International Journal of Fatigue 94 (2017) 288–301

Contents lists available at ScienceDirect

International Journal of Fatigue

journal homepage: www.elsevier.com/locate/ijfatigue

断⼝口显微分析
定量量的 断⼝口⾦金金相 疲劳 传播,繁殖
Quantitative fractography and modelling of fatigue crack propagation in
high strength AerMet!100 steel repaired with a laser cladding process
K.F. Walker a,⇑, J.M. Lourenço b, S. Sun c, M. Brandt c, C.H. Wang d
a
Defence Science and Technology Group, Fisherman’s Bend, Australia
b
Rio Grande do Norte Federal Institute of Technology, IFRN, Natal, RN, Brazil
c
Centre for Additive Manufacturing, RMIT University, GPO Box 2476, Melbourne, VIC 3001, Australia
d
Sir Lawrence Wackett Aerospace Research Centre, School of Engineering, RMIT University, GPO Box 2476, Melbourne, VIC 3001, Australia

a r t i c l e i n f o a b s t r a c t
超⾼高强度钢
Article history: Ultra-high strength steels employed in safety-critical applications, such as AerMet!100 used in aircraft
Received 15 January 2016 拒绝;退回;剔除 标准，条件 瑕疵
起落架 landing gear structures, are managed on very conservative rejection criteria for small defects and repair
Received in revised form 26 June 2016 options are limited. A novel repair technique using laser cladding has recently been developed. In the pre-

_____
Accepted 27 June 2016 疲劳耐久性
sent paper we report a study of the fatigue endurance of AerMet!100 steel components repaired by the
Available online 12 July 2016 断裂⼒力力学
laser cladding process, and a fracture mechanics based model to predict the fatigue endurance of repaired
基准线 !
零件 components. Three different types of samples were tested; baseline AerMet 100 sample with a small
Keywords: 凹⼝口、刻槽、缺陷 引起
Laser cladding repair 电⽕火花加⼯工 electro-discharge machining notch to initiate a crack, as-clad repaired, and as-clad repaired followed
试样，样品。 受到了了。 交变载荷
Crack growth
by heat treatment to relieve residual stresses. The specimens were subjected to cyclic loading under a
序列列 残余应⼒力力 应⼒力力⽐比
Ultra-high strength steel special sequence consisting of constant amplitude segments at two different stress-ratios (ratio of min-
振幅 ⽚片段 扩展，传播
Variable-amplitude loading 变幅加载 imum to maximum cyclic stress). The test results showed that the crack propagation lives from a com-

_____
Spectrum loading 频谱加载 mon initial depth of 0.25 mm for the as-clad samples were significantly longer than the baseline
系数,倍 压缩的
Crack closure samples by a factor of three to four. The longer life is attributed to the beneficial compressive residual
与...密切相关
Stress-intensity factor 应⼒力力强化因⼦子 stresses resulting from the repair process. The model predictions are found to correlate well with the
Quantitative Fractography results of quantitative fractography measurements from samples tested under variable amplitude cyclic
loads.
Crown Copyright " 2016 Published by Elsevier Ltd. All rights reserved.

1. Introduction Recent developments [4–8] around additive layer repair technol-

复原
极端的；过分的 强度
ogy offer a path to a potential solution that can not only restore
⼏几何的
Ultra-high strength steel components, such as aircraft landing the original geometric shape, but to also restore the static strength 静⼒力力强度
疲劳寿命

______
gear components made of AerMet!100, are complex and very and fatigue life properties back to the original as-manufactured
expensive
典型地
to manufacture and replace. Ultra-high strength steels condition or better. 特性 ⼆二次硬化钢
typically account for 5–15% of the airframe structural weight AerMet!100, known as a Co-Ni secondary hardening steel. The
航空与航天
[1,2]. Safety concerns for aerospace applications lead to very strict addition of Co and Ni; (1) increases the dislocation of the FCC
⻢马⽒氏体的
and conservative acceptable damage and repair limits. Because 格⼦子 lattice during martensitic transformation, and位错 (2) promotes the
循环的；周期的 点阵
these components are subjected to very high cyclic operating formation of M2C carbides and precipitation of M3C carbides. The
成型 晶格 碳化物。 析出

___
stress, the formation and propagation of fatigue cracks is a threat high strength and toughness is achieved through secondary hard-
完整性 刚度,韧性
to structural integrity which must be managed based on damage ening precipitation of the M2C carbides during age tempering. 回⽕火
⼀一套⽅方法
tolerance or safe life methodologies. Cracks can start from small Aircraft landing gear components are subjected to very high-
组分，组成物。 粒⼦子。 空洞洞、⽓气孔。 不不连续性 冲击载荷 腐蚀
constituent particles and voids, discontinuities from manufactur- impact loads and corrosive environments. There have been many
失效,故障
ing such as scratches, burrs, nicks and excessive pitting from sur- cases of aircraft landing gear failures, mostly due to corrosion
刻痕,划痕。 ⽑毛刺刺。 断⼝口。 ___ 点蚀 萌⽣生
face treatments, design inadequacies or degradation processes which led to fatigue crack initiation, propagation and final fracture 断裂
不不适当。 ⽼老老化 ____
such as corrosion and fretting [3]. There is therefore a very strong [2,9,10]. Aircraft landing gear design requires minimum weight,
锈蚀。 微振磨损 体积。 性能
motivation to develop viable and cost effective repair strategies. minimum volume, high performance, and long fatigue life. Ultra-
⾦金金属材料料
可⾏行行的 成本效率 high strength metallic materials, such as AerMet!100 ultra-high
strength steel, are commonly employed for manufacturing aircraft
⇑ Corresponding author.
landing gear. However there is lack of knowledge in the literature
E-mail address: kevin.walker@dsto.defence.gov.au (K.F. Walker).

http://dx.doi.org/10.1016/j.ijfatigue.2016.06.031
0142-1123/Crown Copyright " 2016 Published by Elsevier Ltd. All rights reserved.
 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301 289

regarding the repairability of this material by the laser cladding
method.
Fatigue properties for laser cladding repair has been reported on
a number of steels. Chew and Pang [11] described the role of stress
应⼒力力集中系数
concentration factor on the fatigue life of AISI 4340 steel. However,
the surface of the clad layers was not machined and the stress con-
不不匀整
centration factors were as a consequence un-even build height due
重叠,搭接
to the overlapping clad tracks. Hutasoit et al. [7] showed the fati-
表⾯面硬化
gue life of laser cladded hardfacing alloy powders on AISI 4140
steel, where the difference in fatigue life between the different clad
残余拉应⼒力力
materials was due to the tensile residual stresses in the clad layer.
The effect of laser cladding on the fatigue crack growth of ultra-
high strength steels has not been well understood or documented
in the literature. Understanding the fatigue crack growth rate and
behaviour, especially through the clad layer and HAZ, plays a cru-
证明 再现性
cial role for future certification and reproducibility of the repair
process, and hence are necessary for the analysis of this work.
Earlier work [12] involving laser cladding repair of Aermet 100
幅度
steel included fatigue testing under variable amplitude cyclic load-
ing with a maximum stress of 1000 MPa. As-clad and post heat
后热处理理 treated (PHT) samples were compared with a baseline sample with
切⼝口 模拟
an artificial notch to simulate a damaged condition. The variable
__
amplitude loading featured a special sequence of loads at two
平均应⼒力力 mean stress/stress ratio (R), i.e. low R (0.1) and high R (0.7). The
sequence was engineered to create marks on the fatigue crack sur-
manufactured AerMet!100 material, so it was assumed that the
properties for the clad layer and HAZ were both the same as for
分成薄层的
the substrate. The clad repaired case is effectively like a laminated
复合的
composite structure, but the modelling here treated it as a
continuum.
2. Experimental details
2.1. Material
直径
The AerMet!100 round bars, with a diameter of 31.85 mm,
正常化
were supplied in normalised overage condition from Carpenter
Technology (see Table 1 for composition). The bars were then solu-
tion heat treated, by heating to 885 #C for 1 h, followed by air cool,
!73 #C for 1 h, and finally overage tempering at 482 #C for 5 h, in
符合
accordance with AMS standard 2759/2F [24], to achieve a repre-
sentative in-service hardness level of 53–55 HRC.
The powder used for the deposition repair was AerMet! 100
steel (composition shown in Table 1). It was supplied by Sandvik
原⼦子化的；雾化的
Materials Technology, in the form of gas atomized particles in
球状的
the range 45–75 lm, which was mainly spherical in shape, as
筛分
shown in Fig. 1. The powder was further sieved to 50–80 lm just
before the deposition repair. The microstructure of the Aer-
Met!100 substrate consists of soft ferrite and hard martensite.

定量量断层扫描
帮助
faces to facilitate post-test evaluation using Quantitative Fractog-
初步的 执⾏行行
raphy (QF) [13,14]. Preliminary QF was carried out in the earlier
___
_____ 2.2. Laser cladding
铁素体. ⻢马⽒氏体

work [12], but further QF was needed, particularly into the clad
Heat Affected
Zone 热作⽤用and HAZ layers. That work was recently completed and the details Laser cladding process was performed on a TRUMPF TruLaser
区 详情 给料料
are included in this paper. Cell 7020 system, with injecting powder feedstock. The cell is
光纤传输 同轴的
The aims of the research detailed in this paper were as follows: equipped with a fiber-delivered 3.0 kW disk laser and a coaxial
焦点的
laser cladding head with a focal length of 250 mm. The coaxial
氦⽓气
1. Conduct extended Quantitative Fractography including into the powder flow had a focal distance of 8.0 mm. Helium was used as
氩⽓气 防护的
clad and HAZ layers for AerMet!100 specimens repaired using a the carrier gas and argon was used as the shielding gas, at 10
分别地
laser cladding process and tested under fatigue. and 16 l/min respectively. Laser cladding was performed at room
熔化
2. Evaluate the fatigue crack growth characteristics of the three temperature. The shielding gas was used to protect the melt pool
⽅方案 scenarios previously tested, including ‘‘as-damaged”, ‘‘as-clad” from any environmental effects. 熔池
and ‘‘as-clad + PHT”.
3. Develop an analytical crack growth model for the three scenar-
ios and compare the results.

The crack growth model was developed using the strip-yield,

采⽤用
plasticity-induced crack-closure concept adopted in Newman’s
FASTRAN code [15]. The FASTRAN approach has also been included
in the DST Group code known as CGAP [16]. CGAP was used here
because it includes additional features such as a user-friendly
图形界⾯面
graphical interface. The model is based on the Dugdale yield model
修改的,改良的 塑性变形
[17] but modified to leave a region of plastically deformed material
⾮非线性的。 弹塑性
紧紧跟随, in the wake of the advancing crack. The non-linear, elastic-plastic,
随着...⽽而来
plasticity-induced closure approach used here has been exten-
保真度
sively demonstrated to produce high-fidelity fatigue crack growth
simulations [18–22]. The approach has been shown to be superior
弹性断裂⼒力力学
to purely Linear Elastic Fracture Mechanics (LEFM) methods
because it accounts for non-linear plastic behaviour [23]. The
model also accounted for the residual stress field induced by the
cladding repair process. Baseline fatigue crack growth properties
for the model were developed from the best data available from
形态
the literature. Those properties were derived for conventionally Fig. 1. SEM morphology of the AerMet! 100 powder particles.

Table 1
Chemical compositions (wt%) of the AerMet!100 steel substrate and AerMet!100 steel powder as provided by supplier.

C Mn Ni S Cr Si Mo Fe Co
AerMet!100 steel substrate 0.23 0.01 11.13 0.001 3.0 0.02 1.17 Bal. 13.43
AerMet!100 steel powder 0.24 0.86 11.3 0.00 3.1 0.96 1.21 Bal. 13.4
 290 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301

最佳条件,最佳的 ⼯工艺参数
试验
The optimum processing parameters, as determined from clad
trials [12], are summarized in Table 2. These parameters result in
厚度 ⾜足够的; 充⾜足的. 稀释度
a clad layer with a specific clad thickness, with sufficient dilution
(mixing percentage between clad and substrate), no micro-
孔隙率
cracking, and very little porosity. As a high strength alloy, Aer-
!
Met 100 steel is a very defect-sensitive material. Any local stress
集中 过早的
concentration will result in a premature crack initiation and reduce
its fatigue life. Selection of the best laser processing parameters is
very important to minimise defects such as porosity which can be
the initiation site for fatigue cracking.
The microstructure of the AerMet!100 clad layer consists of a
柱状的。 等轴晶粒结构
mixture of fine columnar and equaxial grain structure. Fine equax-
占优势的
ial structure is dominant near the surface of the clad layer, as Fig. 3. Dimensions of the AerMet!100 fatigue specimens, as-manufactured,
细⻓长的
shown in Fig. 2a. Columnar and elongated equaxial grain structure dimensions in mm.
is dominant near the clad interface, as shown in Fig. 2b. The solid-
固化 ification growth is consistent with a decrease in temperature gra-
分界⾯面 All specimens were machined as per Fig. 3 from the 31.85 mm
梯度
dient/solidification speed ratio from the clad interface to the
surface of the clad layer. The formation of the microstructure is round bar material supplied by Carpenter Technology. The samples
奥⽒氏体 板条 all initially looked as per the lower image in Fig. 4.
austenite grain boundary (bright colour region) and lath marten- 配置
Three configurations were tested. Four samples were tested for

__
site within the boundary (dark colour region), as shown in Fig. 2c.
熔融 each configuration in order to obtain statistical data confidence.

A good fusion bond is observed between the AerMet!100 clad layer
and substrate, as shown in Fig. 2d.
The measured hardness of the Aermet 100 clad layer was
approximately 521 HV. After post-clad heat treatment, the hard-
导线,⾦金金属丝
ness of the clad layer increased to 590 HV [12].
2.3. Fatigue test samples and loading
试样
Fatigue specimen was manufactured in accordance with ASTM
尺⼨寸规格
E466-07 (dimensions shown in Fig. 3). The stress concentration
factor K was approximately 1.04 [25].
t
___
The first of the three configurations, ‘‘as-damaged”, was an as-
manufactured specimen with a 0.25 mm macroscopic notch defect
simulating a crack, as shown in detail (a) of Fig. 4. The notch was
wire cut using an Electrical Discharge Machine (EDM) brand Elec-
tronica Eurocut Mark II.
The second configuration, ‘‘as-clad”, was created as follows.
Starting with an as-manufactured specimen, a groove was
轴向的 径向的
machined with an axial length of 10 mm and a radial depth of
__
0.5 mm, to simulate removal of the macroscopic defect (top image
in Fig. 4). Laser cladding was then performed to re-build the

Table 2
Laser cladding processing parameters.

Laser power Number Laser spot Traverse speed Powder flow Step-over Carrier gas Shielding gas
(W) of pass size (mm) (mm/min) rate (g/min) width (mm) flow (L/min) flow (L/min)
800 1 1.3 1400 5.15 0.6 10 16

Fig. 2. Optical microscope and SEM images of the AerMet!100 clad layer (a) 0.2 mm below the surface of the clad layer (b) 0.4 mm below the surface of the clad layer (c)
magnified SEM image 0.2 mm below the surface of the clad layer (d) magnified SEM image of the clad interface.
 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301 291

Fig. 4. Repair process for the AerMet!100 steel fatigue sample showing in the detail (a) the notch for baseline as-damaged specimens. Also is shown the CNC machine grind-
out, laser clad over the grind-out area using AerMet!100 steel powder, and post CNC machine to remove excess clad material.

凹槽
geometry in and around the groove (middle image in Fig. 4).
数控⻋车床 Table 4
Finally, a post-clad CNC machining was performed to remove Detail of the loading block employed in the fatigue tests.
excess clad material, thus restoring the geometry to the as manu-
同样的 Description of Number of Cumulative Load Grouping
factured condition. So these samples looked identical to the origi-
load cycles cycles range
nally manufactured specimens before the introduction of the EDM
Constant 1000 1000 0.1–1.0 CAa start
notch (bottom image in Fig. 4).
被称为 amplitude
The third configuration is referred to as ‘‘as-clad + PHT”, where Marker band 500 1500 0.7–1.0 Stripe 1
PHT indicates Post Heat Treatment. The PHT included an additional Division mark 10 1510 0.1–1.0 DIV
回⽕火
heat treatment: standard tempering heat treatment for Aer- Marker band 500 2010 0.7–1.0 Stripe 2
!
Met 100 alloy at 482 #C for 5 h, in accordance with AMS standards Division mark 10 2020 0.1–1.0 DIV
Marker band 500 2520 0.7–1.0 Stripe 3
[15]. The purpose of the PHT was to relieve any residual stresses Constant 500 3020 0.1–1.0 CA
introduced through the laser cladding procedure.三 The ‘‘as-clad amplitude
+ PHT” configuration 三
构造,外形
looked identical to the ‘‘as-clad” or as- Marker band 500 3520 0.7–1.0 Stripe 1
manufactured specimens. Division mark 10 3530 0.1–1.0 DIV
粗糙度 Marker band 500 4030 0.7–1.0 Stripe2
For all specimens, the surface roughness measurements were
Division mark 10 4040 0.1–1.0 DIV
RA = 0.3 and RZ = 1.6 lm. Marker band 500 4540 0.7–1.0 Stripe 3
Cracks were grown naturally with no artificial crack starters Division mark 10 4550 0.1–1.0 DIV
apart from AerMet!100 baseline as-damaged described above for Marker band 500 5050 0.7–1.0 Stripe 4
the ‘‘as-damaged” case. Constant 500 5550 0.1–1.0 CA
详细说明 amplitude
The mechanical properties for AerMet!100 are detailed in Marker Band 500 6050 0.7–1.0 Stripe 1
Table 3. Division Mark 10 6060 0.1–1.0 DIV
取样⽚片 受到了了 全⾃自动化
The coupons were subjected to cyclic loading at 5 Hz in a fully- Marker band 500 6560 0.7–1.0 Stripe 2
伺服液压控制的 Division mark 10 6570 0.1–1.0 DIV
automated, computer-controlled, closed-loop, servo-hydraulic uni-
单轴的 测⼒力力传感器器 Marker band 500 7070 0.7–1.0 Stripe 3
axial test machine. A 500 kN load cell was used in a tension-tension
Division mark 10 7080 0.1–1.0 DIV
Marker band 500 7580 0.7–1.0 Stripe 4
Table 3 Division mark 10 7590 0.1–1.0 DIV
Mechanical properties for AerMet!100. Marker band 500 8090 0.7–1.0 Stripe 5
Constant 500 8590 0.1–1.0 CA end of
ru (MPa) 1999 amplitude block
ry (MPa) 1779 Constant 1000 9590 0.1–1.0 CA new
E (MPa) 194400 amplitude block
p
KIC (MPa m) 140 a
CA, Constant Amplitude.

Fig. 5. Normalised load sequence applied to test coupons.

292 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301

external loading. Tests were conducted under laboratory air condi- and also in range (0.7–1.0, so R = 0.7 and range = 0.3). The Marker
tions at a frequency of 5 Hz. During the period of testing the tem- Band Groups are also referred to as ‘‘Stripe X” where X is a number
湿度
perature was kept within the range 18.5–21.5 #C. Humidity was ranging from 1 to 5. The marker bands are in three distinct groups
in the range 31.38–74.63% RH. The maximum load applied resulted with 3, 4 and 5 cycles at R = 0.7 stripes as detailed in Table 4.
in a peak nominal stress in the 20 mm diameter test section of Specimen numbers for the three configurations are detailed in
1000 MPa. The normalized load sequence is presented in Fig. 5. Table 5.
Each block consists of 8590 cycles.
恒定振幅 ⼆二 in Table 4. There
The details for one block of loading are given
3. Residual stress evaluation
are 1000 and 500 cycles of Constant Amplitude (CA) and at the
beginning and end of each block respectively. This means that at
总计 Residual stresses can have an important effect on the overall
the change over from one block to the next there are a total of 中⼦子。 衍射
外加载荷 fatigue endurance and crack growth behaviour. Neutron diffraction
1500 cycles of CA ranging from 0.1 to 1.0 of the applied load (i.e.
residual stress measurements were conducted as part of the previ-
R = 0.1), which therefore produces a larger band of crack growth
ous work [12], but a review of the work revealed several issues.
than the Marker Band Groups which are fewer in number (500),
Firstly, the geometry for the earlier residual stress measurement
was not the same as for the specimens tested under fatigue load-
ing. In the earlier work [12], the neutron diffraction measurements
Table 5 were performed on samples which were 32 mm diameter, and the
Specimen numbering system. clad length was 45 mm. That is very different to the fatigue test
Configuration Specimen numbers samples where, in the test section, the diameter was 20 mm and
the clad length was 10 mm. Secondly, the earlier neutron diffrac-
As-damaged 11, 12, 13, 14, 15 逼真度
As-clad 6, 7, 8, 9, 10 tion was not well targeted in terms of fidelity of measurement in
轴向分量量
As-clad + PHT 1, 2, 3, 4, 5 the exial direction in particular. The axial component (see Fig. 6)
is most important for fatigue in this case, and the first 2 mm of
the depth is very important in terms of the impact on fatigue crack
development and propagation.
So an updated neutron diffraction residual stress measurement
was performed at the Australian Nuclear Science and Technology
Association (ANSTO). The smallest possible gauge volume was used
to enable the axial measurements at 0.5 mm increments of depth.
The results for both the ‘‘as-clad” and ‘‘as-clad + PHT” are shown in
Fig. 7. The distribution of the resultant residual stress can be
标绘
described by the following relationship [26], and also plotted in
Fig. 7;
2
1 ! ðz=dÞ
rR ðzÞ ¼ rR;max 2 2
ð1Þ
½1 þ ðz=dÞ '

Fig. 6. Definition of regions and directions for the clad repair samples.
where rR;max ¼ !300 MPa and d ¼ 2:0 mm for the ‘‘as-clad”, and
rR;max ¼ !200 MPa and d ¼ 2:0 mm for the ‘‘as-clad + PHT”. The
透⼊入
residual stress was found to penetrate much deeper than those
喷丸硬化处理理
induced by shot peening [26]. This residual stress distribution satis-
R1
fies the zero net force condition, i.e., 0 rR ðzÞdz ¼ 0.

Fig. 7. Neutron diffraction residual stress results for (a) as-clad, (b) as-clad PHT.
 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301 293

Fig. 8. Image of AerMet!100 baseline as-damaged Specimen 14 showing (a) fracture surface taken using flat diffuse illumination; (b) showing the stable crack using crossed
polarizers; and (c) also using flat diffuse illumination.

Fig. 9. Image of the origin area of AerMet!100 baseline as-damaged produced using flat diffuse illumination, a high contrast setting on the camera and subsequent processing
using the CLAHE filter.
 294 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301
Fig. 10. QF based crack growth measurements for baseline as-damaged case.
断⼝口⾦金金相
4. Test results including quantitative fractography
光学的
QF was performed using a contrasting band method, using opti-
cal and scanning electron microscopy. A Phillips XL30 SEM, LEICA
EZ24D optical microscope, Canon EOS 50D macro lens camera were
滤⾊色器器
used to capture the images using various lighting and filter tech-
niques as described in this paper.
4.1. Baseline as-damaged
The QF required careful examination using a variety of illumina-
tion methods. This is shown in Fig. 8. Fig. 8(a) shows the fatigue
外貌。 平⾯面。 漫射。 照明
surface appearance using flat diffuse illumination with the contrast
enhanced. The image shown in Fig. 8(b) was produced using
交叉偏振器器
crossed polarizers where the specimen was illuminated with plane
polarized light and the light reflected from the specimen was
平⾯面偏振
Fig. 11. Fatigue surface for AerMet!100 as-clad Specimen 6.
filtered with a plane polarizing filter where the plane of polariza-
tion was set at 90 degrees to the first filter. The image shown in
Fig. 8(c) was also produced using flat diffuse illumination. Image
processing using a process known as contrast limited adaptive his-
togram equalisation (CLAHE) [27] was used to generate more
detail. The cross polarizers produced clearer images than the flat
diffuse illumination technique.
匹配
The process of matching up the markings identified on the fati-
gue surface with the applied load sequence is shown in Fig. 9. The
3, 4 and 5 stripe marker band sequence separated by the high-
振幅
amplitude low-R constant amplitude loading is very helpful in pro-
ducing identifiable markings to assist the determination of crack
growth curves. The crack growth curve for the as-damaged case
is shown in Fig. 10. Crack depth is defined to include the initial
EDM notch which had a depth of approximately 0.25 mm. The ini-
tial and updated QF measurements produced very consistent
results as shown in Fig. 10.
4.2. As-clad
QF was performed on four specimens (Specimens 6, 7, 8 and 9)
of the as-clad configuration. The fatigue surface for Specimen 6 is
shown in Fig. 11. The clad layer and the associated HAZ are clearly
evident.
The CLAHE image processing method mentioned earlier was
used to generate more detail in the HAZ and clad layers.
Identification of the markings on the fatigue surface became
渐进地
progressively more difficult as the crack initiation site is
approached. Nevertheless, it was possible to identify some mark-
ings in the HAZ and the clad layer. This was performed for one
specimen, Specimen 6, and the results are shown in Fig. 12, indicat-
回到
ing that the crack growth was estimated right back to the early
stage of fatigue. The open black circle symbols indicate data with
very high level of confidence (>95%). The open 1red square symbols
are data which appeared to be in ‘‘pairs”, indicating that these marks
may have been produced by the alternation between the marker
band section and the constant amplitude section. These data are
regarded as probable. The open blue triangle symbols show esti-
mated data based on extrapolation and the measurement using the
1
For interpretation of color in Fig. 12, the reader is referred to the web version of
this article.
 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301
Fig. 12. Crack depth plot for as-clad Specimen 6, based on updated QF.
updated QF method. The location of the HAZ and the clad layer are
included in the plot. The initial effective crack depth was estimated
at 47 lm. The crack origin area was closely examined as shown in
空洞洞。 污染
Fig. 13. The origin was a void or contamination as shown in the
孔隙率 __ 不不对称的
lower edge. The void/porosity was about irregular and about
100 lm deep. The void was exposed at the surface during post clad
machining. It is therefore reasonable to accept the equivalent initiat-
ing crack depth of 47 lm. It is also interesting to note that another
void of about 20 lm diameter was located nearby at about 200 lm
below the surface.
The results for all four as-clad specimens investigated are
shown in Fig. 14.
___
4.3. As-clad PHT
The QF results of the as-clad PHT samples are shown in Fig. 15.
Specimen 3 was subjected to additional detailed QF examination
similar to that for Specimen 6 in the ‘‘as-clad” condition. The
Fig. 13. Crack origin location for as-clad Specimen 6 subjected to updated QF.
295
fatigue surface from Specimen 3 is shown in Fig. 16 with the crack
origin and progression clearly evident.
4.4. Discussion of results
In order to compare the crack growth behaviour for the three
different conditions it is necessary to establish a common starting
point for the crack growth. The as-damaged case has the largest
starting crack size of about 0.25 mm, so a comparative plot was
constructed with the common starting size of close to 0.25 mm,
with cycles accumulated after that size plotted on the horizontal
轴线
axis. The results are shown in Fig. 17. When plotted in this way,
the trend of crack growth behaviour is clearly evident. The ‘‘as-
clad” life of about 78,000 cycles is more than four times longer
than the ‘‘as-damaged” case which showed a life of about 17,000
cycles. The life of the ‘‘as-clad + PHT” is about 23,000 cycles which
is about 30% longer than for the ‘‘as-damaged” case. Given that the
等级,量量
magnitude of the compressive residual stress was significantly less
for the ‘‘as-clad + PHT” case compared with the ‘‘as-clad” case,
有意义
these crack growth results make sense, at least in terms of trend
and direction. The question of how much sense can be made of
the results is explored further in the next section with the develop-
ment and evaluation of the analytical crack growth model.
5. Crack growth model
The crack growth model was developed using the strip-yield,
plasticity-induced crack-closure concept adopted in Newman’s
FASTRAN code [15]. The FASTRAN approach has also been included
in the DST Group code known as CGAP [16]. CGAP was used here
because it includes additional features such as a user-friendly
graphical interface. The model is based on the Dugdale yield model
[17] but modified to leave a region of plastically deformed material
in the wake of the advancing crack. The non-linear, elastic-plastic,
plasticity-induced closure approach used here has been exten-
sively demonstrated to produce high-fidelity fatigue crack growth
simulations [18–22]. The approach has been shown to be superior
to purely Linear Elastic Fracture Mechanics (LEFM) methods
because it accounts for non-linear plastic behaviour [23]. The
model also accounted for the residual stress field induced by the
cladding repair process. Baseline fatigue crack growth properties
for the model were developed from the best data available from
296 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301

Fig. 14. QF based crack growth measurements for as-clad case.

Fig. 15. QF results for three as-clad PHT samples.

Fig. 16. Fatigue surface from Specimen 3, as-clad + PHT.

 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301 297

Fig. 17. Comparison of QF crack growth results for all three conditions from a common starting size of 0.25 mm depth. Note that HAZ and clad layer regions apply only to the
‘‘as-clad” and ‘‘as-clad + PHT” cases.

Fig. 18. FASTRAN/CGAP Baseline crack growth rate model for AerMet!100 Steel in Lab Air conditions.

Table 6 Table 7
FASTRAN Crack growth model input parameters. DKeff vs rate curve for FASTRAN model.
p
Model parameter Value DKeff (MPa m) Rate (m/cycle)
ry 1779 MPa 1.93 2.63e!10
ru 1999 MPa 2.0 1.0e!9
E 194400 MPa 100.0 8.0e!6
p
KIC 140.0 MPa m
C3 3.29
C4 !0.87 分成薄层
C6 5 the substrate. The clad repaired case is effectively like a laminated
C7 2 composite structure, but the modelling here treated it as a

a 1.8
the literature. Those properties were derived for conventionally
manufactured AerMet!100 material, so it was assumed that the
properties for the clad layer and HAZ were both the same as for
⼀一
continuum.
相互关联
The model was required to correlate experimental test results
圆柱的
from a cylindrical bar specimen (test section diameter 20 mm,
see Fig. 3) which was subjected to a special load sequence, see
Fig. 5. The maximum applied stress was 1000 MPa. The fatigue
tests were conducted in laboratory air conditions. Environmental
298 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301

effects can be significant for this material [28], so it was very
important to ensure that the model closely matched the test con-
ditions and environment. The test included an EDM notch about
0.25 mm deep to initiate the cracking, see Fig. 8. The crack quickly
半圆的
adopted an approximately semi-circular shape as seen in Fig. 8.
The fatigue cracking was modelled as a semi-circular surface
flaw in a 20 mm diameter bar. The geometry factors were deter-
mined using the AFGROW code [29]. This was applied for both
the as-damaged case and also the cases with the clad and HAZ lay-
ers. The clad cases effectively created a layered, composite struc-
ture. It is important to note however that the cracking adopted a
consistent and approximately semi-circular shape very quickly
including in the clad and HAZ layers. So the assumption of the
semi-circular geometry is considered reasonable.
A literature review revealed various sources of fatigue crack
growth rate data for AerMet!100 steel. But there was only limited
data available for laboratory air conditions [30]. Lab air data was
available for R = 0.5, and also two load reduction tests at constant
p
DK with constant Kmax values of 22.0 and 55.0 MPa m. The con-
stant Kmax tests therefore produced a varying stress ratio R. In
p
the case of the Kmax = 22.0 MPa m test, R varied from 0.3 to
p
0.87. In the case of the Kmax = 55.0 MPa m test,i the R value varied
from 0.68 to 0.96. The constant Kmax tests were very important
临界,阈值
because they provided information on the threshold behaviour.
The R = 0.5 test provided mid-region steady state data and also
the behaviour approaching fracture. These data are employed as
input to the crack growth model.
The input values of the model are shown in Fig. 18 and Tables 6
and 7. Overall, the model captured the full range of behaviour from
threshold through to near fracture reasonably well. Other model
details are as follows:
The fatigue crack growth rate properties of the clad and HAZ
layers may be different to the substrate. But those data have not
yet been determined. So at this stage it was assumed that the clad

Fig. 19. Model crack growth curve compared with experimental result, as-damaged case.

Fig. 20. Model crack growth curve compared with experimental result, ‘‘as-clad” case.
 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301
与…⼀一致
and HAZ fatigue crack growth rate properties were consistent with
the conventionally manufactured AerMet!100 substrate material.
有效性
Further work is needed to test the validity of this assumption.
5.1. Model predictions
切⼝口
The crack initiated at a tangential EDM slot about 0.25 mm deep
as shown in Fig. 8. The model predictions correlate very well with
the experimental results of the ‘‘as-damaged” specimens, as shown
in Fig. 19.
构造,外形
To apply the crack growth model to the ‘‘as-clad” configuration,
it is important to consider the beneficial effect of the residual com-
pressive stress of about 300 MPa at the outer surface in the clad
layer (see Fig. 7). This effect can be accounted for by adding the
强度因⼦子 与……有关
stress intensity factor pertinent to the residual stress (obtained
from the neutron diffraction measurements presented in Fig. 7)
外加载荷
to that due to the applied load. Here the stress intensity factor
due to the residual stress and the applied stress is given by:
Z a
K R ðaÞ ¼ rR ðzÞGðz; aÞdz
0
Fig. 21. Model crack growth curve compared with experimental result, as-clad PHT case.
Fig. 22. Comparison of analysis and QF for all three conditions from smallest available initial crack size of 0.25 mm.
299
where rR ðzÞ and rapp are the residual stress and the applied stress,
表示
respectively. The term Gðz; aÞ denotes the weight function of stress
intensity factor due to a pair of point load acting at location x in an
边,缘
edge crack of length a. Numerical solution of this weight function is
available in AFGROW [29]. An initial crack size of 47 lm was used in
与…⼀一致
line with the QF. The predicted crack growth curves are shown in
Fig. 20. The analysis result assuming no residual stress is also shown
for comparison. Accounting for the residual stress improved the
analysis in comparison with the test result, but the predicted life
was still about half of the test result (Specimen 6, see Fig. 20).
The difference is more significant than desired, but is considered
固有的
reasonable given the complexity and uncertainty inherent in this
type of analysis.
施加
The model was then applied to the ‘‘as-clad + PHT” case using
the same approach as described earlier for the ‘‘as-clad” case. The
result is shown in Fig. 21. The appropriate comparison is between
the analysis and the test result for Specimen 3 where the QF was
评估为
performed back to the origin of the cracking, assessed as from a
size of 0.104 mm. The analytical life of 50,405 cycles compares well
ð2Þ with the test demonstrated life for Specimen 3 of 69,430. So the
analysis in this case under-predicted the life by about 27%.
300 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301
Fig. 23. Comparison of analysis and QF for all three conditions from 0.7 mm crack size.
A plot was constructed to show the comparison of as much of
the fatigue crack growth as possible for all three conditions, for
both analysis and test. This is shown in Fig. 22. The comparison
was performed from the smallest possible initial crack depth
which was about 0.25 mm. So the cycles shown in the x-axis on
the plot are the cycles applied since the crack size was 0.25 mm.
The plot clearly shown the trend in crack growth behaviours with
the ‘‘as-damaged” case showing the shortest life, the ‘‘as-clad
+ PHT” showing a longer life, and the ‘‘as-clad” showing the longest
life of all. A similar trend is evident for the analysis. The ‘‘as-
damaged” life from the 0.25 mm crack was over-estimated by
the analysis by about 36%, the ‘‘as-clad” life was under estimated
by about 47%, and the ‘‘as-clad = PHT” life was over estimated by
about 34%. These differences are considered reasonable given the
uncertainty expected in this type of modelling.
In order to evaluate the results for larger crack lengths, another
comparison plot was constructed based on crack growth from a
depth of about 0.7 mm. The result is shown in Fig. 23. The ‘‘as-
damaged” life from the 0.7 mm crack was over-estimated by the
analysis by about 30%, the ‘‘as-clad” life was over estimated by
about 3%, and the ‘‘as-clad = PHT” life was over estimated by
110%. Again, these differences are considered reasonable given
the uncertainty expected in this type of modelling.
6. Discussion
The fatigue crack growth model produced prediction of the fati-
gue behaviour of AerMet!100 high strength steel in reasonable
correlation with the results measured using optical microscopy
based quantitative fractography (QF) for the ‘‘as-damaged” case.
The model matched the QF results very well for the first stage of
growth from about 0.25 mm to 0.5 mm. After that point the FAS-
TRAN/CGAP analysis produced slightly slower growth and a longer
life. The final fatigue life according to the FASTRAN/CGAP analysis
was about 23,000 cycles compared to the test/fractography result
of about 17,000 cycles for the as-damaged samples, so the life
was over-predicted by about 36%. This is not a significant discrep-
ancy considering that the crack actually started from a tangential
EDM wire cut slot and quickly transitioned into an approximately
semi-circular type of flaw.
For the ‘‘as-clad” case, the compressive residual stresses
(300 MPa at the surface) induced by the laser cladding process
had a significant beneficial influence, i.e. the crack growth rate
was significantly lower and the life was longer. An initial crack size
the same as the fractography measurement for Specimen 6 was
used in the analysis, i.e. a semi-circular flaw of depth 47 lm. In this
case the correlation with the QF results was reasonable, with the
life being under-predicted by about 47%.
For the ‘‘as-clad + PHT” case, the compressive residual stress
was significantly less in magnitude (200 MPa at the surface) com-
pared with the ‘‘as-clad” case, and as a result the crack growth was
much faster and the life was reduced. In this case the analysis over-
predicted the life by about 34%. Again, these results are considered
to compare reasonably well for an analysis of this type.
A very important outcome from this work is that the crack
growth behaviour in the clad material was very consistent with
the behaviour in the conventionally manufactured AerMet!100
material. The limiting factor in terms of total life is the size of
the crack initiating feature, combined with the residual stress
condition. It is reasonable to expect that if porosity could be com-
pletely eliminated, then the fatigue properties of the clad material
would be very similar to conventionally manufactured material.
7. Conclusion
The effectiveness of a laser cladding repair process to restore
the fatigue properties of a damaged, ultra-high strength
AerMet!100 Steel component has been demonstrated. The repair
process was found to introduce beneficial compressive residual
stresses which retard fatigue crack propagation. These beneficial
stresses resulted in a fatigue crack growth life increase by a factor
of about 4.0 (starting from a common initial 0.25 mm deep crack)
for the ‘‘as-clad” case.
A clad followed by heat treatment case, ‘‘as-clad + PHT”, was
also evaluated. It was found that the PHT reduced the magnitude
of the compressive residual stress at the surface from about 300
to 200 MPa. So although the magnitude was reduced, a significant
amount of residual stress remained present. As a result, the fatigue
crack growth life for the ‘‘as-clad + PHT” case was longer than the
‘‘as-damaged case” was longer, but in this case the life improve-
ment factor was less, i.e. about 1.3.
Quantitative Fractography was applied very effectively which
provided a very important insight into the fatigue crack growth
behaviour. Fatigue crack growth characteristics were determined
 K.F. Walker et al. / International Journal of Fatigue 94 (2017) 288–301
right down into the HAZ and clad layers. For the clad repair cases, [9]
the cracking always initiated at the surface which is a very impor-
[10]
tant observation.
A non-linear, elastic-plastic analytical crack closure base model
was developed and applied to the crack growth scenarios mea- [11]
脚本
sured here. Quantitative measurement of the residual stress profile断⾯面，剖⾯面
[12]
was a key input into the modelling. The fundamental crack growth
rate behaviour for the model was derived from literature data from
[13]
testing of conventionally manufactured AerMet!100 material. The
明确地
residual stress was explicitly modelled and the correlations against
the QF measurements were generally reasonable to good. These [14]
factors suggest that the fundamental crack growth properties of
the clad material are consistent with the conventionally manufac- [15]
tured material. Further tests will be needed to fully explore that
[16]
conclusion. It was also clear that the cladding process produces a
冶⾦金金学的
very good metallurgical bond with no issues of concern at the
interface between the clad and the substrate. [17]
Acknowledgements
The authors would like to acknowledge the financial support
from the Brazilian Research Funding - CAPES/Science without bor-
ders program – process number registration BEX 9296/13-1. The
authors also thank Mr Noel Goldsmith from DST Group for his
work on the updated fractography. Our sincere thanks to Dr Simon
Barter for his contribution to the Quantitative Fractography, and to
Dr Anna Paradowska at ANSTO for the neutron diffraction residual
stress measurements.
References
[1] Campbell FC. Manufacturing technology for aerospace structural
materials. Elsevier; 2011.
[2] Infante V, Reis L, de Freitas M. Failure analysis of landing gears trunnions due
to service. Eng Fail Anal 2014;41:118–23.
[3] Roiko A, Solin J. Measurement of small cracks initiating from inclusions,
Focused Ion Beam notches and drilled holes. Int J Fatigue 2014;62:154–8.
[4] Desale GR, Paul CP, Gandhi BK, Jain SC. Erosion wear behavior of laser clad
surfaces of low carbon austenitic steel. Wear 2009;266(9–10):975–87.
[5] Ganesh P, Moitra A, Tiwari P, Sathyanarayanan S, Kumar H, Rai SK, et al.
Fracture behavior of laser-clad joint of Stellite 21 on AISI 316L stainless steel.
Mater Sci Eng: A 2010;527(16–17):3748–56.
[6] Nagaoka T, Kimoto Y, Watanabe H, Fukusumi M, Morisada Y, Fujii H. Friction
stir processing of a D2 tool steel layer fabricated by laser cladding. Mater Des
2015;83:224–9.
[7] Hutasoit N, Luzin V, Blicblau A, Yan W, Brandt M, Cottam R. Fatigue life of laser
clad hardfacing alloys on AISI 4130 steel under rotary bending fatigue test. Int J
Fatigue 2015;72:42–52.
[8] Chen JY, Conlon K, Xue L, Rogge R. Experimental study of residual stresses in
laser clad AISI P20 tool steel on pre-hardened wrought P20 substrate. Mater Sci
Eng: A 2010;527(27–28):7265–73.
301
Hou W, Zhang W, Tang Q, Liu X, Wang Z, Ding M. Failure analysis of cracks in
the upper shaft of torque arm. J Aircraft 2011;48(2):601–5.
Franco LAL, Lourenço NJ, Graça MLA, Silva OMM, de Campos PP, von Dollinger
CFA. Fatigue fracture of a nose landing gear in a military transport aircraft. Eng
Fail Anal 2006;13(3):474–9.
Chew Y, Hock Lye Pang J. Fatigue life prediction model for laser clad AISI 4340
specimens with multiple surface cracks. Int J Fatigue 2016;87:235–44.
Lourenço JM, Sun SD, Sharp K, Luzin V, Klein AN, Wang CH, et al. Fatigue and
fracture behavior of laser clad repair of AerMet! 100 ultra-high strength steel.
Int J Fatigue 2016;85:18–30.
Barter SA, Molent L, Wanhill RJH. Marker loads for quantitative fractography of
fatigue cracks in aerospace alloys. In: 25th international committee on
aeronautical fatigue symposium, Rotterdam, The Netherlands.
Barter SA. The quantification of fatigue crack initiators in aluminium alloy
7050–T7451 using quantitative fractography. Int J Struct Integrity 2011;2
(3):243–63.
Newman Jr JC. FASTRAN a fatigue crack growth life-prediction code based on
the crack-closure concept version 5.4. Fatigue & Fracture Associates LLC; 2013.
Hu W, Wallbrink C. CGAP Capabilities and Application in Aircraft Structural
Lifing. In: 15th Australian international aerospace congress (AIAC15),
Melbourne, Australia. Melbourne, Australia.
Dugdale DS. Yielding of steel sheets containing slits. J Mech Phys Solids 1960;8
(2):100–4.
[18] Newman Jr JC. A crack-closure model for predicting fatigue crack growth
under aircraft spectrum loading. In: Chang JB, Hudson CM, editors. Methods
and models for predicting fatigue crack growth under random loading, ASTM
STP 748. ASTM; 1981. p. 53–84.
[19] Newman Jr JC, Anagnostou EL, Rusk D. Fatigue and crack-growth analyses on
7075–T651 aluminium alloy coupons under constant- and variable-amplitude
loading. Int J Fatigue 2014;62:133–43.
[20] Newman Jr JC, Walker KF, Liao M. Fatigue crack growth in 7249–T76511
aluminium alloy under constant-amplitude and spectrum loading. Fatigue
Fract Eng Mater Struct 2015;38(5):528–39.
[21] Walker KF, Hu W. A critical evaluation of three models for short and long
fatigue crack growth. In: 24th International committee on aeronautical fatigue
(ICAF) symposium, Naples, Italy.
[22] Walker KF, Wang CH, Newman Jr JC. Closure measurement and analysis for
small cracks from natural discontinuities in an aluminium alloy. Int J Fatigue
2016;82:256–62.
[23] Newman Jr JC, Ruschau JJ. The stress-level effect on fatigue-crack growth
under constant-amplitude loading. Int J Fatigue 2007;29(9–11):1608–15.
[24] Heat treatment of low-alloy steel parts minimum tensile strength 220 ksi
(1517 MPa) and higher. In: Aerospace Material Specification (AMS) 2759/2F.
PA USA: SAE International; 2010.
[25] Pilkey WD, Pilkey DF. Peterson’s stress concentration factors. 3rd ed. New
Jersey, USA: John Wiley and Sons; 2008.
[26] Wang CH, Barter SA, Liu Q. A closure model to crack growth under large-scale
yielding and through residual, stress fields. J Eng Mater Technol–Trans ASME
2003;125(2):183–90.
[27] Zuiderveld K. Contrast limited adaptive histogram equalization. In: Heckbert,
editor. Graphics gems IV. Pittsburg PA, USA: Academic Press; 1994.
[28] Lee EU. Fatigue crack growth in AerMet 100 Steel, NADC-91111-
60. Warminster PA, USA: Naval Air Development Centre; 1991.
[29] Harter JA. AFGROW user’s guide and technical manual. AFRL-VA-WP-TR-2004;
2004.
[30] Newman JA, Forth SC, Everett Jr RA, Newman Jr, JC, Kimmel WM. Evaluation of
fatigue crack growth and fracture properties of cryogenic model materials,
NASA/TM-2002-211673 ARL-TR-2725, NASA; 2002.