Beruflich Dokumente
Kultur Dokumente
DOI: 10.1111/ffe.12654
1
Faculty of Mechanical Engineering,
Abstract
University of Maribor, Maribor, Slovenia
2
Končar‐Generators and Motors Inc.,
The axle pin is one of the most critical components of a wind turbine. Because 2
Zagreb, Croatia bearings are mounted on the axle pin, it is not subjected to torque and is designed
3
Faculty of Mechanical Engineering and only to carry bending stresses. The axle is usually made by a casting process, which
Naval Architecture, University of Zagreb,
can induce defects. After casting of the axle pin, radial surface cracks were detected
Zagreb, Croatia
outside by ultrasonic and magnetic particle inspection. The first goal of this paper is
Correspondence to establish defect tolerance in the design of the axle pin, ie, at what critical crack
N Gubeljak, Faculty of Mechanical
Engineering, University of Maribor, 2000
size can we expect failure of the axle pin? The second aim is assessing the remaining
Maribor, Slovenia. lifetime of the defected axle pin used in operation of a wind turbine. Both issues are
Email: nenad.gubeljak@uni‐mb.si important to prevent unexpected failure and establish a system for on‐line structural
health monitoring. This paper presents a procedure for the structural integrity
assessment of axle pins by considering their material properties and fatigue crack
growth parameters and comparing them with their fatigue design curve.
K EY WO R D S
axle pin in wind turbine, crack's defects, fatigue design, structure integrity assessment
1284 © 2017 Wiley Publishing Ltd. wileyonlinelibrary.com/journal/ffe Fatigue Fract Eng Mater Struct. 2017;40:1284–1294.
GUBELJAK ET AL. 1285
for flaws and finding the lifetime of the axle pin from the
detected crack size until the critical crack depth. Per the
FITNET2 structural integrity procedure,5,6 the flaws are
treated as cracks. The structural integrity assessment
procedures are tools for estimating the effect of the crack size
for given loading amplitude. In the past 20 years, different
structural integrity procedures have been developed.2,7,8
The structural integrity assessment of the axle pin should
provide answers to the following issues:
Critical crack size is important for the determination of metallographic investigation shows presence of MnS inclu-
the fatigue lifetime from an initial detected crack to the sions over 20 μm in size and a cluster of inclusions 100 μm
critical crack size. Each loading amplitude has its own large, as shown in Figures 4A and 4B. Energy‐dispersive X‐
maximum load and therefore its own critical crack length. ray analysis positions the presence of MnS in nonmetallic
If the difference between the crack size in the first and the inclusions. Changes in wall thicknesses and the reduction of
second cases is significant, then the period of fatigue crack diameters can cause defects in materials, such as pores, flaws,
propagation can be treated as the service lifetime. If the ser- or any remaining slag. We used sections of the cast steel slab
vice lifetime is shorter than the required 20 year lifetime of for determination of the mechanical properties. Material prop-
the wind turbine, then the axle pin containing the detected erties are obtained by tensile testing prescribed by the ASTM
flaws is not suitable. The main goal of this paper is to E8M‐2015 standard.10 The obtained tensile material proper-
demonstrate a structural integrity approach containing the ties of the cast steel are given in Table 2. The fracture tough-
determination of the maximum sustained loading capacity, ness values were obtained by crack tip opening displacement
the determination of the critical crack length or depth, and testing per the ASTM E‐1820 standard11 and, with conserva-
finally, the determination of the remaining lifetime of the tive values for the 3 specimens, are listed in Table 2. The
component by using the fatigue design loading curve. obtained fracture toughness value KIQ = 193 MPa m½ was
obtained as the minimum of that of all 3 specimens,
conforming to the ASTM E‐1820 standard testing.
2 | MATERI AL PRO PERT IES OF
The design procedure for the wind turbine components,
CASTING STEEL
such as the axle pin, should be performed per the
Germanischer Lloyd Rules and Guidelines12 and the
2.1 | Tensile properties, chemical composition,
International standard IEC 61400‐1.13 The design procedures
and microstructure
prescribe design loading conditions of components: the so‐
The casting steel G24CrMn6 is a low‐carbon Mn alloyed steel called fatigue design load with a cumulative number of cycles
with the chemical composition shown in Table 1. The grain that the axle pin should survive in the 20 years of a safe
size was measured by using the Jaffries Planimetric Method service lifetime. The loading conditions are prescribed by
according to ASTM E‐112.9 The grain size value was the standardized loading cases regarding the wind region
0.075 ± 0.022 mm (4.6 ASTM) with mainly a ferritic and and conditions.12 The axle pin is often equipped by strain
bainitic microstructure, as shown in Figure 3. During the gauges for deformation measurement and piezo sensors for
solidification process, the nonmetallic inclusions can remain acceleration measurement. For confirming the local most
in the area where the geometry and thickness change. The deformed area, different approaches can be applied.14-16
TABLE 1 Chemical composition of steel G42CrMn6 (wt%) 2.2 | Fatigue crack growth rate testing
C Si (max) Mn S (max) Cr P (max)
The fatigue limit is designed with respect to the Whöler (S‐N)
0.22 0.6 1.6 0.015 1.5 0.020
curve without a crack, with a safety factor (SF) of 2.
GUBELJAK ET AL. 1287
da
¼ C ðΔK−ΔK th Þm ; (1)
dN
critical location. The most critical crack opening‐bending the second set consists of the loading distribution, and the
stress σ22 appears on the upper surface of the axle pin. The third set is the position and size of the flaws.
crack opening stress is 135 MPa on the surface and on the During inspection, the defects are detected within the
inner surface −10 MPa in compression; therefore, the total region of the geometrical change of the axle pin. The position
stress amplitude is 145 MPa. The distribution of the σ22 stress and size of the detected crack in the critical section of the axle
throughout the thickness represents the crack opening stress pin are schematically shown in Figure 9A. The idealized
and is relevant for the structural integrity assessment using unique surface crack in the most critical region, due to con-
the mentioned procedures, such as FITNET.2 servatism, can be considered by the FITNET2 procedure, as
shown in Figure 9B. Per the FITNET2 procedure, it is allow-
able to assume and perform the assessment for the worst‐case
crack geometry. This conservative approach leads to conser-
3.1 | Description of the applied assessment
vative results. An idealization of the crack geometry is shown
procedure
in Figure 9B, where a = 20 mm is the crack depth and 150° is
The structural integrity assessment procedure is based on 3 the radial surface crack length. The wind‐power turbine's axle
sets of input data. The first set represents the mechanical pin is loaded by bending, as 2 bearings are mounted onto the
properties and the fracture toughness of the cast material, pin, shown schematically in Figure 1. Therefore, the bearings
FIGURE 9 A, The critical section of the axle pin with the position and size of the crack. B, The idealized unique surface crack in the most critical
region
1290 GUBELJAK ET AL.
on the pin ensure rotation of the hub with a direct drive to the stress, considering the crack size and loading manner
direct current generator, while on the axle pin, the stator FITNET.2
generator was fixed, as shown in Figure 1. Reference stress takes into account the bending moment
The stress intensity factor is calculated via Equation 2 at yielding stress defined as
below:
4 ⋅M Y
pffiffiffiffiffiffiffi a σ ref ¼ (7)
K I ¼ σ⋅ π⋅a⋅Y ; β ; (2) T ⋅ π ⋅ ðR o þ R i Þ2
T
where T is the thickness of the axle pin, Y(a/T,β) is the stress
intensity function, a/T is the crack depth ratio, β is the surface
2 βa a
crack length in radians, and σ is the principal opening load, MY ¼ TðRo −Ri Þ ⋅ cos − sin β ⋅σ Y (8)
2T 2T
obtained by FE modelling. The stress intensity function for
a hollow thick tube is available in the compendia of the stress where T = 69.8 mm is the thickness, Ro = 355.05 mm is the
intensity factors for FITNET7 procedure. outside diameter, and Ri = 285.25 mm is the inside diameter
" of the axle pin. Plastic collapse at the base of the FITNET7
a
a a β 0:565 procedure is obtained by
Y ; β ¼ 1:1 þ −0:09967 þ 5:0057⋅ ⋅
T T T π
# 1 R p0:2 þ Rm
Lmax ¼ ⋅ (9)
a β r
2 Rp0:2
−2:8329 ⋅ ⋅ (3)
T π
where Rp0.2 and Rm are the yield stress and the ultimate ten-
The distribution of the crack opening stress along the axle sile stress, respectively (see Table 1). In the FITNET proce-
is shown in Figure 8. This loading corresponds to the highest dure, the failure assessment curve‐FAC is defined for
permitted loading in the design of the wind‐power turbine. materials with continuous hardening by
Figure 8 shows that the highest stresses appear within the −12 h
1 i
region of detected defects. In structural integrity analysis, fðLr Þ ¼ 1 þ L2r ⋅ 0:3 þ 0:7⋅e−μ⋅Lr for 0 ≤ Lr ≤ 1
6
The fracture toughness of the material f ðLr Þ ¼ f ðLr ¼ 1Þ⋅Lr 2⋅N for 1 ≤ Lr ≤ Lmax
r : (12)
Kmat = KIQ = 193 MPa m½ was obtained as the minimum
The strain hardening exponent N is given by the empirical
of all 3 specimens as prescribed by ASTM E‐1820 standard
term:
testing. In the case of ductile fracture, the fracture toughness
of the material is usually expressed in the J integral or the Rp0:2
crack tip opening value crack tip opening displacement = δ: N ¼ 0:3⋅ 1 þ (13)
Rm
rffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
J mat ⋅E Rp 0:2 ⋅ δmat ⋅E The axle pin is the most critical component within a wind
K mat ¼ ¼ : (5)
1− ν2 1− ν2 turbine regarding safety because a fail‐safe design is not
available. The failure assessment diagram has been plotted
The normalized load Lr for the FITNET procedure is by applying Equation 10 for the failure assessment curve ver-
defined as sus the normalized load Lr (Equation 6). The normalized
M b σ ref crack driving force Kr (Equation 4) increases with applied
Lr ¼ ¼ ; (6) load and crack length by using Equation 2.
MY σY
The critical loading capacity of the idealized crack
where My is the limit of the bending moment when yielding (a = 20, β = 150°) is obtained as the intersection point
the rest of the noncracked ligament and σref is the referenced between the normalized crack driving force Kr and the failure
GUBELJAK ET AL. 1291
assessment curve f(Lr), as shown in Figure 10. The crack TABLE 3 Critical crack surface length regarding to admissable
angle β = 150° is kept constant because the axle pin is bending stress at constant crack depth a ¼ 20 mm
subjected to bending (without rotation); therefore, the crack Surface Surface Crack Maximum Admissable
is going to propagate preferentially in depth, rather than Length Angle Bending Stress
longitudinally. The normalized load contains 1 value at each l (mm) β (°) σx, N/mm2
intersection point. Therefore, the corresponding value of the 5 0.8° 583
applied bending moment Mb is known if the limit bending 40 6.45° 557
moment MY is calculated. 120 19.36° 530
To estimate the remaining load bearing capacity, it is 200 32.27° 507
possible to repeat the same procedure for different crack
400 64.54° 470
sizes. The limit of the bending moment is determined for
600 96° 450
different surface crack lengths, ie, the angle β. Table 3
935 150° 440
shows the values for the maximum carrying capacities
regarding different surface crack lengths for a constant crack
depth of a = 20 mm. crack length is possible to estimate by using the FITNET2
Usually the SF for a wind turbine is 2.10 Per the FITNET procedure by simulation of an increasing crack depth (under
structural integrity procedure, it is possible to estimate the a constant β = 150°). It has been found that the critical crack
maximum bending stress for 2 different crack depths depth is ac = 56.6 mm (a/T = 0.81) for Δσ22 = 145 MPa, as
(a = 20 and 50 mm under a constant crack angle β = 150°). shown in Figure 12. Consequently, by decreasing the stress
Figure 2 shows that for SF = 2, the maximum allowed bend- Δσ22, the critical crack depth increases.
ing stresses are 238 and 88 MPa for crack depths of 20 and Therefore, the results show that the crack can propagate
50 mm, respectively. According to the FITNET structural from an initial size of a = 20 mm up to the critical crack
integrity procedure, we performed at structural integrity depth ac = 56.6 mm under the cyclic stress of
assessment on an axle pin with multiple cracks within a crit- Δσ22 = 145 MPa. The initial crack size of 20 mm is signifi-
ical section. In the DLC 7.1c50, the distribution of stresses cant; however, even such a crack, if it stays hidden, will not
throughout the wall thickness at the most critical section influence the projected static bearing strength of the axle
was obtained and is shown in Figure 8. The maximum bend- pin, as the bearing strength is still more than twice the
ing (crack opening) stress in the detected defect area is projected value. The potential hazard is the formation of a
145 MPa. The found flaws are not critical for the beginning fatigue crack from such a defect, which grows to critical size
of the wind turbine service because the estimated points during operation/loading cycles.
remain in the safe area reduced by the SF = 2, as is shown The design procedures prescribed the design loading con-
in Figure 11. The results also show that the final collapse of ditions of the component, the so‐called fatigue design load
the axle pin appears at a significant fatigue crack size. To fol- with a cumulative number of cycles, which the axle pin
low the SF of present defect, the on‐line monitoring system is should survive in the 20 years of a safe service lifetime.10
recommended for the critical area of an axle pin. The critical The loading conditions in 20 years are prescribed by the
loading cases regarding wind region and conditions.10 This loading amplitude (with a loading ratio of R = −1). The
curve is shown in Figure 13 by the black step line, where structural integrity damage curve was obtained by consider-
the highest number of cycles to survive corresponds to a min- ing a cumulative number of cycles to failure. Therefore, the
imum stress amplitude Δσ22 = 10 MPa for a loading ratio of structural integrity damage curve was obtained in the same
R = −1 and the maximum stress amplitude does not over- manner as the design loading curve. The design loading
come Δσ22 = 205 MPa (R = −1). Consequently, the number curve consists of a cumulative number of cycles for the cor-
of survived cycles for a shorter lifetime is proportional to the responding stress amplitude and represents the required
number of cycles, while the stress amplitude remains the number of cycles for 20 year survival. The yellow dot points
same. For an example, see the 6 year survival curve in red, in Figure 13 represent the result of the integrated cumulative
shown in Figure 13. number of cycles. The points of the maximum number of
The parameters of the fatigue crack growth rate in the cycles do not reach the design line for a 20 year lifetime.
form of the Paris‐Erdogan relationship are experimentally Because the red line crosses the points, it corresponds to a
determined, as shown in Figure 5. lifetime limit of the axle pin with the detected defects. It
To determine the number of cycles to failure for different was found that, in the region of a stress amplitude of
amplitude crack opening stresses, the integration of the σ22 = 50 to 70 MPa, the design line for 6 years of operation
Paris‐Erdogan relationship (Equation 1) was performed. time will be necessary that the crack grows from the detected
The integration is based on the individual constant stress size to the critical depth.
GUBELJAK ET AL. 1293
6. SINTAP: Structural integrity assessment procedure for European 16. Slavković R, Veg A, Dučić N, Slavković N, Baralić J, Miličevič I.
industry. Final procedure, Britte‐Euram Project No., BE95‐1426, Rigid body dynamics in optimization of the machine tool
British Steel, 1999 vibroisolation. Tech Gaz. 2015;22(1): 87–94.
7. British Standard BS 7910: ‘Guide on methods for assessing the 17. Moriarty P. Safety‐factor calibration for wind turbine extreme loads.
acceptability of flaws in metallic structures’, 2000 Wind Energy. 2008;11:601–612, Wiley Interscience (www.
8. API 579. Recommended Practice for Fitness‐For‐Service. 1st Edi- interscience.wiley.com. https://doi.org/10.1002/we.306
tion March ed. American Petroleum Institute; 2000. 18. Tadic B, Todorovic P, Novkinic B, et al. Fixture layout design
9. ASTM E‐112‐2015. Standard Test Methods for Determining Aver- based on a single‐surface clamping with local deformation. Int J
age Grain Size Using Semiautomatic and Automatic Image of Simul Model. 2015;14(3):379–391. https://doi.org/10.2507/
Analysis, Volume. 03.01 W. Conshohocken, PA: ASTM Interna- IJSIMM14(3)1.280
tional; 2015. 19. ABAQUS 6.13 Solver, Dessault Systems Samulia 2016
10. ASTM E8M‐2015. Standard Test Methods for Tension Testing of 20. GL Garrad Hassan; GH Bladed; Theory manual and user manual,
Metallic Materials, Volume. 03.01 W. Conshohocken, PA: ASTM Version 4.2; 2011
International; 2015. 21. Cvetić M, Gubeljak N. Determination of stresses in the axle pin
11. ASTM E‐1820:15. Standard Test Method or Measurement of Frac- based on the wind model in the wind turbine. Trans FAMENA.
ture Toughness, Volume. 03.01 W. Conshohocken, PA: ASTM 2017;41(1). https://doi.org/10.21278/TOF.41103
International; 2004. 22. Erdogan PP, Paris P, Erdogan F. A critical analysis of crack propa-
12. Germanischer Lloyd Rules and Guidelines; guideline for the certifi- gation Laws. J Basic Eng. 1963, pp;528–534.
cation of wind turbines, Edition 2003
13. INTERNATIONAL STANDARD; IEC 61400‐1, Wind turbines—
How to cite this article: Gubeljak N, Cvetic M,
part 1: design requirements, Third edition 2005‐08
BOŽIĆ Ž, Predan J. Application of structural integrity
14. Detiček E. Kastrevc M.: Design of Lyapunov based nonlinear posi-
assessment procedure on an axle pin of a wind turbine.
tion control of electrohydraulc servo systems: Strojniški vestnik
ISSN 0039‐2480, Mar. 2016, vol. 62 Nr.3 str. 163‐170, doi: Fatigue Fract Eng Mater Struct. 2017;40:1284–1294.
https://doi.org/10.5545/sv‐jme.2015.2921 https://doi.org/10.1111/ffe.12654
15. Božić Ž, Mikota M, Schmauder S. Application of the K, J and
ΔCTOD parameters in fatigue crack growth modelling. Tech Gaz.
2011;18(3): 459–466.
Copyright of Fatigue & Fracture of Engineering Materials & Structures is the property of
Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.