Research Article

Advances in Mechanical Engineering

2016, Vol. 8(4) 113
The Author(s) 2016
Fatigue properties of 6061-T6 DOI: 10.1177/1687814016643454
aime.sagepub.com
aluminum alloy butt joints processed
by vacuum brazing and tungsten inert
gas welding

Huei Lin1, Jiun-Ren Hwang1 and Chin-Ping Fung2

Abstract
Tungsten inert gas welding and vacuum brazing butt joints of AlMgSi alloy 6061 in the artificially aged condition T6
were studied. Constant amplitude and variable amplitude fatigue loading tests were performed. The experimental S-N
curves were compared with the fatigue design curves recommended by the International Institute of Welding, British
Standard, and Eurocode 9. Two mean stress correction methods, Goodman and Gerber, were evaluated. In terms of the
size effect on the fatigue life, this article proposed an innovational thickness correction method based on the ratio of the
ultimate tensile strengths of specimens with different thickness. For vacuum brazing components, the tensile strength
based thickness correction method was better than the thickness correction methods recommended by the
International Institute of Welding and Eurocode 9.

Keywords
Vacuum brazing, tungsten inert gas welding, aluminum alloy, fatigue, mean stress effect, size effect

Date received: 3 November 2015; accepted: 15 March 2016

Academic Editor: Filippo Berto

Introduction conditions.4 The frequently preferred process for weld-

Recently, environmental problems and concerns about
energy consumption have led to the increasing use of alu-
minum alloys in the automobile, aerospace, and other
industries. In particular, 6000 series aluminum alloys
have been studied extensively because they have better
strength, weldability, corrosion resistance, and cost less
than other aluminum alloys.1,2 AA 6061 is one of the
most versatile of the heat-treatable alloys and popular for
medium-to-high strength applications. It has a wide
range of applications such as in transportation compo-
nents, medical equipment, and national defense products.
Butt welding has numerous advantages such as good
connection efficiency, easy workmanship, and weight
reduction.3 The wide range of application of butt welds
in various structures provides researchers a large scope
for analyzing their behavior under various loading
ing aluminum alloy is tungsten inert gas (TIG) welding
owing to its relative simplicity and cost. However, this
process causes grains to coarsen in the fusion zone, dis-
tortion, hot cracking, and residual stresses. Vacuum
brazing (VB) can eliminate the disadvantages of TIG
welding. It is performed at 1:333101 to1:333 104 Pa,
and no oxidation problems arise. The flux-free braze
1
Department of Mechanical Engineering, National Central University,
Taoyuan City, Taiwan
2
Department of Mechanical Engineering, Oriental Institute of Technology,
New Taipei City, Taiwan
Corresponding author:
Jiun-Ren Hwang, Department of Mechanical Engineering, National
Central University, No. 300, Jhongda Rd., Jhongli District, Taoyuan City
32001, Taiwan.
Email: jrhwang@ncu.edu.tw

Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License
(http://www.creativecommons.org/licenses/by/3.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering
joints with high integrity and superior strength are
formed. VB improves the uniformity of temperature of
base metal and reduces residual stresses using a slow
heating and fast cooling cycle. It results in better
mechanical properties of the material. The process has
been used in various industries.
Fatigue is an important issue for welded structures
subjected to cyclic loading.5 However, welded joints
exhibit poor fatigue properties. In welding techniques,
failure is a key problem that is related to the stability
and safety of the welded structure.6 The fatigue proper-
ties of welded structures are function of welding mate-
rial, shape of the welded structure, radius of weld toe,
weld angle, height of the reinforcement, and welding
quality.710 With respect to the studies on the fatigue in
aluminum alloy weldments, most literature discussed
the fatigue behaviors of weldments prepared by the
molten welding methods, such as inert-gas metal-arc
welding,11 TIG welding,1214 and friction stir weld-
ing.1519 Few studies investigated the diffusion-bonding
weldments.
Several associations classify aluminum alloy weld-
ments into different categories based on joint geometry
and load type. Many related experiments have been
performed. The experimental data yield the stresslife
relationship of each category, and design specifications
are established. Research into fatigue culminated in
such design specifications as British Standard (BS)
8118:1991,20 Eurocode 9:2007,21 and International
Institute of Welding (IIW).22
Due to the complexity of the materials used and
local geometry of welded structures, inadequacies exist
when performing fatigue design according to the exist-
ing specifications of aluminum-welded structures. This
study investigates the fatigue behavior of 6061-T6 alu-
minum alloy butt welded by TIG welding and VB.
Figure 1. Configuration of single-V butt joint for TIG welding: (a) 6 mm and (b) 12 mm (all dimensions in mm).
Fatigue tests were performed using constant and vari-
able amplitude loadings. The results of fatigue tests
were compared with fatigue design specifications of
Eurocode 9, BS 8118, and IIW and could provide the
designer with the amendments for fatigue design
required in the proposal. The following two problems
were also considered in this study:
1. Differentiation between VB and TIG welding;
2. The influence of thickness and the variable
amplitude loading on the fatigue life.
Experimental methods
The material used in this research was AlMgSi alloy
(AA 6061-T6). Rolled aluminum alloy plates with
thickness of 6 and 12 mm were jointed, respectively.
TIG-welded joints were fabricated in a single-V butt
joint configuration as shown in Figure 1. The ER5356
filler rod was used in the TIG welding process. Highly
pure argon was the shielding gas. Multi-pass welding
was conducted to fabricate the joints. Figure 2 shows
the butt joint configuration of VB. The BAlSi-4 filler
and nitrogen cooling gas were used in the VB process.
The direction of welding was normal to the rolling
direction. All necessary care was taken to prevent join
distortion and the joints were made after the sheets
were clamped. The welded plates were milled to a thick-
ness of 4 or 10 mm, respectively.
Table 1 presents the chemical compositions of the
base metal and weld metals. Table 2 lists the tensile
properties of the base metal. The process parameters of
VB as shown in Table 3 were optimized in our previous
study.23
Optical microscope images were captured from the
cross sections of the weldments. After the specimens
Lin et al. 3

Figure 2. Configuration of butt joint for vacuum brazing: (a) 6 mm and (b) 12 mm (all dimensions in mm).

Table 1. Chemical compositions of the base metal and weld metals (wt.%).

Material Si Fe Cu Mn Mg Al

6061-T6 0.48 0.4 0.27 0.12 1.00 Balance

ER5356 0.25 0.1 0.05 4.5 Balance
BAlSi-4 12 0.8 0.3 0.15 0.1 Balance

Table 2. Tensile properties of the base metal. Table 3. Process parameters of vacuum brazing.

Material Yield strength Tensile strength Elongation Soak Soak Brazing Brazing
(MPa) (MPa) (%) temperature time temperature time
(C) (min) (C) (min)
6061-T6 259 306.7 17.3
590 50 600 30

were cut, they were cold-mounted and then mechani-
cally ground and polished. After polishing, the speci-
mens were etched using Kellers reagent. Deionized
water and ethanol were subsequently used to neutralize
the specimens. All samples were then cleaned for 20 s in
an ultrasonic bath using ethanol, dried, and placed in a
desiccator until they were analyzed using a microscope.
In order to characterize the welded joints, Vickers
hardness across the welded joint cross section was
obtained using a microhardness tester, with an indenta-
tion load of 0.05 kgf, according to the ASTM E384
standard. The microhardness was measured from the
center of the weld bead to both sides until microhard-
ness was stabilized. The interval of measurement was
0.1 mm for VB specimen and 1 mm for TIG welding
specimen.
Tensile and fatigue properties of the 6061-T6 plates
in the transverse direction were evaluated. The tests
were carried out according to ASTM E8 and ASTM
E466 standards, respectively. Configurations of the ten-
sile and fatigue specimens are shown in Figure 3.
The deformation of the tensile specimens was mea-
sured using an extensometer. The specimen finally
failed after necking, and the load was recorded as a
function of displacement. Axial fatigue tests were car-
ried out at room temperature with a servo-hydraulic
machine at a frequency of 10 Hz and a stress ratio of
R = 21. Before the fatigue tests were performed, the
specimen surfaces were polished in the loading direc-
tion using emery papers with a grade up to 1000.
In order to discuss the influence of thickness correc-
tion and mean stress, this study conducted variable
amplitude fatigue testing on weldments with thickness
of 4 and 10 mm. The transmission (TRN) history devel-
oped by the American Society of Automotive Engineers
(SAE) was adopted. This history contained predomi-
nantly tensile loadings along with significant compres-
sive loads. It has been normalized and has a maximum
value of 999. As most signals in the transmission history
have small fluctuations, they had little effect on the fati-
gue life. Accordingly, the signals with small amplitudes
4 Advances in Mechanical Engineering

Figure 3. Dimensions of fatigue and tensile specimens: (a) 4 mm (TIG welding), (b) 10 mm (TIG welding), (c) 4 mm (vacuum
brazing), and (d) 10 mm (vacuum brazing) (all dimensions in mm).

could be ignored to reduce significantly the length of
transmission history. The racetrack-counting method
could condense and reveal significant events in a com-
plex reversal history. For this procedure, a racetrack
width(s) must be defined first. Only reversal points at
which a racer would have to change from upward to
downward or vice versa are counted. In this study, the
racetrack width was set to be 40% of the greatest range
of the history. The reduced history had 130 reversal
points as shown in Figure 4. We adjusted the scale of
the condensed transmission loading course to the
desired level before each variable amplitude fatigue test.
The rainflow-counting method was adopted to count
the number of cycles of each stress range in the fatigue
Lin et al.
analysis. Dowling24 demonstrated that the rainflow-
counting method accurately identified closed hysteresis
loops in a variable amplitude histogram.
Miners rule had been developed to estimate the
accumulation of fatigue damage of metals subjected to
a cyclic loading with varying amplitudes. The failure
criterion is
X ni
=C 1 control. It can prevent local overheating and the forma-
Ni
where ni is the number of cycles at stress level Si, Ni is the
fatigue life in cycles at stress level Si, and C is the critical
damage value which is selected on a knowledge of the
load history or on a desired factor of safety. The C value
is usually assumed to be 1 for random and pseudo-
random histories. In this study, C was equal to 1.
Results and discussion
Microstructures
As TIG welding requires a large local heat input and
provides a low welding rate, it easily produces high
Figure 4. Condensed SAE transmission history.
Figure 5. Optical micrographs of TIG weldment: (a) weld bead and (b) heat-affected zone.
5
temperature in the weld bead and expands the heat-
affected zone (HAZ). Figure 5(a) and (b) shows the
microstructures of a TIG weld bead and the HAZ,
respectively. The grains grew in the 6061-T6 weld bead
and the HAZ. The precipitates in the HAZ gradually
concentrated to the grain boundary and grain center.
VB is a heating method that uses radiation and
allows uniform heating and automatic temperature
tion of a HAZ. In this method, the deformation of the
workpiece and the residual stress is minimal. Figure
6(a) and (b) displays the microstructures of a brazing
bead and the base metal, respectively, after VB. Unlike
TIG welding, VB resulted in no obvious variations
among brazing bead, HAZ, and base metal. In Figure
6(a), many inter-metallic compounds surround the dif-
fusion layer of the brazing bead after the filler metal
elements diffused toward the base metal. According to
Figure 6(b), since the temperature in the VB process
was high and maintained for over an hour, most of the
precipitates were solid dissolved. After quenching with
nitrogen gas, the solid solution formed a supersaturated
phase, and then, a small quantity of precipitates was
generated through natural aging.
Microhardness
The microhardness test was conducted at the welding
surface of the specimen. Figure 7(a) and (b) plots the
microhardness distribution achieved by TIG welding
and VB, respectively.
Figure 7(a) shows that under the influence of a local
high temperature in TIG welding, the area with the low-
est microhardness was the HAZ with a microhardness
value of HV50. The microhardness of the base metal
was HV100. The HAZ extended to approximately
22 mm to the left and right of the weld bead. The micro-
hardness distribution along the two sides of the weld
6 Advances in Mechanical Engineering

Figure 6. Optical micrographs of vacuum brazing specimen: (a) weld bead and (b) base metal.

Figure 7. Microhardness profile across the weld: (a) TIG weldment and (b) vacuum brazing specimen.

bead was asymmetric because the TIG welding speci-
men was prepared using a two-pass method. During the
second pass, the first weld bead was reheated, causing
annealing, thus reducing the microhardness.
In Figure 7(b), no clear HAZ was observed because
the weldment was heated to a uniform temperature in a
vacuum, so local overheating did not occur. The rapid
cooling and partial coarsening of precipitates reduced
the microhardness of the base metal from HV100 to
HV65, which was close to the microhardness of the
HAZ in TIG weldment.
Tensile properties
Table 4 presents the tensile properties of two types of
AA 6061-T6 weldments. Three specimens were tested
for each type of joint. The averages and standard devia-
tions of tensile properties are given in Table 4. Tensile
strength of the base metal was 306.7 MPa, and that of
the TIG welding joint was 174 MPa, which was 56.7%
of the strength of the base metal. The tensile strength
that was achieved by VB was 177 MPa, which was
57.7% of the strength of the base metal. This experi-
mental result indicated that 6061-T6 had the same ten-
sile strength in both VB and TIG welding, because it
was a heat-treatable aluminum alloy. When the VB
temperature elevated to 600C, and rapidly cooled
using nitrogen gas, precipitation hardening occurred,
improving the mechanical properties of the weldment.
For TIG welding, local high temperature molten
welding minimized the tensile strength of the HAZ,
causing necking there, and fracture surface was
Lin et al. 7

Table 4. Tensile properties of different AA 6061-T6 weldments.

Welding method Thickness (mm) Tensile strength (MPa) Elongation (%)

Mean Standard deviation Mean Standard deviation

TIG welding 4 174 2.9 9.2 2.7

TIG welding 10 168.8 6 23.3 4.8
Vacuum brazing 4 177 0.19 30.4 0.2
Vacuum brazing 10 195.5 0.8 23.7 1.9

TIG: tungsten inert gas.

Table 5. Parameters for the S-N curve of 4-mm-thick AA 6061-T6.

Welding method K m s Z95

TIG welding 5:5331020 6.53 0.17 2.93

Vacuum brazing 4:631029 10.87 0.1 3.08

TIG: tungsten inert gas.

Figure 8. Fracture surfaces for tensile specimens: (a) TIG welding and (b) vacuum brazing.

classified as ductile as shown in Figure 8(a). In VB, the
specimen fractures at the weld toe and the fracture sur-
face did not exhibit obvious necking, so the fracture
surface was classified as brittle as shown in Figure 8(b).
The tensile strength of the 10-mm-thick VB specimen
was 195.5 MPa, which exceeded that of the 4-mm-thick
VB specimen, because increasing the diffusion-bonding
surface improved the mechanical properties. However,
the tensile strength of the 10-mm-thick TIG welding
specimen was 168.8 MPa, which was less than that of
the 4-mm-thick TIG welding specimen, because more
welding passes yielded poorer mechanical properties.
Fatigue properties
To obtain S-N curves, seven VB specimens and eight
TIG welding specimens were tested. Based on the fati-
gue strength assessment recommended by the IIW,22
statistical analysis and processing of fatigue experimen-
tal data were performed.
When Nf \2 3 106 the S-N equation was expressed
as
Log Nf = Log K  m Log Sn f t  Zs 2
where Nf is the fatigue life in cycles, K is the empirical
constant, m is the slope coefficient of the S-N curve, Sn is
the nominal stress, f(t) is the thickness reduction factor,
Z is the number of standard deviations, and s is the stan-
dard deviation of Log Nf . The parameters for the S-N
curve of butt joints of AA 6061-T6 were listed in Table 5.
The fatigue design curve of aluminum butt joint was
classified as IIW FAT 45, BS 8118 class 42, and
Eurocode 9 category 56-7. For effective stress ratios,
based on consideration of both applied and residual
stresses, R \ 0.5 a fatigue enhancement factor f(R)
may be considered by multiplying the fatigue class of
classified details by f(R)
f R = 0:4 3 R + 0:9 for  1 R  0:25 3
8 Advances in Mechanical Engineering

Table 6. Safety factors for TIG welding and vacuum brazing.

N fDs IIW fDs BS 8118 fDs Eurocode 9

TIG welding
104 0.89 1.62 1.98
105 1.36 2.19 1.93
106 2.06 2.97 1.89
23106 2.33 3.25 1.87

N fDs IIW fDs BS 8818 fDs Eurocode 9

Vacuum brazing
104 0.63 1.14 1.39
105 1.1 1.77 1.56
106 1.91 2.76 1.75
23106 2.26 3.16 1.82
Figure 9. S-N curves of 4 mm thickness of AA 6061-T6.
TIG: tungsten inert gas; IIW: International Institute of Welding; BS:
British Standard.

where R is the stress ratio. When R = 21, f(R) = 1.3.

Figure 9 shows the S-N curves of AA 6061-T6 butt
joints processed by TIG welding and VB and the rec-
ommended design curves in three specifications. The
recommended design curves of IIW FAT 45 and
Eurocode 9 category 56-7 had been multiplied by 1.3 in
Figure 9. The stress amplitude in the fatigue experi-
ments was 40%85% of the tensile strength. The linear
regression line obtained from experimental data repre-
sents 50% survival probability. The experimental S-N
curve with 95% survival probability was obtained from
the standard deviation shift with Z95. In the higher life
region (N = 105 2 3 106 ), experimental S-N curves of
TIG weldments were higher than the recommended
design curves of BS 8118 class 42, Eurocode 9 category
56-7, and IIW FAT 45. As comparing with the experi-
mental results, these three specifications were conserva-
tive. In the lower life region (N = 104105), fatigue
design curve of IIW FAT 45 was not applicable to VB
butt joints. However, the experimental fatigue strength
of VB joint was still higher than those of the BS 8118
class 42 and Eurocode 9 category 56-7. Regarding to
the influence of welding process on fatigue strength,
TIG welding was superior to VB. According to Yahr,25
when aluminum alloy specimens were subjected to fati-
gue loading (R = 21), cyclic hardening caused fatigue
strength enhancement. Therefore, the fatigue strength
might be higher than the monotonic yield strength and
close to the ultimate strength of the material.
Safety factors were evaluated from experimental and
design curves
Dsexp Nr
fDs = 4 shows the fracture surface of VB specimen. The clea-
Dscode Nr
where Dsexp and Dscode are the stress ranges relative to
experimental results and that indicated in the fatigue
design curve for the same number of cycles to failure,
respectively.
Safety factors of TIG and VB specimens are listed in
Table 6 and also plotted in Figure 10(a) and (b). The
following observations were made:
1. Safety factor increased with increasing life dura-
tion except that obtained from Eurocode 9 in
TIG weldment.
2. Safety factors depended on the welding process.
They are higher for TIG weldment than for VB
specimen.
3. For short life duration, safety factors of TIG
weldment and VB specimen were below the
value of 2 indicated by IIW FAT 45, BS 8118
class 42, and Eurocode 9 category 56-7.
Since the HAZ was the area with the lowest micro-
hardness, the rupture of the TIG-welded fatigue speci-
men occurred in the HAZ, close to the weld bead.
Figure 11(a) shows the fracture surface of the TIG
weldment, which revealed a large number of smaller
dimples mixed with several larger dimples distributed
on the fracture surface. The fracture surface of the TIG
weldment showed a ductile fracture dominated by dim-
ples due to microvoid coalescence. The rupture of the
VB fatigue specimen occurred at the weld bead. Since
the cracks grew in the diffusion layer of the weld bead
and the bonding surface of the base metal, the thick-
ness of the diffusion layer strongly affected the crack
growth. Furthermore, if the welding temperature was
maintained for longer, then the diffusion of the weld
bead was more complete, and the diffusion layer of the
weld was less likely to contain defects. Figure 11(b)
vage planes were found on the left side of fracture sur-
face, and the dimples were found on the right side of
fracture surface. The fracture surface of VB specimen
showed a combination of both ductile and brittle
regions.
Lin et al. 9

Figure 10. Safety factors: (a) TIG welding and (b) vacuum brazing.

Figure 11. Fracture surfaces for fatigue specimens: (a) TIG welding and (b) vacuum brazing.

Mean stress effect sa sm

Goodman : + =1 5
Se Su
Many service load histories have a non-zero mean
 2
stress. In this study, the transmission history developed sa sm
by the SAE in the United States was adopted; the mean Gerber : + =1 6
Se Su
stress is tensile. Mean stress correction methods have
been developed to eliminate the burden of having to where sa is the stress amplitude, sm is the mean stress,
carry out fatigue tests at different mean stresses. Se is the endurance fatigue limit, and Su is the ultimate
Generally, there are many theories used to evaluate the tensile strength.
influence of the mean stress. In this article, we The predicted life and experimental life were com-
employed two theories that are commonly used to pared using the S-N curves from the experimental TIG
modify the prediction of fatigue life under fluctuating welding and VB data with a specimen thickness of
loading. Experience has shown that test data tend to 4 mm, following mean stress correction methods that
fall between the Goodman and Gerber curves. were carried out using equations (5) and (6), respec-
Goodman is usually a good choice for brittle materials tively. Figure 12(a) plots the predicted life and experi-
and not bounded when using negative mean stresses. mental life of 4-mm-thick weldments that were
Gerber is usually a good choice for ductile materials prepared by TIG welding. The predicted life made for
and bounded when using negative mean stresses the 4-mm-thick weldment using the Goodman mean
10
stress correction method agreed closely with experimen-
tal life. Figure 12(b) plots the predicted life and experi-
mental life of 4-mm-thick weldments that were
prepared using VB. Goodman mean stress correction
method also agreed closely with experimental life. As
TIG welding was a fusion-bonding method and VB
was a diffusion-bonding method, the two types of
bonding methods were different. Hence, the resulting
tensile and fatigue properties were different, which
affected the fatigue life prediction based on the variable
amplitude test. As found in Table 4, the elongation of
VB specimens was better than that of the TIG welding
specimens. In other words, TIG welding specimens
were more brittle than VB specimens. Therefore, using
the Goodman mean stress correction method to predict
the fatigue life of TIG welding specimens would be
more accurate.
As revealed by the above results, for the 4-mm-thick
TIG welding and VB specimens of aluminum alloy
6061-T6, under the transmission history, the Goodman
mean stress correction method could be used to predic-
tion life achieved using both welding.
Size effect
The fatigue strength of weldments usually decreases due
to geometry, residual stress, slag, porosity, surface
defect, crack, and HAZ. In the fatigue analysis of weld-
ment, S-N curves recommended by several standards or
codes are adopted. Thickness correction is also included
in some specifications such as IIW. In this study, we
consider the fatigue strength of 4-mm-thick specimen as
a reference and check the suitability of different thick-
ness correction methods. The fatigue strengths of
welded joints that fail from the weld toe are generally
believed to decrease as the plate thickness increases.26
Recent work has demonstrated that the thickness effect
depends on the overall proportions of the welded
joint.27,28 These effects are accounted for by the fatigue
rules in Eurocode 9. A further refinement in the IIW
recommendations modifies the thickness correction
exponent p for various welds. Values range from 0.3 to
0.1, reflecting the fact that the thickness correction also
depends on the degree of stress concentration that is
induced by the welded joint. The lower fatigue strength
of thicker members is considered by multiplying the
stress range of the structure by the thickness reduction
factor f (t)
 p
teff
f t = 7 predicted using the Goodman mean stress correction
tref
where the reference thickness tref is 4 mm. The thickness
correction exponent p is dependent on the effective
thickness teff and the joint category. In this study, the
effective thickness was 10 mm and the p value was 0.1.
Advances in Mechanical Engineering
When equation (7) is substituted into equation (2), the
S-N equation is expressed as
Sn
Log Nf = Log K  mLog  0:1  Zs 8
t10
t4
The IIW specification provided the above equation,
which was applicable to molten welding (TIG). VB was
not a molten welding (but was classified as a diffusion
bonding), and as the thickness increased, the tensile
strength increased accordingly, so the thickness correc-
tion equation of the IIW specification was not avail-
able. The fatigue strength of aluminum alloy was
directly proportional to its tensile strength, so the
authors proposed a tensile strengthbased thickness
correction method. The S-N equation could be
expressed as
Sn
Log Nf = Log K  m Log    Zs 9
UTS4
UTS10
where UTS10 is the ultimate strength (195.5 MPa) of 10-
mm-thick specimen, and UTS4 is the ultimate strength
(177 MPa) of 4-mm-thick specimen.
According to the experimental S-N curves of 4-mm-
thick TIG weldment and 4-mm-thick VB specimen, the
thickness corrections were carried out using equations
(8) and (9). After the mean stress correction was con-
ducted by re-applying equations (5) and (6), the pre-
dicted life and experimental life were compared. Figure
13(a) plots the predicted life and experimental life of
TIG weldment, with an applied thickness correction
(equation (8)) based on the data from the 4-mm-thick
S-N curve. For a 10-mm-thick weldment under the
Goodman mean stress correction method, the predicted
life agreed closely with the experimental life. Figure
13(b) plots the predicted life and experimental life of
AA 6061-T6, processed using VB, with an applied ten-
sile strengthbased thickness correction (equation (9))
based on data from the 10-mm-thick S-N curve. As
revealed by Figure 13(b), the life predicted using the
Goodman mean stress correction method with the ten-
sile strengthbased thickness correction was better than
others. For the 10-mm-thick weldment, the predicted
life agreed closely with experimental life under the
Goodman mean stress correction method. The above
results indicated that for the 10-mm-thick TIG welding
and VB specimens of aluminum alloy 6061-T6, under
the transmission history, the fatigue life could be well
method.
The 10-mm-thick S-N curves could be derived via
equations (8) and (9) and 4-mm-thick experimental
S-N curves. Figure 14 plots the S-N curve for TIG
welding that was obtained by applying the thickness
Lin et al. 11

Figure 12. Comparison between the predicted life and experimental life of 4-mm-thick weldments: (a) TIG welding and (b) vacuum
brazing.

Figure 13. Comparison between the predicted life and experimental life of 10-mm-thick weldments: (a) TIG welding specimen
with IIW thickness correction and (b) vacuum brazing specimen with and without tensile strengthbased thickness correction.

correction (equation (8)); the S-N curve for VB that

was obtained by applying the tensile strengthbased
thickness correction (equation (9)). Both curves were
compared with those recommended by three specifica-
tions. The S-N curves of both TIG weldment and VB
specimen following thickness correction were higher
than the design curves for the IIW FAT 45, BS 8118
class 42, and Eurocode 9 category 56-7.

Conclusion
This study investigated the mechanical properties of
AA 6061-T6 weldments processed by TIG welding and
Figure 14. S-N curves of 10 mm thickness with corrections. VB. Fatigue tests were conducted on TIG welding and
12
VB of butt-welded joints. The effects of mean stress and
specimen thickness on the fatigue life were presented.
The following conclusions were drawn:
1. For the TIG weldment, the area with the lowest
microhardness was the HAZ. For the VB speci-
men, no clear HAZ was observed.
2. TIG welding was a molten soldering method and
VB was a diffusion-bonding method. Difference
in mechanism between these two joining methods
leaded to different size effect on tensile strength.
The tensile strength of the VB specimen increased
with specimen thickness, whereas that of the TIG
welding specimen decreased.
3. TIG butt joints of AA 6061-T6 achieved higher
fatigue strength as compared to the fatigue
design curves of IIW FAT 45, BS 8118 class 42,
and Eurocode 9 category 56-7. In the lower life
region (N = 104105), it was found that the fati-
gue strength of VB joint was lower than that of
IIW FAT 45, but still higher than those of the
BS 8118 class 42 and Eurocode 9 category 56-7.
4. When a TIG weldment or VB joint of alumi-
num 6061-T6 was subjected to a variable ampli-
tude loading with tensile mean stress, Goodman
method was suitable to modify the mean stress
effect.
5. In terms of the size effect on the fatigue life, the
thickness correction method recommended by
the IIW with p = 0.1 was applicable to the TIG
butt joint of aluminum 6061-T6.
6. This article proposed an innovational thickness
correction method which was based on the ratio
of the ultimate tensile strengths of specimens with
different thickness. For VB components, the ten-
sile strengthbased thickness correction method
was better than the thickness correction methods
recommended by the IIW and Eurocode 9.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial sup-
port for the research, authorship, and/or publication of this
article: This work was supported by the Ministry of Science
and Technology, ROC under the contract no. NSC-101-2221-
E-008-011-MY3.
References
1. Lee SH, Saito Y, Sakai T, et al. Microstructures and
mechanical properties of 6061 aluminum alloy processed
Advances in Mechanical Engineering
by accumulative roll-bonding. Mat Sci Eng A: Struct
2002; 325: 228235.
2. Mansourinejad M and Mirzakhani B. Influence of
sequence of cold working and aging treatment on
mechanical behaviour of 6061 aluminum alloy. T Nonferr
Metal Soc 2012; 22: 20722079.
3. Kim MH, Kim HJ, Han JH, et al. Influence of backing
materials towards the fatigue strength of butt-welded
joints. Proc IMechE, Part C: J Mechanical Engineering
Science 2011; 225: 17981807.
4. Grubisic V. Service strength of welded aluminum struc-
tures influences and validation. Weld World 2007; 51:
116.
5. Fricke W. Recent developments and future challenges in
fatigue strength assessment of welded joints. Proc
IMechE, Part C: J Mechanical Engineering Science 2015;
229: 12241239.
6. Sun Y and Yang X. Study on the correction of S-N dis-
tribution in the welding fatigue analysis method based on
the battelle equivalent structural stress by rough set the-
ory. Stroj Vestn: J Mech E 2014; 60: 600606.
7. PD 6493:1991. Guidance on methods for assessing the
acceptability of flaws in fusion welded structures.
8. Bergmann JP and Herold S. Influence of processing con-
ditions on the mechanical properties of aluminium over-
lap joints: a case study. Weld World 2006; 50: 5564.
9. Gurney TR. Fatigue of welded structures. 2nd ed. Lon-
don: Cambridge University Press, 1979.
10. Drew VN. Fatigue considerations in welded structure.
SAE technical paper 1982-820695, 1982.
11. Zhou C, Yang X and Luan G. Fatigue properties of fric-
tion stir welds in Al 5083 alloy. Scripta Mater 2005; 53:
11871191.
12. Pinho da Cruz JAM, Costa JDM, Borrego LFP, et al.
Fatigue life prediction in AlMgSi1 lap joint weldments.
Int J Fatigue 2000; 22: 601610.
13. Ye N and Moan T. Improving fatigue life for aluminium
cruciform joints by weld toe grinding. Fatigue Fract Eng
M 2008; 31: 152163.
14. Radaj D, Sonsino CM and Fricke W. Recent develop-
ments in local concepts of fatigue assessment of welded
joints. Int J Fatigue 2009; 31: 211.
15. Ericsson M, Jin LZ and Sandstrom R. Fatigue properties
of friction stir overlap welds. Int J Fatigue 2007; 29:
5768.
16. Di S, Yang X, Luan G, et al. Comparative study on fati-
gue properties between AA2024-T4 friction stir welds and
base materials. Mat Sci Eng A: Struct 2006; 435: 389395.
17. Zhou C, Yang X and Luan G. Effect of root flaws on the
fatigue property of friction stir welds in 2024-T3 alumi-
num alloys. Mat Sci Eng A: Struct 2006; 418: 155160.
18. Xu X, Yang X, Zhou G, et al. Microstructures and fati-
gue properties of friction stir lap welds in aluminum alloy
AA6061-T6. Mater Design 2012; 35: 175183.
19. Di S, Yang X, Fang D, et al. The influence of zigzag-
curve defect on the fatigue properties of friction stir welds
in 7075-T6 Al alloy. Mater Chem Phys 2007; 104:
244248.
20. British Standard (BS) 8118:1991. Structural use of
aluminiumpart 1 code of practice for design.
Lin et al.
21. Eurocode 9:2007. Design of aluminium structurespart
1-3: structures susceptible to fatigue.
22. Hobbacher A. Recommendations for fatigue design of
welded joints and components (IIW doc. 1823-07, Welding
Research Council, Bulletin 520). New York: Welding
Research Council, 2009.
23. Lin H. Application of Taguchi method in the optimiza-
tion of vacuum brazing parameters. In: Proceedings of
the 30th national conference on mechanical engineering of
CSME, Taichung, Taiwan, 67 December 2013, paper
no. 1499. Taiwan: Chinese Society of Mechanical
Engineers.
24. Dowling NE. Fatigue failure predictions for complicated
stress-strain histories. J Mater 1972; 7: 7187.
13
25. Yahr GT. Fatigue design curves for 6061-T6 aluminum. J
Press Vess: T ASME 1997; 119: 211215.
26. Gurney TR. The influence of thickness on the fatigue
strength of welded joints. In: Proceedings of the 2nd inter-
national conference on behaviour of offshore structures,
London, 2831 August 1979. Cranfield: BHRA Fluid
Engineering.
27. Maddox SJ. The effect of plate thickness on the fatigue
strength of fillet welded joints. Cambridge: Abingdon,
1987.
28. Maddox SJ. Scale effect in fatigue of fillet welded alumi-
nium alloys. In: Proceedings of the 6th international con-
ference on aluminium weldments, Cleveland, OH, 35
April 1995. Miami: American Welding Society.