applied

sciences
Article
Lifetime Assessment and Optimization of a Welded
A-Type Frame in a Mining Truck Considering
Uncertainties of Material Properties and Structural
Geometry and Load
Chengji Mi 1,2 , Wentai Li 1 , Xuewen Xiao 1 , Haigen Jian 1, *, Zhengqi Gu 1,2 and Filippo Berto 3
1 Hunan University of Technology, Zhuzhou 412007, China; michengji_86@126.com (C.M.);
liwentai2019@126.com (W.L.); 13838@hut.edu.cn (X.X.); guzhengqi63@126.com (Z.G.)
2 State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University,
Changsha 410082, Hunan, China
3 Norwegian University of Science and Technology, 7491 Trondheim, Norway; filippo.berto@ntnu.no
* Correspondence: jianhaigen2001@163.com




Received: 19 December 2018; Accepted: 26 February 2019; Published: 4 March 2019


Abstract: In order to improve the fatigue performance of a welded A-type frame in a heavy
off-road mining truck, a novel method was presented to implement lifetime and weight collaborative
optimization while considering uncertainties in geometry dimension, material properties, and bearing
load. The mechanical and cyclic material parameters were obtained from experimental work to
characterize the base metal and the weldment. The finite element model of a welded A-type frame
was constructed to analyze stress distribution and predict fatigue life, the force time histories of
which were acquired from multi-body dynamics simulation. The simulated failure position and
fatigue life had a good agreement with the actual results. Then, both structural lifetime and weight
were considered as optimization objectives. The thickness of main steel plates and elastic and cyclic
material parameters were chosen as uncertain design variables as well as main loads at connection
locations. The fifty sample points in the light of Latin hypercube sampling method and its responses
calculated by finite element analysis were supposed to build the approximation model based on
the Kriging approximation method. After its fitting precision was guaranteed, the non-dominated
sorting genetic algorithm II (NSGA-II) was utilized to find the optimal solution. Finally, the fatigue
life of a welded A-type frame was increased to 2.40 × 105 cycles and its mass was lessened by 8.2%.
The optimized results implied that good fatigue performance of this welded A-type frame needs
better welding quality, lower running speed for downhill and turning road surface, and thicker
front plates.

Keywords: welded structure; fatigue characterization; multi-source uncertainty; multi-objectives

optimization; Kriging approximation method

1. Introduction
The welded A-type frame of steering system is considered as one main bearing component for
mining dump trucks. Its fatigue performance is seriously concerned when the truck runs on poor
road surface with hundreds of tons of goods. This frame is welded with different thickness of plates;
however, fatigue cracks often start from weld toe located at stress concentration areas [1].
There are several methods to predict fatigue life of a welded structure. These could be categorized
into nominal stress [2,3], structural stress [4], local notch [5], and fracture mechanics approaches [6].
The fracture mechanics method firstly assumes that the crack or crack-like flaw exists in the weld metal,

Appl. Sci. 2019, 9, 918; doi:10.3390/app9050918 www.mdpi.com/journal/applsci

Appl. Sci. 2019, 9, 918 2 of 19

and the stress intensity factor or J-integral is considered as the fatigue damage parameter, related with
lifetime or crack growth rate [7]. The local notch approach is defined by using local stress, local strain,
or local energy at the notch tip to estimate fatigue strength. The advantage of this method is that the
effect of notch characteristics and cyclic material properties could both be included [8,9]. The structural
stress method focuses on a linearly distributed stress along the thickness without considering the
stress concentration. Superposing the mechanical responses caused by forces and moments, the linear
elastic finite element analysis is often implemented to calculate structural stress and strain, which is
coupled into the mathematical model to assess fatigue life [10]. The nominal stress amplitude in
the critical cross-section is treated as a fatigue damage parameter without consideration of the stress
concentration at the weld toe and dissimilar material properties. This method is based on the S-N
curve and preferred in many engineering areas because of its simplicity, such as in vehicles, vessels,
and bridges [11]. According to those rules, Shao et al. [12] estimate lifetime of drive axle housing for
a mining truck by measuring strains at hot spots, which were transformed to stress spectrum. Then,
the stress amplitude and stress mean with their cycles were counted using the Rain-flow method,
and Morrow’s stress-life equation was utilized to obtain the service lives. Mi et al. [13,14] conducted a
fatigue life assessment of the frame by combining multi-body dynamics analysis with finite element
analysis to acquire the stress and strain spectrum. The service life was calculated as Palmgren-Miner’s
linear damage accumulation rule and normal S-N curve. In order to exactly present fatigue life
distribution for a welded mining truck’s frame at critical areas, even the total strain energy density
was used as a fatigue damage parameter to predict lifetime, based on fitting the relationship between
fatigue life and elastic and plastic strain energy density [15]. Moreover, the local strain energy density
averaged over the control volume surrounding the crack tip in the weakest point of the roller was
supposed to characterize the fatigue strength of welded joints [16,17].
When the actual fatigue life or predicted lifetime of engineering structures is less than the
required one, the fatigue optimization design is indispensable. In order to simultaneously enhance the
performance and robustness of the fatigue life for a truck cab, Fang et al. [18] present a dual Kriging
surrogate modeling method to overcome the limitation of classical dual response surface method.
The reliability-based design optimization is proposed by the Aoues et al. [19,20] and aims to find the
most balanced design through a compromise between cost and safety when uncertainties affect the
system. Moreover, Gao et al. [21] constructed a lightweight optimization model with static strength
and weld fatigue constraints for the welded frame of a three-axle bogie, which is solved by an adaptive
sequential approximation optimization method based on Kriging models.
Most of optimization procedures mentioned above predominately focus on the single optimization
objective, such as minimizing structural weight or maximizing lifetime. Sometimes the re-optimization
design is needed in the presence of contradictory results between different required performances.
In the meanwhile, the variables for the general optimization design only depend on the structural
dimensions or process parameters, while the material properties or loads are deemed to be the
deterministic values. However, the material parameters are related with artificial manufacturing
procedures, and they should be random due to uncertain technological conditions, as well as loads in
the utilizations. Therefore, this paper presents a fatigue life and weight co-optimization method to
obtain the compatible optimal results for a welded A-type frame in a mining dump truck. Moreover,
it considers that there are some deviations in structural geometry in the design stage, and material
properties are also varied under different manufacturing processes, as well as loads in different
usage conditions. The multi-source uncertainty variables including structural geometry dimensions,
material properties, and loads were all considered when conducting multi-objective optimization
design in this paper.
Then, the monotonic tensile and fatigue tests of welded joints were studied to obtain the
mechanical material parameters for the both base metal and weld metal and cyclic properties for the
weld metal. The strain-life curve was considered as the calculation basis in the following lifetime
prediction. The micro properties were also observed to understand the fatigue characteristics of this
 Appl. Sci. 2019, 9, 918 3 of 19
Appl. Sci.Sci.
Appl. 2019, 9, 9,
2019, x FOR PEER
x FOR PEERREVIEW
REVIEW 3 of
3 of 19 19

analysis weld metal. A multi-body dynamic modelits was built tolife

obtain thecalculated
load history as
andperits accuracy was
analysisofofthis
thisframe
framewaswasperformed
performed and and its fatigue was
fatigue life was calculated as per the theManson-Coffin
Manson-Coffin
verified by comparing the simulated results with the experimental ones. The finite element model of
equation.
equation. The
weldedThe multi-objective
multi-objective
A-type
optimization
frame was alsooptimization
function
constructed andfunction
was
validated. was
builtand
built
The elastic
byplastic
by considering
considering theanalysis
the
finite element
multi-source
multi-source
uncertainties
of this in
uncertainties indifferent
frame different stages. The
stages.and
was performed Theits50
50 sample
sample
fatigue points
life was wereasobtained
were
calculated obtained fromthe
from
per the Manson-CoffintheLatin
Latin hypercube
hypercube
equation.
sampling
sampling
Themethod.
method.BasedBased
multi-objective on
on the
the Kriging
optimization Kriging approximation
approximation
function was model,
model,
built by considering the optimization
the the optimization
multi-source mathematical
mathematical
uncertainties in
model
model different
was stages. The
wasconstructed
constructed and50 validated
and sample points
validated bywere
by 10 obtained
10 other
other from the
sample
sample LatinTo
points.
points. hypercube
To obtain sampling
obtainthe theoptimalmethod.
optimal result,
result,the
the
NSGA-II
NSGA-II Based
was on the
wasutilized Kriging approximation
utilizedtotodetermine
determinethe model,
thePareto the optimization
Pareto solution
solution sets.
sets. mathematical model was constructed
and validated by 10 other sample points. To obtain the optimal result, the NSGA-II was utilized to
determine the Pareto solution sets.
2. 2. Cracks
Cracks ininLocal
LocalPositions
Positions
2. Cracks in Local Positions
This
This truckequipped
truck equippedwith withmoremore than
than 200
200 tons
tons of of minerals
minerals usually
usuallyruns runson onthe thedownhill
downhilland and
turning road
Thissurface at
truck equipped the speed
with of
more 35
than km/h,
200 tonsshown in
of minerals Figure 1.
usually runs
turning road surface at the speed of 35 km/h, shown in Figure 1. As main load-bearing part, the As
on main load-bearing
the downhill and turningpart, the
welded road surface
A-type at the
frame speed of 35system
km/h, ofshown in Figure 1. mining
As main truck
load-bearing part, the welded
welded A-type frame ininsteering
steering system of heavy
heavy off-road
off-road mining truckhas hastotowithstand
A-type frame in steering system of heavy off-road mining truck has to withstand huge force from
withstandhuge huge force
force
from connection
from connection joints in this case. However, the stress concentration is usually generated
generatedinin
connection joints
joints inin thisthis
case.case. However,
However, the stressthe stress concentration
concentration is usually
is usually generated in structural
structural weakness regions, especially in geometric transition. The crack then appears along the
structural weakness
weakness regions,regions,
especially especially in transition.
in geometric geometric Thetransition. The crack
crack then appears then
along the weldappears
toe underalong the
weld toe under cyclic loads for some trucks, seen in Figure 2. The fatigue performance of this frame
weld toecyclic
under loads
cyclicfor some
loads trucks, seen intrucks,
for some Figure 2. The in
seen fatigue performance
Figure of this frame
2. The fatigue is one of critical
performance of this frame
is one of critical factors to ensurereliability
truck’s running reliabilityanti-fatigue
and safety. Therefore, anti-fatigue
is one offactors to ensure
critical factorstruck’s
to running
ensure truck’s and safety. Therefore,
running reliability and safety. optimization design
Therefore, for
anti-fatigue
optimization design
the welded A-typeforframe
the welded A-type
is conducted frame is
to improve itsconducted to improve its service lifetime.
service lifetime.
optimization design for the welded A-type frame is conducted to improve its service lifetime.

Figure 1. Practical mine pavement.

Figure 1. Practical mine pavement.
Figure 1. Practical mine pavement.

Figure 2. Crack in welded A-type frame.

Figure 2. Crack in welded A-type frame.

3. Experimental Work Figure 2. Crack in welded A-type frame.

3. 3.1.
Experimental Work
Specimen and Elements
The material
3.1. Specimen used in this frame is one kind of high strength low alloy steel under quenched and
and Elements
tempered conditions. The elements of both base metal and the weld seam were listed in Table 1.
 Appl. Sci. 2019, 9, 918 4 of 19

3. Experimental Work

3.1. Specimen and Elements

The material used in this frame is one kind of high strength low alloy steel under quenched and
tempered conditions. The elements of both base metal and the weld seam were listed in Table 1.

Table 1. Chemical elements of base metal and weldment (%).

Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 19
Material Fe Mn S Cr Ni Si Ti C
Base metal
Table 96.19 elements
1. Chemical 2.11 0.30of base
0.25 metal
0.30and0.46 0.19 (%).
weldment Bal.
Weldment 96.15 1.61 0.18 0.39 0.62 0.89 0.05 Bal.
Material Fe Mn S Cr Ni Si Ti C
Basedimension
The geometry metal 96.19 2.11was0.30
of specimen 0.25like0.30
designed 0.46 0.19
a flat dog-bone; Bal. was 2.5 mm
its thickness
Weldment
and 6 mm for the base metal 96.15
and the 1.61 0.18 respectively,
weldment, 0.39 0.62as shown0.89 in
0.05
FigureBal.
3. This butt joint
processed by gas metal arc-welding was used to obtain the welded frame’s material characteristics.

(a)

(b)
Figure 3.Figure 3. Specimen
Specimen geometry
geometry dimension:(a)
dimension: (a)base
base metal;
metal;(b)(b)weldment; (all dimensions
weldment; in mm, in mm,
(all dimensions
roughness in micro meters).
roughness in micro meters).
3.2. Mechanical Properties
3.2. Mechanical Properties
The installation diagram of monotonic tensile test is shown in Figure 4. The monotonic tensile tests
Thewere conducteddiagram
installation on both base metal and weldment
of monotonic tensilespecimens at the rate
test is shown of 0.025 4.
in Figure mm/sec, at normal tensile
The monotonic
room temperature and humidity. Both engineering and actual stress/strain curves for the base metal
tests were conducted on both base metal and weldment specimens at the rate of 0.025 mm/sec, at
and the weld metal are shown in Figures 4 and 5. The mechanical properties under monotonic tensile
normal room
tests aretemperature
listed in Tableand humidity.
2. Their standard Both engineering
deviations andtheactual
signified that stress/strain
repeatability curves
of the tests was for the
base metal and the
acceptable. weld
It was metal
clearly are the
seen that shown in Figures
yield strength of the 4base
and 5. was
metal Theclose
mechanical
to the weldproperties
metal and under
their difference of ultimate tensile strength (UTS) was also small.
monotonic tensile tests are listed in Table 2. Their standard deviations signified that the repeatability
of the tests was acceptable. It was clearly seen that the yield strength of the base metal was close to
the weld metal and their difference of ultimate tensile strength (UTS) was also small.
sts were conducted on both base metal and weldment specimens at the rate of 0.025 mm/sec, at
rmal room temperature and humidity. Both engineering and actual stress/strain curves for the
se metal and the weld metal are shown in Figures 4 and 5. The mechanical properties under
onotonic tensile tests are listed in Table 2. Their standard deviations signified that the repeatability
the tests was acceptable. It was
Appl. Sci. 2019, clearly seen that the yield strength of the base metal was close to
9, 918 5 of 19
Appl. Sci. 2019,
e weld metal and their difference of 9, x FOR PEER REVIEW
ultimate tensile strength (UTS) was also small. 5 of 19
Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 19

800 Engineering
800 Engineering
True
700
True
700
600

Stress σ (M Pa)
600

Stress σ (M Pa)
500
500
400
400
300
300
200
200
100
100
0
0.00 0.05 0.10 0.15
0 0.20 0.25 0.30
0.00 ε (mm/mm)
Strain 0.05 0.10 0.15 0.20 0.25 0.30
(b) Strain ε (mm/mm)
(a) (b)
Figure 4. (a) Installation diagram of monotonic tensile test; (b) Stress and strain curve of base metal
Figurediagram
Figure 4. (a) Installation 4. (a) Installation diagram
of monotonic of test;
tensile monotonic tensile
(b) Stress andtest; (b)curve
strain Stressofand strain
base metalcurve of base metal
under monotonic tensile test.
under monotonic tensile test. under monotonic tensile test.

700 Engineering
700 True Engineering
600 True
600
Stress σ (MPa)

Stress σ (MPa)

500
500
400
400
300
300
200
200
100
100
0
0.00 0.05 0
0.10 0.15 0.20 0.25
Strain
0.00 ε (mm/mm)
0.05 0.10 0.15 0.20 0.25
Strain ε (mm/mm)
Figure5.5.Stress
Figure Stressand
andstrain
straincurve
curveof
ofweldment
weldmentunder
undermonotonic
monotonictensile
tensiletest.
test.
Figure 5. Stress and strain curve of weldment under monotonic tensile test.
Table 2. Mechanical properties under monotonic tensile test.
Table 2. Mechanical properties under monotonic tensile test.
Table 2. Mechanical properties under monotonic tensile test.
Base Metal Weldment
Base Metal Weldment
Young’sYoung’s
modulus modulus
E (GPa)E (GPa) 200.4
200.4 210.1 Base
210.1Metal Weldment
Young’s modulus E (GPa) 200.4 210.1
Yield Strength
Yield Strength σ s (MPa)
σs (MPa) 451
451 456 456
UTS σb (MPa) Yield Strength
621.5σ s (MPa) 451
615.9 456
UTS σ (MPa) 621.5 615.9
Elongation e f b(%) UTS σ b (MPa)
27 18
621.5 615.9
Elongation e f (%) 27
True fracture stress σt (MPa) Elongation 57 e (%) 18 24
27 18
stresset σ(%)
f
True fracture strain
True fracture t (MPa)
0.336
57 24 0.279
True fracture stress σ t (MPa) 57 24
True fracture strain et (%) 0.336 0.279
True fracture strain et (%) 0.336 0.279
3.3. Cyclic Properties of a Welded Joint
3.3. Cyclic Properties
The crack of a of
initiation Welded
weldedJoint
A-type frameJoint
started from the weld toe, as shown in Figure 2, so it
3.3. Cyclic Properties of a Welded
was necessary to figure out the cyclic characteristics
The crack initiation of welded A-type frame started of thefrom
weldthe metal.
weldFully
toe, asreversed
shown cyclic loading
in Figure 2, so
was applied on theto The crack
flatfigure
butt joint initiation of welded A-type frame started from the weld from as
toe, shown
to in Figure 2, so
it was necessary out under straincharacteristics
the cyclic control, and the of strain
the weld amplitude
metal. changed
Fully reversed 0.1%
cyclic
0.45%. it was necessary to figure out the cyclic characteristics of the weld metal. drop reversed cyclic
Fully
loadingThewasgage length
applied onofthe
theflat
extensometer was 10strain
butt joint under mm. The fatigue
control, andfailure criterion
the strain was 50%
amplitude changed
of load. Thetostrainloading
was was applied
measured by anon Epsilon
the flat uniaxial
butt jointextensometer
under strain control, and the strain amplitude changed
from 0.1% 0.45%. The gage length of the extensometer was 10 mm. to Theobtain cyclic
fatigue mechanical
failure criterion
responses. For the from 0.1%
low strain to 0.45%.
amplitude The gage length
testing, the of the extensometer was 10 mm. The fatigue
first failure criterion
was 50% drop of load. The strain was measured by70anMPa cyclic
Epsilon loadingextensometer
uniaxial under load control was
to obtain cyclic
performed was 50% drop of load. The strain was measured by an Epsilon uniaxial extensometer to obtain cyclic
mechanicalonresponses.
the specimen For for
the 0.1
lowmillion
strain cycles to make
amplitude the weld
testing, the 70 metal
MPastable
cyclicand avoidunder
loading a possible
load
buckling mechanical
phenomenon. responses.
For on
thethe
high For the
strain low strain
amplitude amplitude
testing, thetolowtesting,
loading thefrequency
70 MPa cyclic loading
with andno under load
control was first performed specimen for 0.1 million cycles make the weld metal stable
more control was first performed on the specimen for 0.1 million cycles to make the weld metal stable and
avoidthan 1 Hz was
a possible requiredphenomenon.
buckling to keep the material
For thestable.
high The experimental
strain amplitude results
testing,between
the lowallloading
strain
amplitudes avoid
and number a possible buckling phenomenon. For the high strain amplitude testing, the low loading
frequency with no more of reversals
than 1 Hz was(2N) are shown
required in Figure
to keep 6.
the material stable. The experimental results
frequency with no more than 1 Hz was required to keep the material stable. The experimental results
between all strain amplitudes and number of reversals ( 2N ) are shown in Figure 6.
between all strain amplitudes and number of reversals ( 2N ) are shown in Figure 6.
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 6 of 19

Elastic test data

10-1 Plastic test data
Appl.
Appl. Sci. Sci. 2019,
2019, 9, 918PEER REVIEW
9, x FOR 6 of 619of 19

Strain amplitude εa (mm/mm)

Plastic fitting curve
Elastic fitting curve

y=-0.2615x-0.7984
Elastic test data
R2=0.9424
10-1 10-2 Plastic test data

Strain amplitude εa (mm/mm)

Plastic fitting curve
Elastic fitting curve
y=-0.1243x-2.2565
R2=0.9353
y=-0.2615x-0.7984
-2 10-3 R2=0.9424
10

100 1,000 10,000 100,000 1,000,000

y=-0.1243x-2.2565
Number of reversals Nf (2N,Cycles)
R2=0.9353

10-3
Figure 6. Relationship between strain amplitude and cycles under cyclic loading.

In order to acquire cyclic 100 1,000 the 10,000

properties, 100,000 1,000,000
Manson-Coffin equation was utilized to describe the
Number of reversals Nf (2N,Cycles)
fatigue characteristics:
Figure 6. Relationship between strain amplitude and cycles under cyclic loading.
Δε strain
Figure 6. Relationship between σf
Δε amplitude
e p '
and cycles under c cyclic loading.
ε a = + = ε 'f (2 N
(2 N f )b +equation f )
(1)
2
In order to acquire cyclic properties, 2
the E
Manson-Coffin was utilized to describe the
fatigue
In ordercharacteristics:
to acquire cyclic properties, the Manson-Coffin equation was utilized to describe the
∆ε p Δε σ f/ 2 and Δ0ε p / 2 c are elastic and plastic strain
e0
where ε a is total strain amplitude,
∆εe and
fatigue characteristics: εa = + b
= (2N f ) + ε f (2N f ) (1)
amplitude, respectively. The fatigue2 strength
2 coefficient
E and the fatigue ductility coefficient are
Δε andΔε∆εe /2 σ e p '
by σstrain
f and ε f , respectively. The fatigue
' '
f ∆ε /2
where ε a is total amplitude, and p
represented bare elastic
strength and plastic
exponent and thestrain amplitude,
fatigue ductility
respectively. The fatigue strength
ε a = + = (2 N
coefficient and the fatigue
+ ε 'f (2coefficient
f ) ductility N f )c are represented by σ0 (1)
exponent are represented by b 2and c ,2respectively. E f
and ε0Thef , respectively. The fatigue strength exponent
cyclic stress versus strain relationship and the fatigue ductility exponent are
for the weldp metal could be described by the represented
and Δε / 2 and Δε / 2 are elastic and plastic strain
e
whereby εb aandisc,total
Ramberg-Osgood strain
respectively. amplitude,
equation as:
The cyclic stress versus strain relationship for the weld metal could be described by the
amplitude, respectively. The fatigue strength coefficient and the fatigue ductility coefficient are
Ramberg-Osgood equation as: Δσ / 2 Δσ / 2 1/ n '
represented by σ f and ε f , respectively.
' ' ε a = The fatigue
+ ( strength )0 (2)
E
∆σ/2 ∆σ/2K ' 1/n exponent and the fatigue ductility
εa = +( ) (2)
exponent are represented by b and c , respectively. E K0
where Δσ / 2 is the stabilized stress amplitude, and K ' is the cyclic strength coefficient, 0and n '
The
wherecyclic ∆σ/2 stress versus strain
is the stabilized relationship
stress amplitude, andforK 0 is
thetheweld
cyclicmetal could
strength be described
coefficient, by the
and n is the
is the cyclic strain hardening exponent, respectively. The Ramberg-Osgood coefficients were
cyclic
Ramberg-Osgood strain hardening exponent, respectively. The Ramberg-Osgood coefficients were calculated by
calculated by equation as: with power equation to experimental plastic strain amplitude and
fitting a curve
fitting a curve with power equation to experimental plastic strain amplitude and stress, as shown in
stress, as shown in Figure 7. Those parameters Δσ /3.2are listed
Δσ /in2Table 3.
Figure 7. Those parameters are listed in Table 1/ n '
εa = +( ) (2)
E K'
Ramberg-Osgood fitting curve

where Δσ / 2 is the stabilized stress

500 Test data
450
amplitude, and K ' is the cyclic strength coefficient, and n '
is the cyclic strain hardening exponent,
400 respectively. The Ramberg-Osgood coefficients were
Stress σ (MPa)

calculated by fitting a curve with350power equation to experimental plastic strain amplitude and
300
stress, as shown in Figure 7. Those250
parameters are listed in Table 3.
200
150
100 Ramberg-Osgood fitting curve
500 Test data
50
450
0
400 0.000 0.005 0.010 0.015 0.020 0.025 0.030
Stress σ (MPa)

Strain ε (mm/mm)
350
3007. Fitting curve for Ramberg-Osgood coefficients.
Figure
Figure
250 7. Fitting curve for Ramberg-Osgood coefficients.
200 3. Fatigue parameters of the welded joint.
Table
To clarify the material
0 150 parameters mentioned
0 above, 0 the total0 strain amplitude was
σf (MPa) b εf c K n
100
decomposed to elastic and plastic parts, as shown in Figure 6. Then, the cyclic parameters were
calculated by fitting linear 50 −0.1243
1163.4 curves under 0.159 −0.2615
logarithmic 860.99 to 0.1792
coordinate be elastic and plastic strain
0
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Strain ε (mm/mm)

Figure 7. Fitting curve for Ramberg-Osgood coefficients.

To clarify the material parameters mentioned above, the total strain amplitude was
decomposed to elastic and plastic parts, as shown in Figure 6. Then, the cyclic parameters were
amplitude versus
Appl. Sci. 2019, 9, x FOR number of reversals. Those parameters are listed in Table 3. According to7 the
PEER REVIEW of 19
results, the plastic strain amplitude was much higher than the elastic strain amplitude for each strain
amplitude
level. versuscurve
The fitting number
could of reversals. Those parameters
perfectly describe the cyclicare listed inasTable
response, 3. According
displayed in Figure to the
8.
results,
According the plastic strain amplitude was much higher than the elastic strain amplitude
to cyclic stress and strain relationship shown in Figure 9, the second cycle stress for each strain
at 0.25%
Appl. Sci. 2019, 9, 918 7 of 19
level. The fitting curve could perfectly describe the cyclic response,
strain level was much higher than the half cycle, which implied that this material has as displayed in Figure
cyclic8.
According to cyclic stress and strain relationship shown in Figure 9, the second cycle stress at 0.25%
softening.
strain levelTo clarify
was the material
much parameters
higher mentioned
than the above, which
half cycle, the total implied
strain amplitude was material
that this decomposed has cyclic
to elastic and plastic parts, as shown in Figure 6. Then, the cyclic parameters were calculated by
softening. Table 3. Fatigue parameters of the welded joint.
fitting linear curves under logarithmic coordinate to be elastic and plastic strain amplitude versus
number of reversals. Those (MPa)
σ parameters
'
f b listed
are ε in Table c 3. According n ' results, the plastic strain
K ' tojoint.
the '
f
Table 3. Fatigue parameters of the welded
amplitude was much higher than
1163.4 the
−0.1243 0.159 −0.2615 860.99 0.1792level. The fitting curve
elastic strain amplitude for each strain
could perfectly describe the (MPa)response,
σ cyclic b as displayed
'
f
ε c FigureK8.' According
in n' to cyclic stress and
'
f

strain relationship shown in1163.4

Figure 9, −0.1243
the second cycle stress at 0.25%
0.159 −0.2615 860.99 0.1792 strain level was much higher
Test data
than the half cycle, which implied
0.035 that this material has cyclic softening.
Strain amplitude εa (mm/mm) Manson Coffin fitting curve

0.030
Test data
0.035 Manson Coffin fitting curve
0.025
Strain amplitude εa (mm/mm)

0.030
0.020
0.025
0.015
0.020
0.010
0.015
0.005
0.010
100 1,000 10,000 100,000 1,000,000
0.005 Number of reversals N (2N,Cycles)
f

100 1,000 10,000 100,000 1,000,000

Number of reversals N (2N,Cycles)
Figure 8. Manson-Coffin fitting curve.
f

Figure 8. Manson-Coffin fitting curve.

Figure 8. Manson-Coffin fitting curve.
Second cycle
600 Half cycle

400 Second cycle

600 Half cycle
200
Stress σ (MPa)

400
0
200
Stress σ (MPa)

-200
0
-400
-200
-600
-400
-0.003 -0.002 -0.001 0.000 0.001 0.002 0.003
Strain ε (mm/mm)
-600
-0.003
Figure 9. -0.002
Cyclic strain-0.001 0.000 at0.001
and stress 0.25% 0.002 0.003
strain level.
Strain ε (mm/mm)
Figure 9. Cyclic strain and stress at 0.25% strain level.
3.4. Microscopic Properties
In order
3.4. Microscopic to gain a deeper
Properties Figureunderstanding
9. Cyclic strainof thestress
and fatigue characteristics,
at 0.25% the microstructure for
strain level.
the base metal and the weld seam were both observed by transmission electron microscope (TEM),
3.4.In
asorder
shownto
Microscopic ingain a deeper
Figures understanding
10 and 11.
Properties ofthe
The grain size at thebase
fatigue
zone characteristics, the microstructure
was small, fine equiaxed grain. The big for the
base metal and the
grain located weld
at the heatseam were
affected both
zone wasobserved by transmission
stretched along electron
the thermal gradient microscope
direction (TEM), as
during the
In order
welding to gain
process to a deeper
become understanding
middle equiaxed of
grain. the
The fatigue characteristics,
temperature inconsistency
shown in Figures 10 and 11. The grain size at the base zone was small, fine equiaxed grain. The the microstructure
resulted in the local forbig
the
baselocated
grain metal
grain and theheat
coarsening.
at the weld seamfor
In addition,
affected were bothfatigue
the high
zone was observed
stretched bythe
regime,
along transmission
fracture
the electron
site had
thermal microscope
wave strips
gradient and loads
direction (TEM),
during theas
of
shown intiny steps caused
Figures by secondary extrusion under high frequency load, displayed in Figure 12. For the
welding process to 10 and 11.middle
become The grain size atgrain.
equiaxed the baseThezone was small,
temperature fine equiaxed
inconsistency grain. The
resulted in thebig
low fatigue regime, there were persistent slip bands, extrusion spines, and ditches under high strain
grain
local located
grain at the heat
coarsening. affected zone
In addition, washigh
for the stretched
fatiguealong the the
regime, thermal gradient
fracture direction
site had during
wave strips andthe
amplitude loading, which was formed by uneven slip around metal surface, displayed in Figure 13.
welding
loads of tinyprocess to become
steps caused middle equiaxed
by secondary extrusion grain.
under The
hightemperature
frequency load,inconsistency
displayedresulted
in Figure in12.
the
local
For thegrain
low coarsening.
fatigue regime, In addition, for the
there were high fatigue
persistent regime,extrusion
slip bands, the fracture site had
spines, and wave
ditches strips
underand
loads of tiny steps caused by secondary extrusion under high frequency
high strain amplitude loading, which was formed by uneven slip around metal surface, displayed inload, displayed in Figure 12.
For the
Figure 13.low fatigue regime, there were persistent slip bands, extrusion spines, and ditches under
high strain amplitude loading, which was formed by uneven slip around metal surface, displayed in
Figure 13.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 19

Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 19

Appl.2019,
Appl. Sci. Sci. 2019, 9, 918PEER REVIEW
9, x FOR 8 of 19
8 of 19

Figure 10. Microstructure at the base zone.

Figure 10. Microstructure at the base zone.
Figure 10. Microstructure at the base zone.
Figure 10. Microstructure at the base zone.

Figure 11. Microstructure at the heat affected zone.

Figure 11. Microstructure at the heat affected zone.
Figure 11. Microstructure at the heat affected zone.
Figure 11. Microstructure at the heat affected zone.

Figure 12. High-cycle fatigue fracture morphology.

Figure 12. High-cycle fatigue fracture morphology.
Figure 12. High-cycle fatigue fracture morphology.
Figure 12. High-cycle fatigue fracture morphology.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 9 of 19

Appl. Appl. Sci. 2019,

Sci. 2019, 9, 918PEER REVIEW
9, x FOR 9 of 19
9 of 19

Figure 13. Low-cycle fatigue fracture morphology.

4. Fatigue Life Estimation

Figure 13. Low-cycle fatigue fracture morphology.
Figure 13. Low-cycle fatigue fracture morphology.
4.1. Multibody Dynamics Analysis
4. Fatigue Life Estimation
To obtain
4. Fatigue force load history needed in fatigue life estimation of welded A-type frame, a
Life Estimation
4.1. Multibody
multi-body dynamic Dynamics
model Analysis
for the whole vehicle was constructed using the ADAMS software for
simulating real
To obtain
4.1. Multibody operating
force load
Dynamics conditions.
history
Analysis neededThe trucklife
in fatigue model and of
estimation the horizontal
welded A-type surface are shown in
frame, a multi-body
Figure 14. A model
dynamic complete description
for the whole vehicle of was
this constructed
modeling using can be thefound
ADAMS in software
Refs. [1,13,14]. To verify
for simulating real the
To
accuracy obtain
of this
operating force load
multi-body
conditions. history
dynamic
The truck needed
modelmodel, in
and the the fatigue life
acceleration
horizontal estimation
time
surface are of
history
shown welded
of wheel
in Figure A-type
14. A center frame,
complete at thea
multi-body
speeddescription
30 km/h ofdynamic model
this modeling
obtained for
from can the whole
thebesimulation vehicle
found in Refs. was
was[1,13,14].constructed
compared To verify
withthe using the
theaccuracy ADAMS displayedfor
software
of this multi-body
tested results, in
simulating
dynamic
Figure 15. The real operating
model, conditions.
the acceleration
acceleration The
time history
time history andtruck model
ofitswheel center
frequency and thespeed
atspectrum
the horizontal
30 km/hsurface
matched were are
obtained shown
from the in
well-matched
Figure
between 14.them.
simulationA complete
was description
compared
However, the of
with were
there thisresults,
tested
severalmodeling can
displayed
peaks in betested
thein found15.
Figure inThe
Refs. [1,13,14].
acceleration
acceleration Tohistory
time
frequency verify
spectrumthe
accuracy
and itsoffrequency
this multi-body
spectrum dynamic
matched were model, the
well-matchedacceleration
between time them. history
around 30 Hz. It could be caused by the external interferences or operational errors. The differences
However, of wheel
there werecenter
severalat the
speedpeaks in the tested
30 km/h acceleration frequency spectrum
was around 30 Hz. It could be caused by thedisplayed
external in
of the root meanobtained from the
square acceleration simulation
for the wheel compared with
center at different the speed
tested results,
between the test and
interferences
Figure 15. The or operational
acceleration errors.
time in The
history differences
and of the root mean square acceleration
its frequency spectrum matched were well-matched for the wheel
simulation
center were
at small,
different as shown
speed between Figure
the test 16. simulation
and were small, as shown in Figure 16.
between them. However, there were several peaks in the tested acceleration frequency spectrum
around 30 Hz. It could be caused by the external interferences or operational errors. The differences
of the root mean square acceleration for the wheel center at different speed between the test and
simulation were small, as shown in Figure 16.

Figure 14. Multi-body dynamics model.

Figure 14. Multi-body dynamics model.

Figure 14. Multi-body dynamics model.

 spectrum as (m/s ) Acceleration a (m/s )
2
2
1
Appl.
Appl. Sci.Sci. 9,9,x 918
2019,
2019, FOR PEER 10 of 1910 of 19
0 REVIEW
-1
-2

spectrum as (m/s ) Acceleration a (m/s )

2
02 1 2 3 4 5 6 7 8 9 10 11 12 13
160

Accleration frequency
1 Time t (s)

2
140
120 0
100
80 -1
60
40 -2
20 0 1 2 3 4 5 6 7 8 9 10 11 12 13
0160
frequency

Time t (s)
2

0
140 10 20 30 40 50 60
120 Frequency f (Hz)
100
80
60
Acceleration a (m/sAccleration

40 (a)
20
0
0 10 20 30 40 50 60
2
)

2.0 Frequency f (Hz)

1.5
1.0
0.5 (a)
0.0
-0.5
Acceleration a (m/s2)

-1.0
-1.5 2.0
-2.0 1.5
1.0
0
0.5 10 20 30 40
160 0.0
Accleration frequency
(m/s2)

140-0.5 Time t (s)

120-1.0
100-1.5
80-2.0
spectrum as

60 0 10 20 30 40
40160
Accleration frequency
spectrum as (m/s )

Time t (s)
2

20140
0120
100
0 10 20 30 40 50 60
80
60 Frequency f (Hz)
40
20
0
0 10 20 (b)30 40 50 60
Frequency f (Hz)
Figure 15. Acceleration time history and its frequency spectrum of wheel center at the speed 30
(b)(b) simulated results.
km/h: (a) tested results;
Figure
Figure15.15.Acceleration
Acceleration time history
time and its
history and frequency spectrum
its frequency of wheelofcenter
spectrum wheelat center
the speed 30 km/h:
at the speed 30
(a) tested results; (b) simulated results.
km/h: (a) tested results; (b) simulated results.
5.0
Root mean square acceleration aR (m/s2)

Test data
Simulation results
4.55.0
Root mean square acceleration aR (m/s2)

Test data
Simulation results
4.04.5

3.54.0

3.03.5

3.0
2.5
2.5
2.0
10 20 30 40 50
2.0 Velocity u (km/h)
10 20 30 40 50
Velocity u (km/h)
Figure 16. Root mean square acceleration of wheel center at different speed.
Figure 16. Root mean square acceleration of wheel center at different speed.
When the Figure
precision
16. of multi-body
Root dynamic
mean square model of
acceleration was guaranteed,
wheel center at the downhill
different and turning
speed.
road surface was built to simulate bad working conditions. The starting speed of the truck was defined
When the precision
as 30 km/h, of multi-body
and the turning radius wasdynamic
60 m, shownmodel was17.
in Figure guaranteed, thetime
The simulation downhill
was 16.5and
s. turning
When the precision of multi-body dynamic model was guaranteed, the downhill and turning
road surface
Then, thewas
forcebuilt to simulate
time history bad
of welded working
A-type frame atconditions. The
connection sites wasstarting
acquired speed of the truck was
from multi-body
road surface was built to simulate bad working conditions. The starting speed of the truck was
defined as 30
defined as km/h, and
30 km/h, andthethe
turning
turningradius
radiuswas
was 60 m, shown
60 m, shownininFigure
Figure17.17. The
The simulation
simulation timetime
was was
16.5 16.5
s. Then, thethe
s. Then, force time
force history
time historyofofwelded
welded A-type frameatatconnection
A-type frame connection sites
sites was was acquired
acquired fromfrom
multi-body
multi-bodydynamics analysis,
dynamics analysis,shown
shownininFigure
Figure 18, whichwas
18, which wasconsidered
considered asas cyclic
cyclic load
load required
required for for
fatigue life life
fatigue prediction.
prediction.
 Appl.
Appl.Sci.
Sci.2019,
2019,9,9,x918
FOR PEER REVIEW 1111ofof19
19

Appl. Sci. 2019, 9, x FOR PEER REVIEW 11 of 19

dynamics analysis, shown in Figure 18, which was considered as cyclic load required for fatigue
life prediction.

Figure 17. Multi-body dynamics model for downhill and turning condition.

Figure 17. Multi-body dynamics model for downhill and turning condition.
Figure 17. Multi-body dynamics model for downhill
2x10
and turning condition. 5

6x105
Y direction at connection between A type- frame
5x105 X direction at connection between A-type frame and frame
4x105 and front stabilizer bar 1x105
Force F (N)

3x105 Force F (N)

5 2x105
6x105 2x10 0
5 Y direction at connection between A type- frame
5x105 1x10 X direction at connection between A-type frame and frame
5 0 and front stabilizer bar 5
4x10 1x10
-1x105 -1x105
Force F (N)

3x105
Force F (N)

5
-2x10
2x105
-3x105 -2x1005
1x105 -4x105
0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
-1x105 Time t (s) -1x105 Time t (s)
-2x105
-3x105 -2x105
(a) (b)
-4x105
Figure
Figure
0 4Force
18.Force
2 18. time
6 time history
8 history
10 ofwelded
of
12 welded
14 A-typeframe:
16 A-type frame:(a)
(a)connection
connection
0 2 sitebetween
4site 6between
8 A-type
A-type
10 frame
12 frame
14 and
and
16
Time t (s) Time t (s)
frontstabilizer
front stabilizerbar;
bar; (b)connection
(b) connectionsite
sitebetween
betweenA-type
A-typeframe
frame and
and frame.
frame.

4.2.Elastic/Plastic
4.2. Elastic/PlasticAnalysis
Analysis
(a) and andVerification
Verification (b)
TheFE
The FE model
model of of welded
weldedof A-type
A-type frame
frame was generated
wasframe:
generated using
usingHYPERMESH
HYPERMESH software to estimate the
Figure 18. Force time history welded A-type (a) connection site betweensoftware to estimate
A-type frame and
lifetime
the lifetimeunder its force
under time history. TheTheframe structure made up ofupfront plates andand
backbackplates was
front stabilizer bar; its
(b)force time
connection history.
site between frame
A-typestructure
frame andmade frame. of front plates plates
meshed with 83,648 solid elements, including 81,907 C3D8 and 1741
was meshed with 83,648 solid elements, including 81,907 C3D8 and 1741 C3D6. The average length C3D6. The average length of the
meshed elements
of the meshedAnalysis was
elementsand 2 mm. The cracked
wasVerification region of the weld seam was endowed
2 mm. The cracked region of the weld seam was endowed with gradual with gradual meshing,
4.2. Elastic/Plastic
as shownasinshown
meshing, Figurein 19.Figure
The connection site between
19. The connection A-type frame
site between A-type and suspension
frame (marked(marked
and suspension as A5 in
Figure
The FE 19) was
model constrained
of welded with
A-type all degrees
frame of
was freedom.
generated The connection
using
as A5 in Figure 19) was constrained with all degrees of freedom. The connection sites between sites
HYPERMESH between A-type
software frame
to and
estimate
frame
the lifetime (marked
A-type under
frame and as A1
its force in
frametime Figure 19),
history.
(marked right
as A1 Theand left steering
frame 19),
in Figure structurecylinder
made
right and (marked
left up as
of front
steering A3 and A2 in
plates(marked
cylinder Figure
and back 19),
as A3plates
and
was meshed front
and A2 in stabilizer
with 83,648
Figure bar
19),solid (marked as
elements,
and front A4 in Figure
including
stabilizer 19) were applied
81,907asC3D8
bar (marked A4 inand with the
1741
Figure simulated
19)C3D6.
were The force
applied obtained
average
with thelength
from the multi-body dynamics analysis, respectively. The material properties for elastic/plastic finite
of the meshed elements was 2 mm. The cracked region of the weld seam was endowed with gradual
simulated force obtained from the multi-body dynamics analysis, respectively. The material
element analysis
properties for in were based onfinite
elastic/plastic the cyclic
elementcharacteristics,
analysis according
were basedto on the parameters
the and
cyclic listed in Table 3.
characteristics,
meshing, as shown Figure 19. The connection site between A-type frame suspension (marked
The Von-Mises
according to thestress contour
parameters of the
listed welded
in Table A-type
3. The frame under
Von-Mises the
stressdownhill
contour and
of turning
the road
welded surface
A-type
as A5 in Figure 19) was constrained with all degrees of freedom. The connection sites between
condition
frame under is shown
the in Figure
downhill and20. The maximum
turning road stress
surface was at the
condition bottom
is left
shown incorner
Figure between right front
20. The(marked
maximum
A-type frame
plate and and frame (marked as A1 in Figure 19), right and steering cylinder as A3
stress was back
at theplate,
bottomwhich was close
corner between to the actual
right frontfailure
plateposition.
and back The critical
plate, regions
which wasatclose
the bottom
to the
and A2 in
and top Figure
corner 19), and
between front
front stabilizer
plateregions
and back bar (marked
plate all had as A4
high in Figure 19) were applied with the
actual failure position. The critical at the bottom and topstress
corner levels
betweenas well.
front plate and back
simulated force obtained from the multi-body dynamics analysis, respectively. The material
plate all had high stress levels as well.
properties for elastic/plastic finite element analysis were based on the cyclic characteristics,
according to the parameters listed in Table 3. The Von-Mises stress contour of the welded A-type
frame under the downhill and turning road surface condition is shown in Figure 20. The maximum
stress was at the bottom corner between right front plate and back plate, which was close to the
actual failure position. The critical regions at the bottom and top corner between front plate and back
plate all had high stress levels as well.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 12 of 19
Appl. Sci. 2019, 9, 918 12 of 19
Appl. Sci. 2019, 9, x FOR PEER REVIEW 12 of 19

Figure 19. Finite element model of welded A-type frame.

Figure 19. Finite element model of welded A-type frame.
Figure 19. Finite element model of welded A-type frame.

Figure 20. Von-Mises stress contour of welded A-type frame, units in Pa.
Figure 20. Von-Mises stress contour of welded A-type frame, units in Pa.
To verify the finite element model, the strain measurements were conducted by using strain gauge
Figure 20. Von-Mises stress contour of welded A-type frame, units in Pa.
To rosettes
verify under the downhill
the finite element andmodel,
turning road
the surface
strain condition. The measuring
measurements were points were considered
conducted by using strain
as potential high stress ones, shown in Figure 21; the local strain gauge layout at measuring point A1-2
gaugeTo
rosettes under the downhillmodel,and turning roadmeasurements
surface condition. The measuring points
using were
isverify theinfinite
displayed Figureelement
22. Then, the tested the strain
principal stresses were comparedwere
with conducted
the simulatedby
results, strain
considered as
gauge rosettes
which werepotential
under high stress
the downhill
calculated ones, shown
and turning
from the measured in
road as
strain values Figure
surface 21; the local
condition.
the following strain gauge layout
The measuring points were
equation: at
measuring
consideredpoint A1-2 is displayed
as potential high stress in Figure 22. Then,
ones, shown inqthe
Figuretested21;principal
the localstresses
strain were
gaugecompared
layout at
E E 2 2
with the simulated results,
= which
measuring point A1-2 is 2displayed( were
◦ + ε calculated
◦) ± √ from · the
( measured
◦ − ◦ ) + ( ε strain
◦ − ε values
◦) as the following
(3)
(1 − µ) 0 in Figure 22. Then, the 0 tested principal 90stresses were compared
σ1,2 ε 90 ε ε 45 45
2(1 + µ )
equation:
with the simulated results, which were calculated from the measured strain values as the following
where σ1,2 are the principal stress, E and µ are Yong’s modulus and Poisson’s ratio, and ε 0◦ , ε 45◦ , ε 90◦
equation: E E
are the strain componentsσ1,2along ε0 +ε90directions
= the(three  )± (ε0strain
of⋅the  −ε  ) + (ε  −ε  )
2
rosette. 2
(3)
2(1−Eμ) Eμ)
2(1+ 45 45 90

σ1,2 = (ε  +ε  ) ± ⋅ (ε0 −ε45 ) +(ε45 −ε90 )

2 2
(3)
2(1−μ) 0 90 2(1+ μ)
where σ 1,2 are the principal stress, E and μ are Yong’s modulus and Poisson’s ratio, and ε 0 ,

45 , ε 90σ 1,2areare
εwhere  thethe principal
strain components E and
stress, along the μ aredirections
three Yong’s modulus rosette. ratio, and ε 0 ,
and Poisson’s
of the strain 

ε 45 The
, ε 90simulated
 are the strain
 components
principal stresses along
could the three
then directionsfrom
be obtained of thefinite
strainelement
rosette.analysis, which
were compared with principal
The simulated the experimental
stresses results, listedbeinobtained
could then Table 4. From
from the comparison,
finite it was clearly
element analysis, which
seen that the finite element analysis results agreed well with the tested ones. Most of the
were compared with the experimental results, listed in Table 4. From the comparison, it was clearly relative
errors werethe
seen that lessfinite
thanelement
10%, which is acceptable
analysis in engineering
results agreed structure
well with simulation.
the tested ones. Most of the relative
errors were less than 10%, which is acceptable in engineering structure simulation.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 13 of 19
Appl. Sci.
Appl. Sci. 2019, FOR9,PEER
9, x2019, 918 REVIEW 13 of 19 13 of 19

Figure 21. Schematic diagram of measuring points of welded A-type frame.

Figure 21. Schematic diagram of measuring points of welded A-type frame.
Figure 21. Schematic diagram of measuring points of welded A-type frame.

Figure 22. Layout of measuring point A1-2.

Figure 22. Layout of measuring point A1-2.
The simulated principal stresses could then be obtained from finite element analysis, which were
Figure 22. Layout of measuring point A1-2.
comparedTable
with the experimentalbetween
4. Comparison results, listed in Table
simulated and4. From
testedthe comparison,
principal stressit (MPa).
was clearly seen
that the finite element analysis results agreed well with the tested ones. Most of the relative errors
Table
were less than A1-1
4.
10%, which is A1-2
Comparison between
acceptable A2-1 A2-2and
simulated
in engineering A3-1simulation.
tested
structure A3-2 stress
principal A4-1(MPa). A4-2 A5
Simulated stress 101 117 79 98 333 301 75 81 109
Tested stress A1-1 A1-2
92 4. Comparison A2-1
105 between A2-2
71 simulated A3-1
92 and tested A3-2
316principal
285 A4-1
69 A4-2
78 A5
97
Table stress (MPa).
Simulated
Errorstress 101
9.8% 117
11.4% 79
11.3% 98
6.5% 333
5.4% 301
5.6% 75
8.7% 81
5.1% 109
12.4%
Tested stress 92 A1-1 105 A1-2 71
A2-1 92 A3-1316 A3-2 285A4-1 69A4-2
A2-2 78
A5 97
Error 9.8%
Simulated stress 101
11.4%
117
11.3%
79
6.5%
98
5.4% 301 5.6% 75 8.7%81
333
5.1%
109
12.4%
4.3. Fatigue LifeTested
Prediction
stress 92 105 71 92 316 285 69 78 97
4.3. Fatigue Error 9.8% 11.4% 11.3% 6.5% 5.4% 5.6% 8.7%
Life to the elastic/plastic finite element analysis under the simulated load, the stress and
Prediction 5.1% 12.4%
According
strainAccording
response tocould
4.3. Fatigue
be obtained finite
thePrediction
Life elastic/plastic to calculate
elementfatigue
analysislife of welded
under A-typeload,
the simulated frame.the The
stresscyclic
and
material properties
strain response were
couldtobe based on Table 3, and the linear cumulative damage rule was still used in
According theobtained to calculate
elastic/plastic finite element fatigue lifeunder
analysis of welded A-type
the simulated load,frame. The cyclic
the stress
this estimation.
material properties Taking
werethe based mean stress
onobtained
Table 3, into account,
and the fatiguethecumulative
linear Smith-Waston-Topper rule(SWT) correction
and strain response could be to calculate life of welded damage
A-type frame. wascyclic
The still used in
method was
this estimation. utilized in
Taking were
material properties the MSC
the meanbased onsoftware.
stress into
Table 3, Fatigue
andaccount, was considered to
the Smith-Waston-Topper
the linear cumulative have good
damage rule was still used precision
(SWT) in high
correction
in this
plastic
methodstrain field.Taking
estimation.
was utilized Then,
in the
thethemean
MSClifetime
stress contour
into
software. account,ofthe
Fatigue welded A-type frame
Smith-Waston-Topper
was considered wascorrection
to(SWT)
have acquired,
good as shown
method
precision in
in high
Figure
plastic 23.
wasThefield.
strain minimum
utilized
Then,fatigue
in the MSC life of contour
software.
the lifetime this welded
Fatigue was A-type A-type
considered
of welded frame was
to have good11,203
frame wascycles
precision in (about
high
acquired, as24
plastic cyclesina
shown
day) and strain field.
located Then, the
at the bottom lifetime contour
corner of welded A-type frame was acquired, as shown in Figure 23.
Figure 23.
TheThe minimum
minimum fatigue
fatigue life of this ofbetween
lifewelded
thisA-type frontA-type
welded plates frame
and back
frame was 11,203 wasplates.
cycles 11,20324
(about
Visual
cycles inspections
cycles (about
of thea
24 cycles
a day) and
welded
day) andA-type
located frame
at the revealed
bottom the crack
corner between(seenfront
in Figure
plates 2)
andat back
the right
plates.bottom
Visual weld toe after the
inspections
located at the bottom corner between front plates and back plates. Visual inspections of the welded of the
truck
welded ran for 13
A-type
A-type months
frame
frame atthe
revealed
revealed thecrack
downhill
the crack
(seen in and
(seenturning
Figure inatFigure
2) condition,
the right2)bottom and
at theweld thebottom
right
toe simulated
after weldresults
the truck toe
ran wasthe
forafter in
good
truck agreement
ran for 13 with months the attested ones. According
the downhill and turningto thecondition,
calculated and results,
the the fatigue results
simulated life of welded
was in
A-type frame has to be increased, because its current lifetime
good agreement with the tested ones. According to the calculated results, the fatigue life of welded was not in compliance with
engineering
A-type frame requirements.
has to be increased, because its current lifetime was not in compliance with
engineering requirements.
 Appl. Sci. 2019, 9, 918 14 of 19

13 months at the downhill and turning condition, and the simulated results was in good agreement
with the tested ones. According to the calculated results, the fatigue life of welded A-type frame has to
Appl. Sci. 2019, 9, x FOR PEER REVIEW 14 of 19
be increased, because its current lifetime was not in compliance with engineering requirements.

Figure 23. Fatigue life contour of welded A-type frame (cycles).

Figure
5. Lifetime and 23.Co-Optimization
Weight Fatigue life contour of welded A-type frame (cycles).

5.1. Multi-Objective Optimization Function

5. Lifetime and Weight Co-Optimization
The lifetime of an engineering structure is related to geometry dimension, material properties,
and bearing load. However,
5.1. Multi-Objective Optimization there is permissible deviation for the dimensions of structure, and the
Function
manufacturing process would also result in the uncertainty of geometry. The material properties
The lifetime of anbecause
could change engineering structure
of different formingisconditions
related to
andgeometry dimension,
artificial operating. The material properties,
dynamic loads
are bound to keep varying in service time. Therefore, those uncertainty
and bearing load. However, there is permissible deviation for the dimensions of structure, and the variables of geometry
dimensions, material properties, and bearing loads were all taken into account in lifetime and weight
manufacturing process would also result in the uncertainty of geometry. The material properties
co-optimization in this paper. The thickness of front and back plates in this frame would vary in
could change because
ranges, as shown of in
different
Figure 19,forming
as well asconditions and artificial
the cyclic material parameters operating. The
σ0f , ε0f , b and dynamic
c. The loads are
four main
bound tobearing
keep loads
varying
at fourin servicesites
connection time. Therefore,
(marked as A1, A2,those
A3, anduncertainty
A4 in Figure 19) variables of geometry
were also defined
dimensions,
to bematerial
variables. properties, and bearing
Then, the multi-objectives loads were
optimization all was
function taken into as:
described account in lifetime and
weight co-optimization in this paper. The thickness of front and back plates in this frame would vary
min Mass( x0, x1, x2)
in ranges, as shown in Figure max
19, as well
Li f e(as
x0,the cyclic
x1, x2, material
x3, x4, x5, x6, x7,parameters
x8, x9, x10) σ 'f , ε 'f , b and c . The
s.t. Li f e ≥ 2.0e5; mass ≤ 4.0e3kg; 0 ≤ x0 ≤ 100; 15 ≤ x1 ≤ 25; 24 ≤ x2 ≤ 40;
four main bearing loads at four connection sites (marked as A1, A2, A3, and A4 in Figure 19) were
872.6 ≤ x3 ≤ 1454.3; −0.1554 ≤ x4 ≤ −0.0932; 0.119 ≤ x5 ≤ 0.198;
also defined to be variables. Then, the multi-objectives optimization function was described as:
−0.3269 ≤ x6 ≤ −0.1961; −2.50e5 ≤ x7 ≤ −1.50e5; 7.88e4 ≤ x8 ≤ 1.31e5;
min Mass
7.95e4 ≤ x9(≤
x0, x1, x−2)
1.32e5; 5.01e5 ≤ x10 ≤ −3.01e5;

where Mass functionmax stoodLife for(the

x0,optimization
x1, x2, x3, x4, x5, x6, of
objective x7,weight
x8, x9,and
x10)
Li f e function stood for the
s.t. Life ≥ 2.0e5 ; mass ≤ 4.0e3kg ; 0 ≤ x0 ≤ 100 ; 15 ≤ x1 ≤ 25 ; 24
optimization objective of lifetime. The distance of weld metal away from the back≤plates
x2 ≤was40defined
; as
x0, marked as d in Figure 19. The weighting coefficient of two optimization objectives was assigned to
872.6 ≤ x0.73 in≤ 1454.3
be 0.3 and the light of; −engineering
0.1554 ≤experience.
x4 ≤ −0.0932 ; 0.119
The geometry ≤ x5 ≤x10.198
variables and x2;were thickness
− 0.3269 ≤ x6 ≤ −0.1961 ; −2.50e5 ≤ x7 ≤ −1.50e5 ; 7.88e4 ≤ x8 ≤ 1.31e5 ;
of front and back plates. The cyclic material parameters x3, x5 were the fatigue strength coefficient
and the fatigue ductility coefficient, and the parameters x4, x6 the fatigue strength exponent and the
7.95
fatigue ≤ x9 ≤exponent,
e4ductility ; −5.01e5 ≤The
1.32e5 respectively. x10 ≤ −3.01
bearing loads e5x7,; x8, x9, and x10 were the peak vertical
where Massforce function
along Y direction
stoodatfor connection site A1, the peak
the optimization lateral force
objective along X and
of weight directionLife
at connection
function site
stood for
A2–A4, respectively.
the optimization objective
In order to build ofthelifetime.
relationshipThe distance
between of weld
objectives metal away
and variables, the Latinfrom the back
hypercube plates was
sampling
x 0 , marked
defined asmethod was first usedas dto getin Figure 19. The
the 50 sample weighting
points, coefficient
listed in Table 5. of two optimization objectives
was assigned to be 0.3 and 0.7 in the light of engineering experience. The geometry variables x1
and x 2 were thickness of front and back plates. The cyclic material parameters x3 , x 5 were the
fatigue strength coefficient and the fatigue ductility coefficient, and the parameters x 4 , x 6 the
fatigue strength exponent and the fatigue ductility exponent, respectively. The bearing loads x 7 ,
x8 , x9 , and x10 were the peak vertical force along Y direction at connection site A1, the peak
 Appl.
Appl. Sci.Sci. 2019,
2019, 9, x FOR PEER REVIEW
9, 918 1515of of
19 19

Table 5. Sample points from the Latin hypercube method.

Table 5. Sample points from the Latin hypercube method.
Run Counter
x0 x1 x 2 x3 （MPa） x 4 x5 x6 x7 x8 x9 x10
5N）x8
Run x0（mm）
x1 （mm）
x2 （mm）
x3
x4 x5 x6 （×10
x7 （×105N）（×10x9 5N）x10
（×105N）
Counter (mm) (mm) (mm) (MPa) (×105 N) (×105 N) (×105 N) (×105 N)
1 0 18 27 1311.8 −0.09950.119−0.2788 −1.89 0.86 1.01 −3.83
1 2 0 1018 22
27 31
1311.8 −1098.1
0.0995 −0.1287−
0.119 0.169−0.3002
0.2788 −1.89 −1.96 0.86 1.02 1.010.88 −3.83−3.66
2 3 10 2022 31
16 1098.1
29 −1430.5
0.1287 −0.1313−
0.169 0.3002 −1.96
0.197−0.2228 −1.65 1.02 1.07 0.880.81 −3.66−5.01
3 20 16 29 1430.5 −0.1313 0.197 −0.2228 −1.65 1.07 0.81 −5.01
4 30 24 35 1003.1 −0.11350.168−0.2575 −1.85 1.18 1.14 −4.07
4 30 24 35 1003.1 −0.1135 0.168 −0.2575 −1.85 1.18 1.14 −4.07
5 5 40 4019 19
36 36
1193.1 1193.1
−0.1376 −0.1376
0.125 0.125−0.3242
−0.3242 −1.87 −1.87 0.99 0.99 0.95 0.95 −4.43
−4.43
6 6 50 5015 15
26 26
1169.3 −1169.3
0.1439 −0.1439−
0.132 0.132−0.2067
0.2067 −2.04 −2.04 1.11 1.11 1.311.31 −4.03−4.03
7 7 60 6021 21
37 37
1406.8 −1406.8
0.1414 −0.1414−
0.187 0.187−0.2362
0.2362 −1.81 −1.81 0.92 0.92 1.011.01 −4.28−4.28
8 8 70 7020 20
34 34
920.0 920.0 0.189
−0.0982 −0.0982−
0.189−0.2815
0.2815 −2.25 −2.25 1.22 1.22 1.031.03 −4.97−4.97
9 9 80 8017 35
17 1371.2
35 −1371.2
0.1351 −0.1351−
0.155 0.2468 −2.49
0.155−0.2468 −2.49 0.83 0.83 1.061.06 −3.29−3.29
... ... ... ... ... ... ... ... ... ... ...
50
… 100 23
…38
… …
1121.85 −0.1300
…
0.164
…
−0.2175
… −2.22
… 0.84
… 0.84
… −4.93
…
50 100 23 38 1121.85 −0.13000.164−0.2175 −2.22 0.84 0.84 −4.93

When
When thethe
optimization
optimization design
design variables were
variables weredetermined,
determined, thethe
corresponding
corresponding response
responsevalues
values
could
couldbebe
obtained
obtained from
fromthethe
finite element
finite elementanalysis. Then,
analysis. Then,their approximation
their approximation relationship
relationshipcould
couldbebe
constructed to start the multi-objectives optimization. In this paper, the Kriging approximation
constructed to start the multi-objectives optimization. In this paper, the Kriging approximation model
was chosen
model wastochosen
describe
totheir mathematical
describe links, while
their mathematical the NSGA-II
links, while thewas used to
NSGA-II search
was usedfor the optimal
to search for the
solution in this optimization problem. These two methods are described in my previous
optimal solution in this optimization problem. These two methods are described in my previous works [22,23].
The whole
works optimization
[22,23]. The wholeprocess is shownprocess
optimization in Figure
is 24.
shown in Figure 24.

Figure 24. Chart of optimization process.

5.2. Approximation Model Figure 24. Chart of optimization process.

The Kriging approximation model could take the place of the finite element model to describe
5.2. Approximation Model
the relationship between design variables and optimization objectives, which was explained in my
previous The Kriging approximation
publications [22,23]. In thismodel
paper,could take10
the other the place points
sample of the obtained
finite element modelhypercube
from Latin to describe
sampling method were used to testify the accuracy of the approximation model, shown in Figurein25.
the relationship between design variables and optimization objectives, which was explained my
If previous publications
the coefficient [22,23]. In
of determination R isthis paper,
closer the
to 1.0, other 10
it means the sample pointshas
fitting model obtained
the best from Latin
accuracy.
hypercube sampling method were used to testify the accuracy of the approximation model, shown
in Figure 25. If the coefficient of determination R is closer to 1.0, it means the fitting model has the
Appl. Sci. 2019, 9, 918 16 of 19
Appl. Sci. 2019, 9, x FOR PEER REVIEW 16 of 19

best
The accuracy.
Kriging The Kriging
approximation approximation
model model
for both weight andfor both met
lifetime weight and lifetime
the accuracy met the for
requirement accuracy
this
requirement
engineering for this engineering
optimization problem. optimization problem.

Lifetime from the finite element analysis Nc (cycles)

Weight from the finite element model mf (kg)

4,400 250,000
Simulated points
4,300 Approximation model fitting curve Simulated points
200,000 Approximation model fitting curve
4,200

4,100
R2=0.9856 150,000
4,000 R2=0.9708

3,900 100,000
3,800

3,700 50,000

3,600
0
3,600 3,700 3,800 3,900 4,000 4,100 4,200 4,300 4,400 0 50,000 100,000 150,000 200,000
Weight from the approximation model ma (kg) Lifetime from the approximation model Na (cycles)

(a) (b)
Figure Accuracy
25.25.
Figure of of
Accuracy approximation model:
approximation (a)(a)
model: weight of of
weight welded A-type
welded frame;
A-type (b)(b)
frame; fatigue lifelife
fatigue of of
welded A-type frame.
welded A-type frame.

5.3. Optimization Results

5.3. Optimization Results
In this paper, the NSGA-II was selected to conduct this multi-objectives optimization. The Pareto
In this paper, the NSGA-II was selected to conduct this multi-objectives optimization. The
frontiers were obtained after running 27 min and 2450 iterations, from which a compromised solution
Pareto frontiers were obtained after running 27 min and 2450 iterations, from which a compromised
was treated as the optimal result with comprehensive consideration of two conflicting optimization
solution was treated as the optimal result with comprehensive consideration of two conflicting
objectives, listed in Table 6. The distance of weld metal away from the back plates increased to 40 mm
optimization objectives, listed in Table 6. The distance of weld metal away from the back plates
and the thickness of front plates increased to 24mm as well, which could weaken the stress level at
increased to 40 mm and the thickness of front plates increased to 24mm as well, which could weaken
stress concentration areas, shown in Figure 26. The maximum stress from the optimized structure
the stress level at stress concentration areas, shown in Figure 26. The maximum stress from the
decreased to 285 MPa. According to the optimized results, it was beneficial to extend fatigue life of
optimized structure decreased to 285 MPa. According to the optimized results, it was beneficial to
this welded A-type frame when four absolute cyclic material parameters increased and four loads at
extend fatigue life of this welded A-type frame when four absolute cyclic material parameters
connection sites were decreased. It could be deduced that better welding quality would improve the
increased and four loads at connection sites were decreased. It could be deduced that better welding
fatigue performance of welded joints and lower speed could result in small load. As such, enhancing
quality would improve the fatigue performance of welded joints and lower speed could result in
the weldment quality and limiting drive speed for mining truck are able to ensure fatigue performances
small load. As such, enhancing the weldment quality and limiting drive speed for mining truck are
of welded A-type frame. As shown in Table 6, the reduction of the thickness of back plates mainly
able to ensure fatigue performances of welded A-type frame. As shown in Table 6, the reduction of
contributed to lightweight of this frame. The minimum fatigue life of welded A-type frame went
the thickness of back plates mainly contributed to lightweight of this frame. The minimum fatigue
up to 2.40 × 105 cycles, and its weight was reduced by 8.2%, which was also achieved by reducing
life of welded A-type frame went up to 2.40 × 105 cycles, and its weight was reduced by 8.2%, which
the thickness of back plates. The simulated fatigue life contour for the optimal solution is displayed
was also achieved by reducing the thickness of back plates. The simulated fatigue life contour for the
in Figure 27. The minimum fatigue life was close to the calculated one (around 2.33 × 105 cycles)
optimal solution is displayed in Figure 27. The minimum fatigue life was close to the calculated one
obtained from the approximation model and satisfied with the design requirement.
(around 2.33 × 105 cycles) obtained from the approximation model and satisfied with the design
requirement.
Table 6. Comparison of the initial and optimized results.

x0 x1 x2 x3 x7 x8 x9 x10 Lifetime Weight

x4 x5 x6
(mm) (mm) (mm) (MPa) (×105 N) (×105 N) (×105 N) (×105 N) (Cycles) (kg)
initial 0 20 32 1163.4 −0.1243 0.159 −0.2615 −2.03 1.05 1.06 4.01 11203 4135
optimized 40 24 27 1381.1 −0.1393 0.189 −0.2956 1.84 0.89 0.93 −3.32 240180 3797
 Table 6. Comparison of the initial and optimized results.
x 0 x1 x 2 （mm） x3 （MPa） x 4 x5 x 6 x 7 x8 x9 x10 LifetimeWeight
x 0 x1 x 2 （mm） x3 （MPa） x 4 x5 x 6 x 7
（mm） （mm） （×10 N） x8
（×10 N） x9N）（×10
（×10 x10N） (Cycles)
5 5 (kg)
LifetimeWeight 5 5

initial 0 20
（mm） （mm） 32 1163.4 −0.12430.159 −0.2615 −2.03 1.05 1.06 4.01
（×10 N）（×10 N）（×10 N）（×10 N） (Cycles)
11203
5 (kg)
4135
5 5 5

optimized
initial 40 24 27 1381.1 −0.13930.189
−0.12430.159−0.2956 1.84 0.89 0.93 −3.32 240180
11203 173797
Appl. Sci. 02019, 9, 20
918 32 1163.4 −0.2615 −2.03 1.05 1.06 4.01 4135
of 19
optimized 40 24 27 1381.1 −0.13930.189 −0.2956 1.84 0.89 0.93 −3.32 240180 3797

Figure 26. Stress contour of optimized welded A-type frame (MPa).

Figure
Figure 26. Stress
26. Stress contour
contour of optimizedwelded
of optimized weldedA-type
A-type frame
frame(MPa).
(MPa).

Figure 27. Fatigue life contour of optimized welded A-type frame (cycles).
Figure 27. Fatigue life contour of optimized welded A-type frame (cycles).
6. Conclusions
Figure 27. Fatigue life contour of optimized welded A-type frame (cycles).
6. ConclusionsThis paper firstly aimed to analyze the fatigue failure of welded A-type frame based on experiment
works and simulation method. The monotonic tensile properties for the weld metal and base metal
6. Conclusions
This
werepaper firstly and
both studied aimedfoundtotoanalyze the fatigue
be very similar. The fatiguefailure of the
test of welded
weldedA-typejoints wasframe based on
implemented
This
experiment paper
works firstly
and aimed
simulation to analyze
method. the
The fatigue
monotonic failure
tensileof welded
properties
to obtain cyclic properties used for the lifetime estimation. The multi-body dynamic model was A-type
for the frame
weld based
metalbuiltandon
experiment
base metal works
to getwere
load bothand
history simulation
studied
for method.
and found
the welded A-typeto be The monotonic
very
frame, similar.
and tensile
The
its accuracy properties
fatigue
was test offor
validated thethe
with theweld
welded metalwas
joints
experimental and
base metal
implemented were
results. The both
finite studied
to obtain and found
cyclic model
element properties toframe
of thisused be for
very
wasthesimilar.
lifetime
generated The fatiguethe
toestimation.
observe teststress
of the
The welded joints
multi-body
distribution, dynamic
and its was
accuracy
implemented was
to also
obtain verified
cyclic by comparing
properties the
used test
for results
the with
lifetime the simulated
estimation.
model was built to get load history for the welded A-type frame, and its accuracy was validated with stresses.
The The fatigue
multi-body life
dynamicof
model
the welded
was built
experimental A-type
to frame
get
results. load was
The predicted
history
finite for the
elementand found to
welded
model ofbe
A-type
thisfarframe
less than
frame, the
and
was itsdesign requirement,
accuracy
generated towas while
validated
observe its
with
the stress
predicted
the experimental
distribution, failure
and its position
results.
accuracy was
The finite in good
element
was also agreement
verifiedmodel with experimental
byofcomparing
this frame the observations.
wastest
generated
results to observe
with the stress
the simulated
Then, considering the multi-uncertainties in geometry dimension, material properties, and bearing
distribution,
stresses. and its life
The fatigue accuracy was also
of welded A-type verified
frameby wascomparing
predictedthe andtest
foundresults
to bewith
far the
lesssimulated
than the
load, a co-optimization method for increasing lifetime and lightweight was presented, based on the
stresses.
design The fatiguewhile
requirement, life ofitswelded
predictedA-type
failure frame was predicted
position was in good andagreement
found to be withfarexperimental
less than the
Kriging approximation model. The multi-objectives optimization function was built. The 50 sample
designpoints
requirement,
observations. while its predicted failure position was in good agreement with experimental
were obtained by Latin hypercube sampling method, and their responses were calculated by
observations.
Then, consideringfinite
the elastic/plastic the element
multi-uncertainties in geometry
analysis. The constructed dimension,
Kriging approximationmaterialmodelproperties,
was verified and
Then,
bearingwith
load,considering
a co-optimization
other the multi-uncertainties
10 sample points,method for increasing
and its accuracy in geometry
was foundlifetime dimension,
to beand lightweight
satisfactory. material properties,
was presented,
The NSGA-II was utilized and
based
bearing
on the to load,
Kriging a approximation
optimizeco-optimization
this multi-objective method
model. Theformulti-objectives
optimizationincreasing
problem. lifetime
The and lightweight
optimization
distance function
of weld waswas
metal presented,
awaybuilt.
fromThe based
the 50
on theback
sample Kriging
points
plates approximation
were obtained
and the model.
byofLatin
thickness The
front multi-objectives
hypercube
plates and four sampling optimization
absolute method, function
and
cyclic material their was built.
responses
parameters The
increased, were 50
while
sample points
calculated the
by thethickness
were of back
obtained by
elastic/plastic plates and
Latin
finite four loads
hypercube
element at connection
sampling
analysis. sites decreased.
method, and
The constructed Finally, the lifetime
their responses
Kriging approximation of
were
this
waswelded
calculated
model by theA-type
verified withframe otherincreased
elastic/plastic 10 finite
sample toelement
2.40e5
points,cycles
and and
analysis. itsThe
weight
its accuracy decreased
foundby
constructed
was to8.2%.
Kriging approximation
be satisfactory. The
modelAuthor
NSGA-IIwas
was verified with
utilized other
to optimize
Contributions: 10 sample points, and
this multi-objective
Conceptualization, C.M. and its accuracy was
optimization
F.B.; Methodology, found
problem.
C.M.; to
Thebedistance
Software, satisfactory.
W.L.; The
of weld
Validation,
NSGA-II
metal W.L.,was
away utilized
from
X.X. and to Formal
theZ.G.;
back optimize thisthe
platesanalysis,
and multi-objective
thickness
C.M.; optimization
of front
Investigation, plates
C.M.; andproblem.
fourZ.G.;
Resources, The
Datadistance
absolute ofX.X.;
weld
cyclic material
Curation,
Writing—Original Draft Preparation, C.M.; Writing—Review & Editing, F.B.; Visualization, H.J; Supervision, H.J.;
metal away
parameters from the back plates and the thickness of front plates and four absolute
increased, while the thickness of back plates and four loads at connection sites
Project Administration, C.M.; Funding Acquisition, C.M. and H.J.
cyclic material
parameters increased, while the thickness of back plates and four loads at connection sites
Appl. Sci. 2019, 9, 918 18 of 19

Funding: This research was funded by [Natural Science Foundation of Hunan Province] grant number [2017JJ3059
and 2018JJ4060], and [Postdoctoral Science Foundation] grant number [2017M622569], and [State Key Laboratory
of Advanced Design and Manufacturing for Vehicle Body] grant number [31715012].
Acknowledgments: The authors gratefully acknowledge the support of the Hunan Province Natural
Science Foundation (Grant No.:2017JJ3059 and 2018JJ4060), China Postdoctoral Science Foundation
(Grant No.:2017M622569), and Open Foundation of State Key Laboratory of Advanced Design and Manufacturing
for Vehicle Body (Grant No.:31715012).
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to
publish the results.

References
1. Gu, Z.; Mi, C.; Wang, Y.; Jang, J. A-type frame fatigue life estimation of a mining dump truck based on modal
stress recovery method. Eng. Fail. Anal. 2012, 26, 89–99. [CrossRef]
2. Wannenburg, J.; Heyns, P.S.; Raath, A.D. Application of a fatigue equivalent static load methodology for the
numerical durability assessment of heavy vehicle structures. Int. J. Fatigue 2009, 31, 1541–1549. [CrossRef]
3. Tovo, R.; Livieri, P. A numerical approach to fatigue assessment of spot weld joints. Fatigue Fract. Eng.
Mater. Struct. 2010, 34, 32–45. [CrossRef]
4. Kang, H.T. Fatigue prediction of spot welded joints using equivalent structural stress. Mater. Des. 2007, 28,
837–843. [CrossRef]
5. Varvani-Farahani, A.; Kodric, T.; Ghahramani, A. A method of fatigue prediction in notched and un-notched
components. J. Mater. Process. Technol. 2005, 169, 94–102. [CrossRef]
6. Wang, P.C.; Ewing, K.W. Fracture mechanics analysis of fatigue resistance of spot welded coach-peel joints.
Fatigue Fract. Eng. Mater. Struct. 1991, 14, 915–930. [CrossRef]
7. Ayala-Uraga, E.; Moan, T. Fatigue reliability based assessment of welded joints applying consistent fracture
mechanics formulations. Int. J. Fatigue 2007, 29, 444–456. [CrossRef]
8. Fischer, C.; Fricke, W.; Rizzo, C.M. Review of the fatigue strength of welded joints based on the notch stress
intensity factor and SED approaches. Int. J. Fatigue 2016, 84, 59–66. [CrossRef]
9. Jahed, H.; Varvani Farahani, A.; Noban, M.; Khalaji, I. An energy based fatigue life assessment model for
various metallic materials under proportional and non-proportional loading conditions. Int. J. Fatigue 2007,
29, 647–655. [CrossRef]
10. Jiang, C.; Liu, Z.C.; Wang, X.G.; Zhang, Z.; Long, X.Y. A structural stress-based critical plane method for
multi-axial fatigue life estimation in welded joints. Fatigue Fract. Eng. Mater. Struct. 2016, 39, 372–383.
[CrossRef]
11. Michael, K. Improvements in the fatigue assessment of large welded structure using the nominal stress
approach. SAE 2012, 1, 1910.
12. Shao, Y.; Liu, J.; Mechefske, C.K. Drive axle housing failure analysis of a mining dump truck based on the
load spectrum. Eng. Fail. Anal. 2011, 18, 1049–1057. [CrossRef]
13. Mi, C.; Gu, Z.; Yang, Q.; Nie, D. Frame fatigue life assessment of a mining dump truck based on finite
element method and multi-body dynamic analysis. Eng. Fail. Anal. 2012, 23, 18–26. [CrossRef]
14. Yang, Q.; Gu, Z.; Mi, C. An analysis on the fatigue life of frame in SF33900 mining dump truck. Autom. Eng.
2012, 23, 1015–1019.
15. Gu, Z.; Mi, C.; Ding, Z.; Zhang, Y.; Liu, S.; Nie, D. An energy-based fatigue life prediction of a mining truck
welded frame. J. Mech. Sci. Technol. 2016, 30, 3615–3624. [CrossRef]
16. Berto, F.; Campagnolo, A.; Chebat, F.; Cincera, M.; Santini, M. Fatigue strength of steel rollers with failure
occurring at the weld root based on the local strain energy values: Modelling and fatigue assessment.
Int. J. Fatigue 2016, 82, 643–657. [CrossRef]
17. Berto, F.; Vinogradov, A.; Filippi, S. Application of the strain energy density approach in comparing different
design solutions for improving the fatigue strength of load carrying shear welded joints. Int. J. Fatigue 2017,
101, 371–384. [CrossRef]
18. Fang, J.; Gao, Y.; Sun, G.; Xu, C.; Li, Q. Multi-objective robust design optimization of fatigue life for a truck
cab. Reliab. Eng. Syst. Saf. 2015, 135, 1–8. [CrossRef]
Appl. Sci. 2019, 9, 918 19 of 19

19. Aoues, Y.; Pagnacco, E.; Lemosse, D.; Khalij, L. Reliability-based design optimization applied to structures
submitted to random fatigue loads. Struct. Mulidisc. Optim. 2017, 55, 1471–1482. [CrossRef]
20. Hu, W.; Choi, K.K.; Cho, H. Reliability-based design optimization of wind turbine blades for fatigue life
under dynamic wind load uncertainty. Struct. Mulidisc. Optim. 2016, 54, 953–970. [CrossRef]
21. Gao, Y.; Liu, Q.; Wang, Y.; Zhao, W. Lightweight design with weld fatigue constraints for a three-axle bogie
frame using sequential approximation optimization method. Int. J. Veh. Des. 2017, 73, 3–19. [CrossRef]
22. Mi, C.; Gu, Z.; Zhang, Y.; Liu, S.; Zhang, S.; Nie, D. Frame weight and anti-fatigue co-optimization of a
mining dump truck based on Kriging approximation model. Eng. Fail. Anal. 2016, 66, 99–109. [CrossRef]
23. Mi, C.; Gu, Z.; Jian, H. Anti-fatigue and lightweight design for frame structures of electric wheel dump truck.
China Mech. Eng. 2017, 28, 2455–2462.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).