Research Article

Advances in Mechanical Engineering

2018, Vol. 10(4) 1–9
Ó The Author(s) 2018
Fracture analysis and improvement of DOI: 10.1177/1687814018769003
journals.sagepub.com/home/ade
the main shaft of wind turbine based
on finite element method

Ruiming Wang1, Tian Han2, Wenrui Wang2, Yang Xue1 and Deyi Fu1

Abstract
For the main shaft of wind turbine of certain type, shaft fracture occurs at the variable section of the shaft during early
stage of operation. In order to validate the failure analysis, finite element analysis of the main shaft was performed. The
analysis results demonstrate that there is a severe stress concentration that leads to the formation of initial cracks at
variable inner diameters of the main shaft. Also, the stress in the variable part mainly resulted in the pressure of the bul-
ging joining sleeve applied on the main shaft via the gearbox planet carrier shaft. According to the above analysis, local
structure improvements were carried out through increasing the action area between the bulging joining sleeve and the
main shaft. The finite element simulation results show that the stress concentration in the variable section of the
improved shaft decreases significantly. The improvement of shaft strength decreases the possibility of crack formation
and its growth, thus enhancing the reliability of the main shaft. This analysis process and the results of this study can pro-
vide a reference in shaft fracture analysis and also technical support for improvement in the design of wind turbine main
shafts.

Keywords
Fracture analysis, wind turbine, shaft design, structure improvements

Date received: 19 December 2017; accepted: 12 March 2018

Handling Editor: Shun-Peng Zhu

Introduction complexity of the systems and the components of a

wind turbine, it can be challenging to identify the core
Wind is a clean and reproducible energy, which is the causes of a failure.
fastest growing energy source. Furthermore, wind According to the guidelines and standards for wind
energy market has achieved substantial growth over turbine generating system (WTGS), generally wind tur-
years, especially in the United States and China. In bines are designed with a working life of at least 20
most situations, wind turbines are not easily accessed
due to the remote locations of mountain areas or off-
1
shore. Once certain components of the unit fail, the loss State Key Laboratory of Operation and Control of Renewable Energy
& Storage Systems, China Electric Power Research Institute, Beijing,
of power-generating capacity is inevitable. In the mean-
China
time, a mass of manpower and material resources are 2
School of Mechanical Engineering, University of Science & Technology
required to replace the components. Although the Beijing, Beijing, China
design working life of wind turbine is over 20 years, the
operation and maintenance costs of the units may Corresponding author:
Tian Han, School of Mechanical Engineering, University of Science &
account for 10%–15% of the revenue of the wind Technology Beijing, 30 Xueyuan Road, Haidian District, Beijing 100083,
power stations.1 Especially, unplanned failures of the China.
key components can be very expensive. Because of the Email: [email protected]

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering

Figure 2. Schematic diagram of the wind turbine.

Figure 1. Outages and downtime weights of wind turbine
faults (Guohua Energy Investment Co., Ltd.).

years.2,3 As shown in Figure 1, the fault statistics of the

wind turbine shows that the outages caused by the main
shaft failure account for less than 0.1% (marked in
blue), but its downtime accounts for a very large share,
about 24.8% (marked in red).4
Although many scholars have been engaged in fail-
ure detection and fault diagnosis of wind turbine,5–8 Figure 3. The supporting arrangement of the main shaft and
there are still many issues that are unresolved such as fracture position.
life prediction of key components and efficient prog-
nosis health management. Many papers focusing on Based on the observation and actual measurement,
fault diagnosis of various key components of turbine the fracture position located in the variable diameter
blades, support bearings, and transmission gearboxes section of the main shaft, which is shown in Figure 4.
have been published.9–12 The main shaft is a critical To analyze the root cause of shaft failure, the
component in the wind power station, which carries mechanical tests of the material of the shaft were first
variable forces transmitted by the system. It has been carried out to verify the quality of the material. Second,
reported that certain type of main shafts of wind tur- the static stress analysis and strength calibration based
bine units have fractured prematurely after 4 years of on finite element method (FEM) were performed
performance. The failure leads to significant economic according to the Fatigue, Aerodynamics, Structures,
losses to the wind farm. and Turbulence (FAST) procedure. Finally, in order to
A model of WTGS is the research object that solve this problem, local structure improvement of the
belongs to the double-fed wind turbine shown in main shaft is carried out.
Figure 2. The parameters of the power system are 850
kW rated output power and 1500 rev/min rated rota- Fracture analysis of the main shaft
tion speed. The weight of the main shaft is 1260 kg and
fracture
its highest speed is 24.3 rev/min.
The WTGS that produced the main shaft fracture in For the purpose of thoroughly analyzing the dominant
this study was designed for European environments cause of main shaft fracture, this study consists of two
with operation temperatures ranging from –20°C to parts: (1) material verification of the main shaft and (2)
+ 40°C. In 2007, this batch of WTGS was installed in static stress analysis and strength calibration of the
Northern China where the temperature can be as low main shaft based on FEM.
as –40°C in cold winter and finally crashed between
October 2011 and February 2012, with a working life Experimental analysis
of just 4 years.
To verify the quality of the main shaft material and
The studied main shaft is made of 34CrNiMo6 steel
determine whether or not the material defect caused its
with heat treatment.13 The material properties of the
fracture, in this study the mechanical property tests of
main shaft steel can be referred from Siemia˛tkowski
the material of the main shaft were carried out.
et al.14 The shaft is supported by two-point self-align-
ing roller bearings, which endure transmission torsion,
transverse moment, and axial force from the hub and Mechanical property test
gearbox. The fracture position and the connection of Tensile strength and impact strength tests were per-
the shaft output are illustrated in Figure 3. formed to verify the quality of the main shaft material.
Wang et al. 3

Figure 4. Image of fracture surface. Figure 6. The stress–strain curve of one tested specimen.

Figure 7. Specimens after mechanical test.

Figure 5. Dimensions of the specimen for tensile test.

Based on the ISO Standards, the specimens were pro-

cessed15–17 to test the properties of the shaft material
such as tensile strength and impact strength (see
Figure 5).
Two groups of six test specimens were used in the
tensile and impact tests by the universal testing Figure 8. Sampling points for hardness test.
machine, respectively. The stress–strain curve and the
specimens after the mechanical test are shown in
Figures 6 and 7.
From the test results, the mechanical properties are
listed in Table 1.
The tensile strength and impact properties of the
main shaft at low temperature are insufficient com-
pared with the EN10083-3:2006 Standard. As men-
tioned previously, the WTGS crashed in winter, so it
can be considered that the insufficient mechanical prop-
erties of the shaft material have some influence on the
fracture of the main shaft.
Rockwell Hardness machine was used to test the
hardness of the specimens with the sampling points
taken along the longitudinal axis of the shaft. The sam-
pling points for hardness test are shown in Figure 8.
The hardness values ranged from 26.7 to 29.9 HRC Figure 9. Data of hardness test.
with a mean value of 28.1 HRC as shown in Figure 9.
The HRC value of the quenched and tempered (QT)
34CrNiMo6 steel has a typical hardness value of 271– the main shaft from the inner diameter to the mean dia-
331 HB (27.1–35.3 HRC).18 The results of the hardness meter are below the standard value (see Figure 9). As
test show that the measured values at certain points of the hardness can reflect the strength of the materials to
4 Advances in Mechanical Engineering

Table 1. Mechanical properties (using mean value).

Test items Rp0.2 (N/mm2) Rm (N/mm2) Z (%) A (%) K (KV2/J)

Room temperature –30°C

Measured value 725 883 66 19 58 31

Standard value 700 900–1100 55 12 45 32

directions and the rotational degrees of

freedom about the X axial direction were
constrained to express the aligning perfor-
mance of the bearings;
(b) The interface between the planet carrier
and the bulging joining sleeve: setting the
interface as the contact;
(c) The outer surface of the planet carrier:
The displacement degrees of freedom of
the planet carrier center in the 3D direc-
Figure 10. The model of the wind turbine main shaft.
tions and the rotational degrees of free-
dom about the X axial direction were
constrained.
a great extent, the hardness result indicates that the
material strength of the main shaft is insufficient. 2. Imposing loads

(a) Hub center: A reference point was estab-

Finite element analysis lished in the hub center and coupled with
For a more detailed understanding of the stress distri- the flange surface. The wind loads
bution of the shaft, the analysis of the stress state of the (Fx , Fy , Fz , Mx , My , Mz ) imposed on the
wind turbine main shaft was carried out by FEM. In shaft and the wind wheel gravity were
this article, the stress values were analyzed under the applied on the reference point;
rated load condition and the impact condition, respec- (b) The interface between the planet carrier
tively. For the convenience of analysis, the screw holes and the bulging joining sleeve: Radial pres-
and pin holes on the flange were ignored as they have sure was applied on the outside surface of
very little effect on the stress state of the main shaft. the planet carrier to simulate the force that
As shown in Figure 10, the discrete model of the the bulging joining sleeve imposed on the
main shaft was presented by 137,675 elements with 8- planet carrier.
node hexahedron reduced integration elements (C3D8R The FEM analysis of the main shaft was conducted
unit). To obtain a more accurate description of the with the applied loads on the main shaft under the rated
stress state in the key structural positions of the shaft load condition and impact condition, respectively.
that may result in large stress, including the transition
fillet, shaft shoulder, and unloading slot, the shaft grid Load acquisition. In order to obtain a reasonable stress
was moderately refined. To avoid local stress distortion distribution of the main shaft, the key point is to obtain
caused by direct application of the boundary conditions the realistic loads of the main shaft. The wind loads are
on the main shaft,19 a part of the planet carrier of the obtained by aerodynamics analysis using the FAST
gearbox that is in direct contact with the main shaft, on procedure in this article.20 The FAST code can be used
which the radial pressure was applied to simulate the to aeroelastically simulate the extreme and fatigue
effect of the bulging joining sleeve, was added in the loads. FAST with AeroDyn21 was validated by
model: Germanischer Lloyd WindEnergie.22 So it is of practi-
cal significance to utilize the FAST procedure to con-
1. Applying constraints duct the dynamic simulation analysis in this article.
The analysis progress is shown in Figure 11. The
(a) The front bearing surface: The displace- nonlinear equations of motion are simulated with time
ment degrees of freedom of the rigid sur- marching. While the simulation is ongoing, the aerody-
faces in the three-dimensional (3D) namic and structural response of the wind turbine to
Wang et al. 5

Figure 11. Analysis flow chart of wind loads.

Table 2. Parameter list.

Parameter Value Nomenclature/definition

d Fit normal diameter Fit normal diameter

W 850 kW Rated power
l 180 mm Fit length
f 0.2 Friction coefficient
n 24.3 rev/min Rated rotation speed

wind inflow conditions is timely decided. In the process

of simulation analysis, many aspects of turbine opera-
tion can be controlled actively. FAST calls the full-field
background wind files generated by the TurbSim23 sub-
Figure 12. The stress distribution of the main shaft under
routine and uses the AeroDyn subroutine package to
rated condition (MPa).
generate aerodynamic forces along the blade. The simu-
lation results consist of the time series data of aerody-
namic loads and the load and deformation of the wind
turbine structural members. the definition of the linear elastic material constitutive
In this article, the wind models, namely, Normal model is adopted, mainly considering the elastic modu-
Turbulence Model (NTM) and 3EWM50 (Class III tur- lus, Poisson’s ratio, density of the input, without con-
bulent Extreme Wind Speed Model, 50-year recur- sidering the plastic part of the material properties, and
rence), were used to simulate the rated wind loads and the material of the stress–strain curve after yield input.
the extreme loads of the wind turbine. In addition, the
radial pressure of the shaft tail is calculated according
FEM analysis under rated load condition. Applying loads
to the formulae in Section 2.6.2 of Zhang et al.24 The
and constraints, the FEM analysis of the main shaft
wind wheel gravity is calculated according to its mass
under the rated load is conducted, the results of which
(11.5 t).
are shown in Figures 12–14.
4T 4 9549W As shown in Figures 13 and 14, there is serious stress
p =  = 150:7 MPa ð1Þ concentration in the variable radius section of the shaft
pd 2 lf pd 2 lf n
tail and the front edge of the front main bearing sur-
where p is the radial pressure, T is the rated torque, d is face. The maximum equivalent stress of the main shaft,
the fit nominal diameter, f is the friction coefficient, about 441.3 MPa, appears in the front edge of the mini-
and l is the length. And the values of these parameters mum inner diameter at the end of the shaft. The equiv-
are listed in Table 2. alent stress in the front edge of the front main bearing
Based on the previous research and simulation surface roughly equals 304 MPa. The equivalent stress
results, it is found that the structure of the load under in the front edge of the front main bearing surface
the current setting has not reached the yield. Therefore, equals 247.6 MPa.
6 Advances in Mechanical Engineering

Figure 13. The internal stress distribution of the main shaft under rated condition (MPa).

Figure 14. Magnified view of stress distribution of the shaft for

rated condition (MPa).

The results of the FEM analysis are consistent with Figure 15. The image of the fracture surface.
the characteristics of the variable radius section of the
fracture main shaft. The marked area in Figure 15 is
one of the fatigue sources, which was caused by stress where the parameter S represents the minimum safety
concentration of the transition fillet in the variable dia- margin of the shaft material.
meter section of the main shaft inner diameter and is The safety margin of the main shaft is above zero,
characterized by the cowrie pattern lines, with the which indicates that the strength of the main shaft can
direction of extension from the internal to the external transmit the normal working load.
surface.
Based on the guideline for the certification of wind
turbine,2 the yield strength, ss , of the shaft material FEM analysis under impact load condition. The main shaft
(34CrNiMo6 steel) is equal to 700 MPa, the material endures huge impact load in case of variable wind loads
partial safety factor g m = 1:1, and the allowable stress, and WTGS emergency start or stop. The effect of the
½s, can be obtained as follows impact load is mainly reflected in the twisting action on
the main shaft. As reported in Xie et al.,25 during emer-
ss 700 gency braking, the maximum torque is capable of
½s = = = 663:5 MPa ð2Þ
gm 1:1 reaching 2–3 times the rated torque. In this article, dur-
ing the process of FEM analysis of the main shaft under
In this case the impact load condition, the torque (Mx ) trebled the
rated torque T, with other invariable loads (Fx , Fy , Fz ,
½s 663:5
S=  1=  1 = 0:504.0 ð3Þ My , Mz , G_F, pressure). The simulation results are
s 441:3 shown in Figures 16–18.
Wang et al. 7

shown in Figure 16, the variable radius section and the

interaction zone between the elliptical arc transition
surface and the front bearing surface experienced severe
stress concentrations.
In contrast, the stress distribution of the main shaft
under the rated load conditions and impact conditions,
the stress in the elliptical arc transition surface increased
significantly, especially in the interaction zone between
the elliptical arc transition surface and the front bearing
surface, from about 247 to 573.5 MPa. While the stress
in the shaft tail which is held by the bulging joining
sleeve experienced little changes, it increased from
about 304 to 305.5 MPa, which draws a conclusion that
Figure 16. The stress distribution of the main shaft under stress in this position is mainly affected by the radial
impact condition (MPa). pressure that the bulging joining sleeve imposed to the
shaft via the planet carrier of the gearbox.

The maximum equivalent stress of the main shaft,

about 573.5 MPa, appears in the front edge of the front Structure improvement
main bearing surface. The equivalent stress of the tran- As reported in the previous analysis, there is severe
sition fillet in the variable radius section in the main stress concentration in the interaction zone between the
shaft tail equals 305.5 MPa. As shown in Figures 13 elliptical arc transition surface and the front bearing
and 14, there is serious stress concentration in those surface and the variable radius section at the end of the
two places. main shaft. In order to resolve the stress concentration
Under the impact load condition, we have of the main shaft and finally improve the reliability of
wind turbines, the methods of surface treatment, mov-
½s 663:5 ing forward the inner diameter and others will be con-
S=  1=  1 = 0:157.0 ð4Þ
s 573:5 sidered in this article.
The safety margin of the main shaft is also above
zero.
The main shaft material contains defects, which
Surface treatment and heat treatment
makes it easy to produce initial cracks. The initial In general, the improvement of the surface condition
cracks extend continuously under large alternating can be used to improve the fatigue strength of the main
loads. When the actual bearing area cannot sustain the shaft. Cold process techniques, such as shot-peening
limit impact load, the main shaft fractured. and rolling, can be used to improve the surface strength
From the above results, it can be seen that the major- and residual compressive stress, which are effective to
ity of the stresses are at the elliptical arc transition sur- decrease the average tensile stress and reduce the initia-
face and the tail which is held by the bulging joining tion and extension of cracks. In addition, the surface of
sleeve is much larger than the other positions of the the main shaft can be spray coated to improve the fati-
main shaft (see Figure 16). From the detailed view gue strength.

Figure 17. The internal stress distribution of the main shaft under impact condition (MPa).
8 Advances in Mechanical Engineering

Figure 18. Magnified view of the internal stress distribution

under impact condition (MPa).
Figure 19. The stress distributions of the main shaft (MPa).
From the view of engineering knowledge related to
the cyclic behavior of fatigue cracks, the influence of
heat treatment is necessary to be considered. Different
heat treatments can result in different fatigue crack
growth rates due to different hardness levels as well as
different ductility.26

Local structure improvement

To enhance the reliability of the main shaft, many
structural modifications have been attempted based
on the FEM, such as adding relief grooves, increasing
the action area between the bulging joining sleeve and
the main shaft, moving the variable radius section. As Figure 20. The internal stress distributions of the main shaft
severe stress concentrations locate within the area of (MPa).
radial pressure imposed by the bulging joining sleeve
with variable inner diameters, the method of moving
the variable radius section is the most effective
method. The basic method is to decrease the radial
pressure from the bulging joining sleeve. The trans-
mitted radial pressure, p, should not make circumfer-
ential slip under the impact torque Mf (usually 2–3
times the rated torque T). In other words, the friction
resistance moment should be greater than or equal to
the impact torque

d
Mf = p  d  l  p  f  ð5Þ
2
where d is the fit nominal diameter and f and l are the
friction coefficient and fit length, respectively. Figure 21. Magnified view of the internal stress distributions of
According to the above formula, it is shown that the main shaft tail (MPa).
decreasing the pressure p and increasing the action area
can both transmit the load and improve the stress con-
centration at the main shaft tail. The modified finite decreased significantly from about 441.3 to 302.4 MPa;
element model of the main shaft loaded with the rated especially, the stress at the variable radius section at the
loads and newly calculated p was analyzed, and the end of the main shaft decreased from 304 to 94.7 MPa.
simulation results are shown in Figures 19–21. By means of structure improvement, the stress con-
In contrast to the analyzed results in section 2.3.2, centrations in the main shaft have been improved,
after redesigning the main shaft structure, the stress at which can decrease crack initiation and extension
the region affected by the bulging joining sleeve effectively.
Wang et al. 9

Conclusion 8. Liua WY, Zhang WH, Han JG, et al. A new wind tur-
bine fault diagnosis method based on the local mean
Based on the mechanical property test, the impact decomposition. Renew Energ 2012; 48: 411–415.
property at low temperature and the hardness of the 9. Marı́n JC, Barroso A, Parı́s F, et al. Study of fatigue
main shaft are insufficient, which indicate the potential damage in wind turbine blades. Eng Fail Anal 2009; 16:
influence from the fracture mechanics point of view. 656–668.
Based on the FEM analysis, the stress concentrations 10. Chou J-S, Chiu C-K, Huang I-K, et al. Failure analysis
of the main shaft result in the generation of fatigue of wind turbine blade under critical wind loads. Eng Fail
sources, which is in good agreement with the analysis Anal 2013; 27: 99–118.
results of Aleksandra et al.27 Based on the results of dis- 11. Ma H, Yu Y, Jiang T, et al. Failure analysis of gear in a
crete models of the main shaft for strength analysis, wind power generator. J Mater Eng 2009; 11: 88–91.
12. Evans M-H, Richardson D, Wang L, et al. Serial section-
structure improvement of the main shaft was carried
ing investigation of butterfly and white etching crack
out. The simulation result shows the improvement of (WEC) formation in wind turbine gearbox bearings.
moving the variable radius section is effective. This arti- Wear 2013; 302: 1573–1582.
cle provides a reference for shaft fracture analysis and 13. DIN 17201/E DIN EN10083, 1-3:2006-10. Vergütungs-
these results provide technical support for improvement stähle—Teil 3: Technische Lieferbedingüngen für legierte
in the design of wind turbine main shafts. It seems justi- stähle; Deutsche norm.
fied to perform further works and analyses in order to 14. Siemia˛tkowski Z, Gzik-Szumiata M, Szumiata T, et al.
evaluate the potential lifetime extension of the main Metallurgical quality evaluation of the wind turbine main
shaft. shaft 42CrMo4 steel: microscopic and Mössbauer studies.
Nukleonika 2017; 62: 171–176.
15. ISO 377:2013. Steel and steel products—location and pre-
Declaration of conflicting interests paration of samples and test pieces for mechanical testing.
The author(s) declared no potential conflicts of interest with 16. ISO 6892-1:2009. Metallic materials—tensile testing—
respect to the research, authorship, and/or publication of this part 1: method of test at room temperature.
article. 17. ISO 148-1:2006. Metallic materials—Charpy pendulum
impact test—part 1: test method.
18. DIN EN ISO 18265:2004. Metallic materials—conversion
Funding of hardness values.
The author(s) received no financial support for the research, 19. Soo KH, Mall S, Ghoshal A. Two-Dimensional and
authorship, and/or publication of this article. Three-Dimensional Finite Element Analysis of Finite
Contact Width on Fretting Fatigue[J]. Materials Transac-
tions, 2011, 52(2): 147–154.
References 20. Jonkman J. FAST (NWTC computer-aided engineering
1. Lu B, Li Y, Li X, et al. A review of recent advances in tools), http://wind.nrel.gov/designcodes/simulators/fast/
wind turbine condition monitoring and fault diagnosis. 21. Laino DJ. AeroDyn (NWTC computer-aided engineering
In: Proceedings of the IEEE power electronics and tools), http://wind.nrel.gov/designcodes/simulators/aerodyn/
machines in wind applications, Lincoln, NE, 24–26 June 22. Manjock A. Evaluation report: design codes FAST and
2009, pp.109–115. New York: IEEE. ADAMSÒ for load calculations of onshore wind turbines.
2. Lloyd G. Guideline for the certification of wind turbine. Report no. 72042, 26 May 2005. Humburg: Germa-
Berlin: Key Wind Energy GmbH, 2010. nischer Lloyd WindEnergie GmbH.
3. IEC 61400-1:2005. International standard. 23. Kelley N and Jonkman B. TurbSim (NWTC computer-
4. Ren Yu-ting. Application research of vibration supervi- aided engineering tools). http://wind.nrel.gov/design-
sion technology on wind power generating units. Inn codes/preprocessors/turbsim/
Mong Elect Power 2010; 28: 8–12. 24. Zhang Z, Yin Z, Han T, et al. Fracture analysis of wind
5. Liu W, Tang B and Jiang Y. Status and problems of turbine main shaft. Eng Fail Anal 2013; 34: 129–139.
wind turbine structural health monitoring techniques in 25. Xie D, Yang Y, Yang G, et al. Analysis on impact prop-
China. Renew Energ 2010; 35: 1414–1418. erties of cast wind turbine main shaft by finite element.
6. Amirat Y, Benbouzid MEH, Al-Ahmar E, et al. A brief Hot Work Technol 2011; 40: 49–51.
status on condition monitoring and fault diagnosis in 26. Lesiuk G, Duda M, José AFO, et al. Fatigue crack
wind energy conversion systems. Renew Sust Energ Rev growth of 42CrMo4 and 41Cr4 steels under different heat
2009; 13: 2629–2636. treatment conditions. Int J Struct Int, in press.
7. Tang B, Liu W and Tao S. Wind turbine fault diagnosis 27. Aleksandra K, Grzegorz L and Miko1aj K. Root causes
based on Morlet wavelet transformation and Wigner- analysis of differential pinion shaft assembly failure in
Ville distribution. Renew Energ 2010; 35: 2862–2866. WRX class car. Int J Struct Int 2017; 8: 694–706.