously explained procedure using EXCEL.
Spar (0/90)13 (Glass/epoxy fabric) 3.25 The required fatigue life is more than 20 years, and
Core (Polyurethane foam) 40.5 (max)
the fatigue life is estimated by Miner’s rule [12–16]. The
C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type 389
Figure 9 FE mesh generation results using PCOMP 4-node shell elements.
Figure 10 Spanwise (Y-axis) stress contour on the 13th ply of spar for load case 2 (MPa).
cumulative damage is calculated using the S-N curve of From stresses calculated by the load spectrum and
the applied composite materials and the obtained load the S-N curve used for spar shown as Figure 14, the fatigue
spectrum. The fatigue life can be estimated using Miner’s life Ni at average stress level Si can be estimated. If the
rule. number of applied load cycles ni at average stress level
390 C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type
4 M
anufacturing of prototype blade
A B C
and structural test
In manufacturing, the hand lay-up and the matched die
molding methods are applied [3]. In the manufacturing
process, the styrofoam mold is first manufactured using
steel plate templates and hot wires due to economic
reasons, and then glass fabrics for the second mold are
layered up on the styrofoam mold with a special coating.
Again the glass fabrics are layered on the second mold for
the final mold, and then glass fabrics for the upper and
1st mode (1.4 Hz) 2nd mode (4.8 Hz) 3rd mode (5.7 Hz) lower surface skins of the blade are layered on the final
mold according to the structural design result. The cured
D E F upper and lower surface skins are bonded by epoxy, and
then the polyurethane foam is injected into the space
between the upper and the lower skins. After the blade
is completely cured, the proper coating is applied. The
manufacturing process and the first prototype blade are
shown in Figure 15.
The design loads for the static structural test are sim-
ulated by the three-point loading shown in Figure 16. The
test is performed by the structural test equipment shown
in Figure 17. The test results are compared with FEM
structural analysis results. The deflection measured at
4th mode (7.7 Hz) 5th mode (16.6 Hz) 6th mode (17.6 Hz) the blade center is 51 mm, and the estimated deflection at
the same position is 54 mm. The upper and lower surface
Figure 11 Natural vibration mode shapes and frequencies.
spanwise strains measured at 50 mm from the blade
center are 1060 με and 870 με, respectively. Estimated
results at the same locations are 1010 με and 762 με,
Si is given, the required 20-year fatigue life can be con-
respectively. Figure 16 shows static strength test loads
firmed by Miner’s rule: If the blade is accumulatively used
simulated by the three-point loading method, and Figure
at average? stresses Si during ni times, its fatigue life safety
17 shows a picture of the static structural test using three-
is evaluated by the following formula:
point loading. In the structural experimental test, the
k
ni aerodynamic loads are simulated using the three-point
∑ N <1
i= 1
(9)
loading method shown in Figure 15 because the loading
i
actuation system is limited. Therefore, the simulated
where k; total sets of applied load cycles at constant stress bending moment is slightly different from the aerody-
level Si. namic bending moment. It is surmised that the mesured
C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type 391
15 4 P.R.O
Hz
3 P.R.O
10
2 P.R.O
5
1 P.R.O
0
0 20 40 60 80 100 120 140 160 180
RPM
Table 6 Load spectrum obtained by the aerodynamic design program and the wind data measured at Ganyeoam, Yeosu, Jeollanam-do in
Korea.
Wind speed Operating Load/max load Cycles during Normalized cycles (divided by cycles
(m/s) time years (h) 20 years during max wind speed region)
Max Min
600
400
Load (N)
200
0
0 50 100 150 200
-200
-400
Normalized cycle
Figure 13 Normalized load spectrum calculated by the proposed load spectrum estimation program.
392 C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type
1.1
1.0
Normalized maximum stress (S/So)
0.9
0.8
R=0.1 R=0.5
0.7
0.6
0.5
0.4
0.3
0.2
100 101 102 103 104 105 106 107
Cycles to failure
R=0.1 R=0.5
IEC 450/06
5 P
erformance test of prototype
small VAWT
To evaluate the target design performance, the perfor-
mance test of the prototype small VAWT is done using a
test purpose tower, generator, electrical loader, perfor-
mance measuring instruments, etc. Because the wind
tunnel for the aerodynamic test of the full-scale wind
turbine system is not available, a truck is used for the per-
Figure 15 Applied manufacturing process and the first prototype formance test. In order to simulate various wind speeds,
blade.
1000
Aerodynamic load
500 Total static test load
Moment (N.m)
0
0 0.5 1.0 1.5 2.0 2.5 3.0
-500
-1000
Distance from blade tip (m)
Figure 16 Static strength test loads simulated by the 3-point loading method.
C. Kong and H. Lee: Development of a high-efficiency and long-life 500W class H-Darrieus-type 393
1200
Measure power
1000
Estimated power
Power (w)
6th order fitting
600
400
200
0
2 3 4 5 6 7 8 9 10
Wind speed (m/s)
Figure 17 Static structural test of the prototype blade using the
3-points loading. Figure 19 Comparison between test results and estimation results
of electrical power produced by prototype small VAWT.
6 Conclusion
[Wind gauge] [Photo sensor]
References
[1] Kim KH, Lee JO. Trans. KSME B 1979, 3, 60–67. [10] Solum SA, Bernhoff H, Leijon M. Renewable Energy 2008, 33,
[2] Templin RJ. Aerodynamic Performance Theory for the NRC 674–681.
Vertical Axis Wind Turbine. LRT-LA 160. National Research [11] Lee WL, Kim DH, Park GG, Oh MW, Choi HC, Kim YH, Kim SH.
Council: Canada, 1974. In: Proceedings of the 2010 International Symposium, Yanbian
[3] Project Eole – a retrospective. Bulletin, October 1983. University of Science & Technology, 28 June, 2010.
[4] Pourboghrat F. Symposium on Bio-fuels and Wind Energy, [12] IEC 61400-02. Part 2: Design Requirements for Small Wind
Michigan State University, December, 2010. Turbine, IEC, 2006.
[5] Kong C, Bang J, Sugiyama Y. Energy 2005, 30, 2101–2114. [13] IEC 1400-1. Wind Turbine Generator System. Part I. Safety
[6] Kong C, Kim T, Han D, Sugiyama Y. Int. J. Fatigue 2006, 28, Requirement, 1st ed., IEC, 1994.
1382–1388. [14] Seoyoung Tech. Available at: http://www.evsmotor.co.kr.
[7] Kong C, Choi S, Park H. Adv. Compos. Mater. 2011, 20, 105–123. Accessed 11 May 2010.
[8] Eggleston DM, Stoddard, F. Wind Turbine Engineering Design, [15] Miner MA. J. Appl. Mech. 1945, 12, 159–164.
Van Nostrand Reinhold Co. Inc., 1987. [16] Mandell JF, Reed RM, Samborsky DD. Fatigue of Fiberglass
[9] Le Gourieres D. Wind Power Plants: Theory and Design, Wind Turbine Blade Materials, SAND92-7005, Sandia
Pergamon press, 1982. National Laboratories: Albuquerque, NM, 1992.