Energy 30 (2005) 21012114

www.elsevier.com/locate/energy

Structural investigation of composite wind turbine blade

considering various load cases and fatigue life
C. Konga,*, J. Banga, Y. Sugiyamab
a
Division of Aerospace and Naval Architectural Engineering, Chosun University, Gwangju, South Korea
b
Department of Aerospace Engineering, Osaka Prefecture University, Sakai-shi, Japan

Abstract
This study proposes a structural design for developing a medium scale composite wind turbine blade made of
E-glass/epoxy for a 750 kW class horizontal axis wind turbine system. The design loads were determined from
various load cases specified at the IEC61400-1 international specification and GL regulations for the wind energy
conversion system. A specific composite structure configuration, which can effectively endure various loads such
as aerodynamic loads and loads due to accumulation of ice, hygro-thermal and mechanical loads, was proposed.
To evaluate the proposed composite wind turbine blade, structural analysis was performed by using the finite
element method. Parametric studies were carried out to determine an acceptable blade structural design, and the
most dominant design parameters were confirmed. In this study, the proposed blade structure was confirmed to be
safe and stable under various load conditions, including the extreme load conditions. Moreover, the blade adapted
a new blade root joint with insert bolts, and its safety was verified at design loads including fatigue loads. The
fatigue life of a blade that has to endure for more than 20 years was estimated by using the well-known SN linear
damage theory, the service load spectrum, and the Speras empirical equations. With the results obtained from all
the structural design and analysis, prototype composite blades were manufactured. A specific construction process
including the lay-up molding method was applied to manufacturing blades. Full-scale static structural test was
performed with the simulated aerodynamic loads. From the experimental results, it was found that the designed
blade had structural integrity. In addition, the measured results of deflections, strains, mass, and radial center of
gravity agreed well with the analytical results. The prototype blade was successfully certified by an international
certification institute, GL (Germanisher Lloyd) in Germany.
q 2004 Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: C82 62 230 7188; fax: C82 62 230 7188.
E-mail address: cdgong@mail.chosun.ac.kr (C. Kong).
0360-5442/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2004.08.016
2102 C. Kong et al. / Energy 30 (2005) 21012114

1. Introduction

Advancements in wind energy system technology have required the standardization of the main
components of wind turbines to reduce costs. Therefore, standardization states large volume production
and manufacturing methods as well as structural test methods and design rules [1]. In South Korea, since
the introduction of the New and Renewable Source of Energy (NRSE) Development & Promotion Act
in 1987, the Korean government has considerably invested in the development of a new wind turbine
system. One of these development initiatives resulted in the design and manufacturing of an E-glass/
epoxy composite blade for a 750 kW medium-scale horizontal axis wind turbine system (HAWTS) [2],
and it has carried out extensive structural design and analysis, and manufacturing and structural tests on
this medium scale composite blade [38].
The blades of a wind turbine rotor are generally regarded as the most critical component of the wind
turbine system [911]. Therefore, designers must carefully consider the fatigue life of the blades in their
structural design and must test the full-size structure. Structural design requirements, such as minimum
blade tip clearance limit, strain limits along the fiber direction, surface stress limit, and fatigue life time
over 20 years, are specified by the design requirements of the IEC 61400-1 international specification
and the Germanischer-Lloyd (GL) regulations [1214]. Often, the fatigue requirement drives the design
of the primary structural members of a wind turbine because wind turbines must have long operation life
of 2030 years to be cost effective. Fatigue life is usually expressed in term of cycles to failure, which is
the number of repeated significant loads that a structure can endure before cracks appear or start to grow
to an allowable length. Palmgren [15] and Miner [16] proposed the linear damage hypothesis: the stress
cycle, which remains constant throughout a fatigue lifetime, is equal to N, and the fraction of that
lifetime consumed in every cycle is constant and equal to 1/N. Mandel et al. [17] found the SN data,
which were expressed as a power law equation, in mostly unidirectional glass fiber composites. Veers
et al. [18] proposed a computer program FAROW that evaluates the fatigue and reliability of wind turbine
components using structural reliability methods. Bishop et al. [19] showed a fatigue analysis of wind
turbine blades using frequency domain techniques. A fracture mechanics model for fatigue damage
process was proposed by Broek [20].
The effect of material defects, imperfections introduced during the manufacturing process, and
assumptions used in the design and analysis of the component must be fully tested. Mayer [9] presented
various test examples such as specimen tests, component tests for the spar, the blade root joint test, the
T-bolt root attachment test, and full-scale test characterizing static and fatigue behaviors. Delft et al. [21]
carried out a full-scale fatigue test on a laminated wood-epoxy blade to be used in a 500 kW wind
turbine. Also, Inomata et al. [22] tested operating stress and presented its results for the blade of a
prototype 500 kW-wind turbine developed by the NEDO in Tappi Wind Park in Japan. Even though a
full-scale test is quite expensive, the test must be performed to verify the design and manufacturing
procedure as well as to gain confidence of the blades in-service structural integrity.
This study proposed a structural design procedure of a medium scale E-glass/epoxy composite wind
turbine blade. With the proposed procedure, it performed aerodynamic design, dynamic and static load
analyses, structural design and analysis, fatigue life estimation from the random load spectrum, modal
analysis, structural test with the specific test rigs, fixtures and measuring equipments, etc. Moreover,
it showed a new idea for a blade root joint with insert bolts and a light structural configuration such as the
cell box type using the skin-spar-foam sandwich structure not only to minimize the blade weight but also
to get the structural stability against buckling and vibration.
 C. Kong et al. / Energy 30 (2005) 21012114 2103

2. General design consideration

In the present design, the blade had a skin-spar-foam sandwich structure with total span-wise length of
23.3 m, maximum chord of 2.1 m, tip chord of 0.5 m and a twisting angle of 128. For simplicity, the
blade had no coning angle and no blade-to-shaft hinge, but it got a full active pitch control to reduce
cyclic aerodynamic loads. The blade was manufactured with a straight leading edge and a straight
tapered trailing edge to make it easier to build, and the modified thick NACA 63(6)-0xx airfoil was used
at the blade root for good structural performance. In order to get high structural strength and stiffness at
the 0.286R station, the modified NACA 66-0xx (40% of total thickness) airfoil was applied for the upper
surface, and the modified NACA 63-0xx (60% of total thickness) airfoil was used for the lower surface.
However, the NACA 63-218 airfoil was used to keep good aerodynamic performance at the tip of the
blade, and airfoils between the 0.286R station and the tip were linearly interpolated. Fig. 1 shows the
results of the designed aerodynamic configuration. For weight reduction, an insert bolt type joint
structure was introduced, and hub structure had a smooth geometry to avoid low local stress
concentration. Provision for lightning strike protection was inserted in the blade, and a round blade tip
was prepared to minimize wake effect and noise. An appropriate gel coat was applied to the blade to
minimize aerodynamic drag, penetration of humidity and solar deterioration. In operation, the wind
turbine may experience different loads such as aerodynamic, inertial, icing, hygro-thermal and
mechanical loads.
A cell-box type with a skin-spar-foam sandwich construction would be able to sustain these
complex loads quite effectively. Spar flange was manufactured by using a unidirectional E-glass/epoxy
tape; two webs had the sandwich construction with urethane foam core, and G458-fabric skin, similar
to the construction of the outer sandwich shell of the blade. The E-glass/epoxy composite material was
selected because of low cost, appropriate fatigue lifetime, desirable specific strength and stiffness, and
ease of manufacture. Table 1 shows the mechanical properties of materials used [11].

Fig. 1. Sectional airfoil configurations with modified NACA 63(6)-0xx.

2104 C. Kong et al. / Energy 30 (2005) 21012114

Table 1
Material mechanical properties [11]
Material, property UD Tape-GFRP Fabric-GFRP Urethane foam
2
E1 (N/mm ) 35,700 22,147 60.86
E2 (N/mm2) 10,600 2658 59.86
G12 (N/mm2) 2810 1617 19.18
n12 0.324 0.3 0.2
Xt (N/mm2) 711.0 367.3 2.63
Xc (N/mm2) 1200.0 411.0 1.41
Yt (N/mm2) 38.0 40.0 2.49
Yc (N/mm2) 183.0 141.0 1.41
S (N/mm2) 65.7 52.8 0.71
r (g/cm3) 1.80 1.87 0.1197

Structural design requirements such as the minimum clearance limits between the blade tip and the
tower wall, the strain limits along the fiber direction, the surface stress limit and fatigue life time more
than 20 years, were specified by the design requirements of the IEC1400-1 international specification
and the GL regulations [12,14]. Various design load cases including aerodynamic, centrifugal,
gravitational, thermal and ice loads and operating conditions such as stand-by, start up, power
production and shut-down due to faults were analyzed and compared with GL Regulations for the
certification of wind energy conversion systems. For design purpose, four major design load cases
were considered: (1) the rated wind speed condition of 12.5 m/s (load case 1), (2) the rated wind speed
plus the extreme gust condition of 9 m/s (load case 2), (3) the cut out wind speed of 25 m/s plus the
extreme wind speed condition (load case 3) and (4) the stationary limit load condition in storm of 55 m/s
(load case 4). Load cases 13 are simultaneously considered with the condition of wind direction
changes between K40 and C408 at maximum rotational wind speed of 35 rpm in operation.
At aerodynamic loads at the rated wind speed, the flap- and chord-wise bending moments for the
calculated loads (load case 1) were very close to those of the simplified loads as specified by the GL
Regulation, as shown in Fig. 2.

Fig. 2. Comparison between the calculated aerodynamic loads and the simplified aerodynamic loads by GL Regulations in
load case 1.
 C. Kong et al. / Energy 30 (2005) 21012114 2105

3. Structural design and analysis

In the current design approach, the optimal blade structural configuration was determined through a
parametric study by using a finite element method (FEM) [23]. The skin thickness, thickness and width
of the spar flange, and thickness, location, and length of the front and rear spar web were varied until
design criteria were satisfied.
The finite element program used for the structural analysis is a well-known commercial code NISA II
(Numerically Integrated Element for System Analysis) [24]. The element types used for the static stress
analysis are the 3D laminated composite shell element and the 3D laminated sandwich element with 12
nodes. In the study, the most dominant design parameters are as follows: the thickness of the spar flange
and skin for reducing weight, the thickness of the spar flange and the location of the front and rear spar
web for sustaining maximum stress and/or strain, the thickness and the width of the spar flange for the
blade tip clearance and the location of the front spar web for minimizing the distortion angle. Through
linear static stress analysis, modal analysis, buckling analysis and additional stress analysis due to ice
and thermal loading, it was confirmed that the blade structure was safe and stable to operate at various
loads conditions, described above. The blade root joint with the insert bolts is a new design concept.
Furthermore, the finite element stress analysis proved that the joint was safe against the design and
fatigue loading conditions. Figs. 35 shows detailed dimensions at the joint, root, and middle of the
blade. Fig. 7 shows lay-up sequence for the skin and the spar flange of compressive upper blade part.
Table 2 presents the stress analysis results of four design load cases. Fig. 6 shows the stress contours of
s11 along the fiber direction developed in the first ply (G450) of the skin during the design load case 2.
As shown in Table 2, the blade structure satisfies the design criteria such as the blade tip clearance
limit from the tower wall of 50%, strain limits of 0.25% in compression and 0.35% in tension, and
strength safety factor of 2.7. Table 3 presents the results of the modal analysis that calculated the natural
frequencies for the first and second flap- and chord-wise bending modes and the first twisting mode.
There was no resonance between the blade and excitation loadings. The buckling load factors from the
first to the fifth buckling modes of the blade are shown in Table 4. The most important buckling mode
was the first mode (the critical mode) and its load safety factor was 3.43.

4. Fatigue design

Fatigue life was estimated by using the SN linear damage equation, Goodman diagrams, the load
spectrum, and the Speras empirical formulae [25]. Using these methods, a dynamic structural problem
for the fatigue can be converted into a static structural problem [25].

Fig. 3. Cut-view of the blade root station (0.076R0.225R).

2106 C. Kong et al. / Energy 30 (2005) 21012114

Fig. 4. Cross-section of the blade at 0.286R station.

4.1. Fatigue load estimation

Spera proposed the empirical equations, which were based on a set of test data that was broad
enough in scope to include the sizes and types of rotors and towers, the types of terrain, and the types
of wind conditions expected for future HAWT power stations. Calculating dynamic loads with these
equations, therefore, is essentially a process of interpolation rather than extrapolation, and structural-
dynamic computer models are not required. With the equations presented by Spera, the log-normal
probability distributions of the following four important dynamic loads can be calculated from the
basic configuration and site data: (1) blade cyclic flap-wise moments, (2) blade cyclic chord-wise
moments, (3) generator cyclic power density, and (4) rotor cyclic thrust density [25]. However,
because the blade design was considered only in this study, so only blade cyclic flap- and chord-wise
moments will be calculated. Table 5 shows calculation results for blade flap- and chord-wise cyclic

Fig. 5. Cross-section of the blade at 0.598R station.

 C. Kong et al. / Energy 30 (2005) 21012114 2107

Fig. 6. s11 (fiber direction) stress contours in the first ply of the G458 skin (design load case 2).

bending moments (dMy,n and dMz,n), where n is the number of standard deviation (nZ0 for the 50th
percentile load, nZ1 for the 84th percentile load, nZ2 for the 98th percentile load, and nZ3 for
99.9th percentile load).

4.2. Allowable fatigue stress

One of the simplest models for fatigue damage accumulation during repeated cycles of stress is the
linear damage hypothesis proposed by Palmgren and Miner [15,16]. Based on this model, the allowable

Table 2
Results of linear static stress analysis
Load case analysis Case 1 Case 2 Case 3 Case 4
Max. stress (MPa) On skin (1st ply) At xZ0.811, yZ12.76 K16.4 K22.6 K23.8 K14.6
At xZ0.797, yZ13.10 18.2 24.3 26.9 14.7
On spar (15th ply) At xZ0.642, yZ13.10 K53.0 K74.9 K82.1 K39.7
At xZ0.466, yZ13.10 73.5 97.6 105.3 63.1
Max. displacement 1.21 1.64 1.77 1.03
at tip (m)
Clearance from 69 57 55 74
tower (%)
Safety factor (Tsai 4.6 3.5 3.0 7.1
Wu failure criterion)
Max. strain (%) in For tension 0.23 0.31 0.34 0.19
fiber direction For comp. 0.16 0.23 0.25 0.13
x, distance from the leading edge (m); y, distance from the root (m) (K, compressive; C, tensile).
2108 C. Kong et al. / Energy 30 (2005) 21012114

Fig. 7. Lay-up sequence for the skin and the spar flange of compressive upper blade part.

fatigue stress Smaxmax for the loading condition can be calculated by Mandell et al. and Spera [17,25]:
 P K1=a a
0 ns
Smaxmax Z SI Nf Pi i (1)
ni
where S 0 I, modified empirical stress coefficient; si, stress parameter in the cycle; Nf, fatigue lifetime
(3.72!108 cycles); Ni, fatigue lifetime at a constant stress level; and ni, number of cycles applied at
stress level.
Table 3
Results of modal analysis
Mode shape Natural frequency (Hz)
First flap-wise bending (FK1) 1.856
First chord-wise bending (LK1) 2.034
Second flap-wise bending (FK2) 5.280
Second chord-wise bending (LK2) 6.610
First torsional (TK1) 18.08
First tower bending (TBK1) 0.659
 C. Kong et al. / Energy 30 (2005) 21012114 2109

Table 4
Buckling load factors
Mode shape Load factor
1st mode 3.43
2nd mode 3.54
3rd mode 4.28
4th mode 4.39
5th mode 4.60

Therefore, the allowable fatigue stresses based on laboratory specimens were 218.8 MPa for tensile
stress and 366.0 MPa for compressive stresses. For practical values of allowable stresses, knockdown
factors should be considered. In this study, a knockdown factor was assumed to be 0.7; that is, allowable
fatigue stresses were 180.74 MPa for tensile stress and 302 MPa for compressive stress. The value of
Smaxmax becomes the allowable fatigue stress for the loading condition. This method was used to convert
the structural dynamic analysis considering a load into a static structural design problem.

4.3. Evaluation of fatigue life

To compare the allowable fatigue stress for 20 year fatigue life with the estimated fatigue stress
caused by type II fatigue load from the Speras empirical equations, the calculated maximum flap- and
chord-wise bending moments in Table 2 (nZ2) were applied to the finite element model of the wind
turbine rotor blade. According to the results of this analysis, the maximum tensile and compressive
stresses developed in the blade due to cyclic bending were 62.87 and K55.19 MPa, respectively. Since
the safety factors for the allowable fatigue stresses were 3.5 for tensile and 6.6 for compressive, which
were calculated from the allowable fatigue stress and the predicted fatigue stress, the designed wind
turbine blade satisfied the design criteria for fatigue life of 20 years.

5. Manufacturing

In this study, two manufacturing processes were used to make the proposed blade: the contact
molding [9] for the spar webs, a method in which the dry reinforced fiber is laid on a mold and is then
impregnated with resin and consolidated; and the prepreg molding for the skins and the spar flanges, a
method in which the impregnated UD tapes or fabrics is laid on a mold and consolidated with a proper
temperature and pressure. Before manufacturing the blade, the molds should be manufactured. The
master molds for the GFRP molds were made of wood. The total length of the master mold for the
blade was divided into eight pieces, and each piece was manufactured with lots of blade

Table 5
Blade flap- and chord-wise cyclic bending moment calculated by Spera [25]
n 0 1 2 3
dMy,n (kN m) 76.63 119.15 186.33 395.17
dMz,n (kN m) 269.54 276.34 288.67 328.72
2110 C. Kong et al. / Energy 30 (2005) 21012114

Fig. 8. Complete assembled proto type blade.

profile templates. And then the surface of the assembled master mold was well polished and then coated
by a Gel coat. The manufacturing molds were fabricated by laying-up the GFRP fabric prepregs on the
master mold with proper stiffeners. Fig. 8 shows an assembled proto type blade using the proposed
manufacturing process.

6. Structural test

In the static loading test, the load case 1 is discussed. As mentioned, this load case includes the
aerodynamic load at the rated wind speed reached in practical operation including wind direction change
and gust. Moreover, this load will be used in the qualification test of the final full-size blades to be
mass-produced. To simulate the aerodynamic load, three concentrated loads were applied at points
located 17.2, 20.26 and 21.85 m from the blade root [4,8]. Fig. 9 shows the deflected blade under
bending.
A prototype E-glass/epoxy blade was tested statically, and results were compared with those obtained
from the analytical model and the TsaiWu failure criterion. In the bending test of the full-size blade,

Fig. 9. Full-size blade under bending.

 C. Kong et al. / Energy 30 (2005) 21012114 2111

Table 6
Measured and predicted values of the blade
Mass of blade (kg) Span-wise center of Tip deflect. (mm) Flap-wise 1st N.F.
mass (m)
Measured 2951 8.451 1978.6 1.820
Predicted 2883 8.786 1921.0 1.856
Error (%) 2.36 3.81 3.0 2.0

no local failure, debonding, delamination, or local buckling were observed up to the maximum design
load. The previous manufacturing and testing experiences helped this study in testing for material
characterization and improving the final manufacturing process, whose procedures that are now well
established in authors laboratory. Accurate fiber volume fraction (vf) measurements are essential in the
stress analysis calculations since large variations could result inaccurate estimates. In coupon tests,
the average fiber volume fraction of the specimens obtained from an actual blade was 52.5%, which is
less than the 60% suggested by the materials manufacturer. Axial strength and modulus strongly
depended on the fiber volume fraction, so these values were corrected [26]. Table 6 shows both
measured and predicted results for the mass of the blade, the location of center of gravity (span-wise),
the tip defection, and the first flap- natural frequency. The difference between the measured and
predicted values is between 2 and 4%.
Fig. 10 and Table 7 present the measured and predicted deflections at the maximum thickness of the
blade profile along the span-wise stations when the maximum design load is applied. There was a small
difference between the measured and predicted results when the reference fiber volume fraction of 60%
is used in the analysis; however, these values agreed well when the actual fiber volume fraction of 52.5%
used, Table 8.
Fig. 11 and Table 8 show comparison between measured and predicted strain results for the upper part
of the blade. Table 8 presents three different locations at the maximum thickness positions of the
blade profile, 0.236R (5.56 m), 0.493R (11.59 m) and 0.574R (13.43 m) for applied loads of 20, 40, 60,
80 and 100% of the maximum design load. The theoretical values agreed well with experimental
measurements.

Fig. 10. Comparison between measured and predicted deflections along the blade at maximum design load.
2112 C. Kong et al. / Energy 30 (2005) 21012114

Table 7
Comparison between measured and predicted deflections along the blade at maximum design load
Distance from root (m) Measured deflect. (mm) Predicted deflect. VfZ52.5% (mm)
15 525.6 470.9
17 809 737.3
18.13 1012.6 892.1
19.97 1331.9 1215
23.9 1978.6 1921

Table 8
Comparison between measured and predicted strains (311) on the upper part of the blade
Loading Strains at RZ5.56 m [10-6] Strains at RZ11.59 m [10-6] Strains at RZ13.43 m [10-6]
(%) Test FEM Error (%) Test FEM Error (%) Test FEM Error (%)
20 K57 K45 K21 K181 K167 K8 K189 K153 K19
40 K163 K141 K13 K360 K392 9 K368 K350 K5
60 K275 K263 K4 K522 K577 10 K537 K550 2
80 K323 K369 14 K655 K755 15 K687 K730 6
100 K428 K482 13 K878 K966 10 K896 K940 5

7. Conclusion

An E-glass/epoxy composite blade for a 750 kW medium scale HAWTS was designed, manufactured,
tested, and evaluated with the proposed design flow.
The various load cases specified by the IEC61400-1 international specification and GL Regulations
for the wind energy conversion system were considered, and a specific composite structure configuration
that can effectively endure various loads such as aerodynamic and inertial loads, icing loads, hygro-
thermal and mechanical loads was proposed. The structural analysis was performed to evaluate the
proposed design configuration using the FEM.
An acceptable blade structural configuration was determined through the parametric study. The most
dominant design parameters were the thickness of spar flange and skin for minimization of weight,

Fig. 11. Comparison between measured and predicted strains on the upper part of the blade.
 C. Kong et al. / Energy 30 (2005) 21012114 2113

the thickness and the location of front and rear spar webs for maximum stress and strain, the thickness
and the width of spar flange for the blade tip clearance, and the location of front spar web for the
distortion angle. The fatigue life of more than 20 years was estimated by using the well-known SN
linear damage equation, the load spectrum, and Speras empirical equations. Based on the developed
construction process and lay-up molding method, the prototype composite blades were manufacture.
Finally, a full-scale static structural test was carried out at the simulated aerodynamic loads. The
experimental results showed that the designed blade had structural integrity. The predicted mass, span-
wise center of gravity, blade tip deflection and first flap- natural frequency agreed well with the
corresponding measured values within 4% error. Furthermore, the measured strain results had good
agreement with the analytical results.

Acknowledgements

This study was supported by research funds from Chosun University, 2002.

References

[1] Ackermann T, Soder L. Wind energy technology and current status. Renew Sustain Energy Rev 2000;4:31574.
[2] Kim JS. Renewable energy development and supply in Korea, Ministry of Commerce. Industry and Energy and Korean
Energy Management Corporation; 2000.
[3] Kong C, Bang J, Kim H. A study on aerodynamic analysis and starting simulation for horizontal axis wind turbine blade.
J KSPE 1999;3(3):406.
[4] Kong C, Jeong J. Improvement of design by structural test for 750 kW HAWT composite blade. J KSPE 2000;4(3):229.
[5] Kong C, Bang J, Jeong S, Ryu J, Kim Y. Structural design of medium scale composite wind turbine blade. Proceedings of
the third Asian-Pacific conference on aerospace technology and science (APCATS2000); 2000, pp. 37684.
[6] Kong C, Jeong S, Jang B, Bang J. Design improvement on wind turbine blade of medium scale HAWT by considering
fatigue life. J KSPE 2000;4(3):2937.
[7] Kong C, Kim J. Structural design of medium scale composite wind turbine blade. KSAS Int J 2000;1(1):92102.
[8] Kong C, Bang J, Kang M, Jeong S, Yoo J. Structural design of medium scale composite wind turbine blade. Thirteen
international conference on composite materials (ICCM-13); 2001.
[9] Mayer RM. Design of composite structures against fatigue applications to wind turbine blades. Chippenham: Antony
Rowe Ltd; 1996 pp. 195208.
[10] Gourieres DLE. Wind power plants theory and design. New York: Pergamon Press; 1982.
[11] Kong C, Kim H, Kim J. A study on structural and aerodynamic design of composite blade for large scale HAWT system.
Final report, Hankuk Fiber Ltd; 2000.
[12] IEC International Standard. Wind turbine generator systemPart I: safety requirements; 1994.
[13] Technical Note: IEC 1400-1 GL Test Regulation, 2000.
[14] Germanischer Lloyd. Regulations for the certification of wind energy conversion system. Germany: Germanischer Lloyd;
1999.
[15] Palmgren A. Die Lebendauer von Kugellagern. Zeitschrift von Deutche Ingenieurring 1924;68:33941.
[16] Miner MA. Cumulative damage in fatigue. J Appl Mech 1945;12A:15964.
[17] Mandell JF, Reed RM, Samborsky DD. Fatigue of fiberglass wind turbine blade materials. Albuquerque, New Mexico,
SAND92-S7005: Sandia National Laboratories; 1992.
[18] Veers PS, Lange CH, Winterstein SR. Farow: a tool for fatigue and reliability of wind turbines. Windpower93
1993;3429.
2114 C. Kong et al. / Energy 30 (2005) 21012114

[19] Bishop NWM, Zhihua H. The fatigue analysis of wind turbine blades using frequency domain techniques. Amsterdam
EWEC 91 1991.
[20] Broek D. Elementary engineering fracture mechanics. Dordrecht (Boston): Martinus Nijhoff; 1982.
[21] Delft DRV, Corbet DC, van Leeuwe JL. Full scale fatigue tests of wood-epoxy blades. Amsterdam EWEC91 1991.
[22] Inomata N, Tsuchiya K, Yamada S. Measurement of stress on blade of NEDOs 500 kW prototype wind turbine. Renew
Energy 1999;16:9125.
[23] Bechly ME, Clausent PD. Technical note: structural design of a composite wind turbine blade using finite element
analysis. Comput Struct 1997;63(3):63946.
[24] EMRC, NISAII-Users Manual, version 5.2; 1992.
[25] Spera DA. Dynamic loads in horizontal-axis wind turbines Part II: empirical equations. Windpower93 1993;2829.
[26] Zweben C, Thomas HH, Chou TW. Mechanical behavior and properties of composite materials, vol. 1.: Technomic
Publishing Co., Inc.; 1989.