Professional Documents
Culture Documents
A R T I C LE I N FO A B S T R A C T
Keywords: This paper investigates the effect of laser welding parameters, such as beam power, welding speed, incident
Laser welding beam angle, incident beam position and beam diameter, on the weld geometry, microstructure, porosity and
Welding parameters mechanical properties of successive double-sided laser beam welded AA2024-AA7075 T-joints using 4047 filler
Dissimilar T-joints wire. A change in the welding parameters influences the weld geometry and porosity, but does not cause sig-
Defects
nificant variations in the weld microstructures, though some liquation cracking was observed in the heat-af-
Mechanical properties
fected zone of alloys AA7075 and AA2024. The macroporosity occurs more in the second weld seam than in the
first one. The pull-out test results presented higher values than those obtained by other authors. The ultimate
tensile load in pull-out test is influenced by the laser power, laser beam diameter and incident beam position.
Macroporosity plays a relevant role in fracture initiation during pull-out tests. Porosity and liquation cracking
influenced the fracture mode of the pull-out test specimens, but they do not significantly affect the results.
1. Introduction plate. According to them, one of the reasons for this is the fluctuation in
wire feeding rate and/or position during the welding process which
Some aircraft manufacturers are exploring the possibility of repla- leads to a fluctuation in laser energy that might cause keyhole in-
cing riveting by LBW as joining technology in manufacturing of alu- stability. Moreover, the gap between the skin and stringer plates that is
minum alloy stiffened panels in lower fuselage [1]. The goal is to not covered by the weld bead affects the keyhole instability leading to
achieve weight reduction and an increase in strength of the airplane’s macropore formation in an autogenous single-sided laser welded T-joint
structure as well as time and cost reduction in aircraft production [2]. [10]. For a single-sided laser welded T-joint, Ventzke et al. [11] verified
The aeronautical industry considers both, simultaneous and successive that the macropores were formed mainly on the rear side of the T-joint.
double-sided LBW, as acceptable methods for joining skin-stringer T- This occurred because the solidification rate is higher in this region
joints, since they can produce defect-free and symmetrical weld seams than on the side where the laser irradiation occurs. Thus, the gas bubble
[3]. formed due to the collapse of the keyhole does not have enough time to
Porosity is a common defect in aluminum alloys welded by fusion escape from the molten pool, it therefore becomes entrapped as a
techniques. The main reason for this is the rejection of hydrogen during macropore in the weld seam.
weld pool solidification due to the much higher hydrogen solubility in The porosity is significantly affected by the welding parameters.
molten aluminum than in solid aluminum [4]. Furthermore, the highest Thus, Tao et al. [1] stress that the amount of porosity increases with the
cooling rate of LBW is harmful for weld degassing [5]; however, it also wire feeding angle and it is higher when the filler metal is added in the
reduces the nucleation and growth rates of hydrogen pores [6], gen- trailing direction than in the leading direction. The reduction of the
erating microporosity. Micropores might be formed by vaporization of incident beam angle (from 25° to 11°), that is the angle formed by the
alloying elements [7] as well. Another reason for porosity formation in laser beam and the skin surface, also reduces the porosity significantly
LBW is keyhole instability [6]. This type of porosity is characterized by [11]. Other authors report a narrower range (10° to 15°) of this angle to
its large size, irregular shape and rough walls [8]. For bead-on-plate obtain welds with reduced porosity in an autogenous one-sided T-joint
weldments, Yu et al. [9] observed that susceptibility to macroporosity is of AA6013-T4 [10]. Leo et al. [7] state that the microporosity caused by
higher in laser welding with filler wire than in autogenous laser magnesium vaporization increases with the laser power in butt joints of
welding when the keyhole does not penetrate the full thickness of the AA5754-H111 welded by arc assisted fiber laser with ER5356 as filler
⁎
Corresponding author at: CEMMPRE, Department of Mechanical Engineering, University of Coimbra, Portugal.
E-mail address: altino.loureiro@dem.uc.pt (A. Loureiro).
https://doi.org/10.1016/j.jmapro.2018.08.003
Received 25 May 2018; Received in revised form 16 July 2018; Accepted 1 August 2018
1526-6125/ © 2018 Published by Elsevier Ltd on behalf of The Society of Manufacturing Engineers.
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
wire. On the other hand, the porosity due to collapse of keyhole in butt Al-Mg-Si or Al-Cu-Li series [15] and special alloys which have their
joints of AA5754-H111 welded by arc assisted fiber laser reduces with weldability improved by modification of their chemical composition
the increase in laser power since the increase in vaporization of mag- [25]. On the other hand, few works have studied dissimilar laser
nesium raises the metal vapor pressure in the keyhole [12]. Moreover, welded T-joints of 2xxx and 7xxx series but focus on single-sided LBW
Ola and Doern [13] mention the use of Ar as being more effective in [3] or on autogenous successive double-sided LBW [26].
minimizing porosity than He in the welding of AA7075-T651. Fur- Thus, the main objective of this study is to investigate the effect of
thermore, increasing the beam diameter is beneficial for degassing the laser welding parameters, on the morphology, microstructure and me-
melted material due to the enlargement of the keyhole [3]. Ad- chanical properties of dissimilar AA2024-T3/AA7075-T6 T-joints per-
ditionally, Yang et al. [14] state that supplying the filler wire and formed successively on both sides using a high-power disk laser and a
shielding gas on the same plane as the laser beam stabilizes the welding filler wire.
process, thereby reducing the keyhole-porosity.
The other welding parameters have influence on the weld strength 2. Materials and experimental methods
as well. The welding speed and the feed rate of the filler wire affect the
strength of the welds, because they influence the weld build up [15] 2.1. Base materials and filler wire
but, Cicalã et al. [16] mention that the second factor is more relevant
for the T-joint’s mechanical strength. Furthermore, weld penetration Dissimilar T-joints composed of AA2024-T3 skin and AA7075-T6
can be reduced, and mechanical strength improved in T-joints, by using stringer materials were welded using a high Si content filler wire
AA4047 filler wire instead of AA5356 [17]. AA4047 of 1 mm diameter, in order to prevent hot cracking [27]. The
Solidification cracking was the main defect observed by Siqueira skin and stringer sheets of 500 mm x 160 mm x 2 mm and 500 mm
et al. [18] in autogenous one-sided laser welded T-joints of AA6013 x 40 mm x 2 mm, respectively, were removed perpendicular to the sheet
alloy. This difficulty can be avoided by using filler wire with a high rolling direction. Table 1 shows the chemical composition of the alu-
silicon content [19]. Nevertheless, cracks will arise even with this metal minum alloys and filler wire used in this study and Table 2 shows the
filler if its feed rate is not enough to compensate the formation of cracks mechanical properties of the base materials.
during the solidification [16]. For autogenous welds, solidification
cracking can be avoided by decreasing the welding speed or enlarging 2.2. Experimental set-up and welding conditions
the diameter of the laser beam [20].
Aluminum alloys of 2xxx and 7xxx series have been widely used in Before carrying out the welding, the oxide and contaminated layers
the manufacturing of riveted aeronautic structures due to their high were removed from the work-piece surfaces by burnishing with a steel
strength to weight ratio, but both are considered not easily weldable by brush and then the surfaces were cleaned with acetone. These proce-
fusion welding [2], due to defect formation during welding. Apart from dures were performed to minimize hydrogen porosity [28].
the porosity caused by hydrogen during molten pool solidification [21], Then, the skin and stringer sheets were fastened as shown in Fig. 1.
one of the main defects observed during the laser and hybrid laser-arc The successive double-sided laser beam welded T-joint was performed
welding of these aluminum alloys was macro-porosity, caused by the using a continuous wave disk laser TruDisk 16002 with a focusing optic
instability of the keyhole [22]. Besides this, 7xxx alloys have high hot BEO D70. The maximum power of the laser equipment was 12 kW. The
cracking susceptibility in welding, increasing significantly with the Cu wavelength of the laser beam was 1.030 μm. Moreover, the focal posi-
content [23]. In the same way, the AA2024 alloy is also susceptible to tion was on the stringer surface (0.0 mm) and the focal length was of
hot cracking [21]. 200 mm. Therefore, the enlargement of the laser beam’s diameter was
According to Tan et al. [24], the investigation in advanced struc- carried out by replacing the optical fiber for one with a larger diameter.
tural engineering has given great attention to dissimilar joints of metals The shielding gas used was argon with a flow rate of 10 l/min. Fig. 2a
with very different melting points, being the laser welding-brazing an shows the incident beam position, and the incident beam angle α,
alternative joint technique for this kind of metal combination. In the which is 5° lower than the angle formed by the skin’s surface and the
case of dissimilar joints of aluminium alloys, the majority of works supply plane. The filler wire and the shielding gas were supplied on that
about laser beam welding have addressed weldable alloys such as the plane according to the scheme presented in Fig. 2b. This figure also
Table 1
Chemical composition of the base materials and filler wire (Wt. %).
Alloy Si Fe Cu Mn Mg Cr Zn Ti
AA2024 Max 0.50 Max 0.50 3.80 4.90 0.30 0.90 1.20 1.80 – Max 0.25 Max 0.15
AA7075 Max 0.40 Max 0.50 1.20 2.00 Max 0.30 2.10 2.90 0.18 0.28 5.10 6.10 Max 0.20
AA4047 11.00 13.00 Max 0.6 Max 0.30 Max 0.15 Max 0.10 – Max 0.20 Max 0.15
150
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Table 2
Mechanical properties of the base materials.
Ultimate stress (MPa) Yield stress (MPa) Young’s modulus (GPa) Strain at ultimate stress (%) Microhardness (HV0.2)
Table 3
Range of welding parameters studied.
P (kW) s (mm/s) R (m/min) α (º) δ (mm) D (μm)
2.3 – 3.5 35, 40, 45 3.8 – 4.8 10, 15, 20 0.0 – 0.6 400, 600
Table 4
Welding parameters used in the series studied.
Weld series D (μm) α (º) s (mm/s) R (m/min) P (kW) δ (mm)
shows that the gas nozzle was set behind the laser beam.
The welding parameters, such as the laser beam power, the welding
speed, the feed rate of the filler wire, the incident beam angle, the in-
cident beam position and the laser beam diameter, were varied ac-
cording to Table 3.
The welding parameters producing weld seams with the best visual
appearance and appropriate welding penetration depth in the skin were
chosen for this research. Table 4 presents the weld series chosen.
151
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
152
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Fig. 5. a) Geometry of pull-out test specimens; b) Device for clamping specimen to perform pull-out testing.
For the successive double-sided laser welded T-joints studied in this According to Yang et al. [14] an Over-Aged Zone also exists, where both
work, the macroporosity formation has the same mechanisms as those second-phase particle size and number decrease with the distance from
reported in the literature [11]. However, the interaction between the the fusion line, but it is difficult to see in optical microscopy. As can be
first and second weld seams leads to both a higher number and a greater seen in these figures, there is a chill zone near the fusion line between
size of macropores in second seam than in the first one, because the the weld and the stringer base material, whereas there is no chill zone
macropores at the first seam root are absorbed at the second seam. near the fusion line between the weld and the skin base material.
Thus, the macropores in the second seam are formed by gas cavities Fig. 13c shows that the center of the fusion zone has a fine cellular
from the first seam and by gas bubbles formed due to keyhole collapses dendrite structure like the microstructure presented by Squillace and
that occurred during the second weld seam. Prisco [17] for simultaneously double-sided laser welded T-joints.
According to Fig. 12, the only welding parameter that significantly Fig. 13d shows the grain refinement of the previous structure of the first
affects the macro-porosity is α. Thus, the percentage of macroporosity seam, generating very fine equiaxed grain structure along the whole
increases when α is increased. This occurs, according to Oliveira et al. fusion line between the weld seams.
[10], because the gap between the skin and stringer plates affects the Furthermore, scanning electron micrographs of the weld series
instability of the process for high angles. Additionally, the data scatter showed some liquation cracking in the StHAZ and little liquation in the
presented in Fig. 12 is associated to the variation of other welding SkHAZ, as is illustrated in Fig. 14a and b respectively, for weld series
parameters that are not considered significant by the t-test. Therefore, 11. This can be explained by the high liquation cracking susceptibility
the highest scatter showed when the incident angle is equal to 15° is of 7xxx and 2xxx alloys during welding, as mentioned by Huang et al.
related to the variation of laser power, incident beam position, laser [30].
beam diameter and welding speed whilst the lowest scatter presented
for angle of 10° is linked only to the variation in the position of the
incident beam. 3.2. Mechanical testing
Fig. 6. Weld appearance of the first and the second weld seams respectively of: a) and d) weld series 5; b) and c) weld series 8.
153
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Table 5
Relevant morphological results.
Weld Series a (mm) b (mm) c (mm) d (mm) Porosity (%)
hardening precipitates but also to the use of a filler metal (AA4047) Comparing these microhardness results with those obtained by Song
with less mechanical strength. Furthermore, the valley hardness in the et al. [31] for friction stir welded lap-joints of AA2024-T3/AA7075-T6,
SkHAZ is 89.5 ± 4.5% of the hardness of the skin base material and is it is possible to note that the minimum hardness value is 15 HV0.2 higher
currently attributed to the coarsening of the strengthening precipitates in laser beam welded T-joints.
[14].
As shown in Fig. 15b, the hardness profile through the stringer has
3.2.2. Pull-out testing
the lowest hardness at the FZ due to the same mechanisms mentioned
Table 6 shows the mean values of the maximum load required for
above. Furthermore, the SkHAZ hardness decreases rapidly near the FL
the rupture of the test pieces. The values of the mean ultimate tensile
while the StHAZ hardness increases rapidly from the FL until it reaches
stress obtained by dividing the load by the stringer cross-section are
the same hardness as the StBM. The causes involved in the variation of
also presented in the same table, in order to allow a comparison with
hardness are the same as those already mentioned; however, it is ob-
the results of other authors. These results are higher than the maximum
served that the stringer hardness is much higher than the skin.
stress of the pull-out tests obtained by Enz et al. [3] for single-sided and
154
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Fig. 9. Relation between measured b, and b calculated by the regression. Fig. 12. Relation between α and porosity.
strong influence on the UTL, as can be seen in Fig. 17. Thus, the UTL
decreases with an increase in d. Considering that the filler wire volume
per weld length is almost the same for all weld series, this effect cannot
be attributed mainly to filler feed rate, as observed by Cicalã et al. [16],
but to the parameters mentioned above.
The fractured specimens presented in Fig. 18 show the features of
the two different failure modes. In the first, the complete failure took
place mainly along the fusion boundary of the skin, as can be seen in the
Fig. 18a. In the second, the complete failure occurred mainly along the
fusion boundary of the stringer, as shown by Fig. 18b. Furthermore, for
all cases in which the second failure mode occurred, another crack
occurred with the same characteristics of the first failure mode, but
without leading to a complete failure of the specimen. The first failure
Fig. 10. Relation between c and D.
mode was more common than the second one, which occurred in some
specimens of weld series 2, 5 and 6 as well as in every one of weld series
successive double-sided LBW of 2 mm thick AA2024-T351/AA7050- 4, as shown in Table 6.
T76 T-joints whose approximate values were 230 and 210 MPa, re- No direct correlation between the pull-out test results and the level
spectively. In the same way, the present results are higher than the pull- of porosity was observed. However, the welds that display the second
out test results obtained by Yang et al. [14] for a simultaneous double- fracture mode present many macropores on the fracture surface.
sided LBW of 1.8 mm thick AA6056-T4/AA6156-T6 T-joint whose mean Furthermore, the crack initiation always occurred on the second weld
value was approximately 204 MPa. In all cases mentioned above the seam side, which has higher porosity than the first seam. These facts
AA4047 filler wire was used. suggest that macroporosity influences fracture initiation.
The welding parameters that significantly affect the UTL are D, P For any failure mode, the fractures always started at the second
and δ . Thus, the UTL increases when D enlarges, P decreases and δ weld seam. For failure mode 1, the crack initiation occurred at the lo-
decreases considering that only one welding parameter is varied at a cation with the highest strain concentration on the weld seam surface
time, as shown by the coefficients presented in Fig. 16. The UTL rises causing the crack to start within the weld metal, as can be seen in
with the increase in D because when this occurs, for the same welding Fig. 19a and b. Thus, the crack initiation did not occur at the weld toe,
speed, the increase in laser interaction time (D/s) leads to a lower as expected. This can be explained by the high deformation in the thin
cooling rate which reduces the susceptibility to cracking and hence plate of skin during the pull-out test [3] and by the presence of mac-
improves the mechanical strength of the welded joint [20]. Further- ropores within the weld bead, resulting in higher deformations near the
more, the UTL decreases when P increases because the higher the laser weld seam surface. Then, after the crack initiation, the crack grew to-
power, the higher the dissolution of precipitate phases in the material wards the fusion boundary of the skin and propagated in this region
[28], as well as the bigger keyhole size caused by the increase in P, the until the complete failure, as shown in Fig. 19c.
higher the reduction in the UTL caused by the increase of the partici- The crack initiation for failure mode 2 occurred in the following
pation of the filler wire in the weld which has lower mechanical way. A first crack initiated with the same mechanism of failure mode 1,
strength than the base material. The pull-out results also decrease with as can be observed in Fig. 20a and b. This crack grew until reaching a
an increase in δ, because of the lower interaction of the laser beam with macropore, which reduced the restraint for opening the first crack,
the skin material. That is, the highest UTL occurred when δ was equal to resulting in a high deformation at the region indicated by the arrow in
0.0 mm. This result is not in agreement with the results of Yang et al. Fig. 20c. After that, a second crack initiated on the weld seam surface at
[14] for simultaneous double-sided LBW of a dissimilar AA6056-T4/ the region indicated in Fig. 20c. Thus, the second crack grew towards
AA6156-T6 T-joint since, for them, the highest ultimate tensile stress the stringer FL and then it began to grow along the stringer fusion
occurred when the incident beam position was 0.2 mm. boundary up to the full failure, as can be seen in Fig. 20d. Moreover, the
Among the weld dimensions mentioned in this work, only d has a
155
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Fig. 13. Optical micrograph of weld series 10 showing: a) the boundary between FZ and StBM, b) the boundary between FZ and SkBM, c) the FZ center and d) the
transition between first and second layers in the FZ.
second crack always led to a complete fracture since the cross-section of of the pull-out test specimens.
the area to resist the test load was smaller for the second crack than for
the first, as well the susceptibility of liquation cracking in the StHAZ is 3.2.3. Fracture morphology
higher than in the SkHAZ. These observations show that defects such as SEM fractographies show that the specimens with different failure
porosity and liquation cracks play an important role in the failure mode modes have different fracture morphologies. Furthermore, each
Fig. 14. Scanning electron micrographs of welding series 11 showing: a) the liquation cracking in the StHAZ and b) in the SkHAZ.
156
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Fig. 15. Microhardness profiles of welding series 6: a) along the skin and b) along the stringer.
Table 6
Average ultimate tensile load and stress in pull-out tests for each weld series.
Weld Average Ultimate Average Ultimate Pull-out Failure
Series Tensile Load (kN) Tensile Stress (MPa) Mode
1 11.7 343 1
2 12.5 367 1 and 2
3 11.5 337 1
4 13.5 397 2
5 12.5 366 1 and 2
6 12.1 355 1 and 2
7 12.6 370 1
8 10.7 314 1
9 12.1 355 1
10 11.9 350 1
Fig. 17. Relation between d and the UTL.
11 12.3 363 1
surface presents two different features. The part of the fracture surface
that occurred through the FZ presents zones of dimples (example
marked with the number 1 in the image) intercalated with zones of
quasi-cleavage appearance (example marked with the number 2), as
can be seen in Fig. 21a. At the other part of fracture surface, along the
skin fusion boundary, the fracture surface presents a mixture of both
brittle and ductile zones, since they show brittle fracture areas with flat
facets and ridges with dimples, both marked in the same way, as shown
in Fig. 21b. Furthermore, Fig. 21b shows a lot of micropores at the
fracture surface, marked with arrows.
For failure mode 2 of the pull-out test specimens, the fracture sur-
face presents three different regions. At the FZ, the fracture surface
presents a lot of dimples (example marked with the number 1 in the
Fig. 16. Relation between the measured UTL and the UTL calculated by re- image) intercalated with zones of quasi-cleavage appearance (example
gression. marked with the number 2 in the image), as shown in Fig. 22a. Fur-
thermore, Fig. 22b shows the presence of macropores in the FZ, in-
dicated by arrows. Along of the stringer fusion boundary, the fracture
fracture surface presents areas with different features. To identify the
presents areas of flat facets (example marked with the number 1 in the
location of each one of these areas more easily, Figs. 21 and 22 show a
image) but also zones with plastic strain (example marked with the
macrograph of a fractured pull-out specimen. The fracture started at the
number 2 in the image), as can be seen in Fig. 22c. At the junction of the
second weld seam in both cases, as can be observed.
stringer fusion boundaries of the first and second weld seams, the
For the first failure mode of a pull-out test specimen, the fracture
157
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Fig. 18. Macrograph of (a) the first and (b) the second pull-out failure modes.
Fig. 19. Strain fields during the crack initiation and propagation in failure mode 1: (a) before crack initiation, (b) after crack initiation and (c) at the complete
fracture.
Fig. 20. Strain fields during the crack initiation and propagation in failure mode 2: (a) before first crack initiation, (b) after first crack initiation, (c) before second
crack initiation and (d) at the complete fracture.
fracture surface presents areas with intergranular fractures suggesting significantly to this second mode of fracture.
the presence of liquation cracking, as can be observed in Fig. 22d.
Therefore, the presence of liquation cracking seems to contribute
158
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
Fig. 21. SEM fracture surface images of failure mode 1: a) at the FZ and b) along the fusion boundary for the specimen 1 of weld series 8.
Fig. 22. SEM fracture surface images of failure mode 2: a) at the FZ, b) at the region with macropores in the FZ, c) along the fusion boundary of the stringer and d) at
the junction of the stringer fusion boundaries of the first and second weld seams for specimen 3 of weld series 4.
4. Conclusions (2) The porosity is more frequent in the second weld seam than in the
first one and increases with the increase of the incident beam angle;
The following conclusions can be drawn based on the results pre- (3) The microstructure of the welds is little influenced by the welding
sented in this study for the dissimilar successive double-sided Laser parameters, but liquation cracking occurs in the heat-affected zone
Beam Welding of dissimilar T-joints: of the stringer (AA7075) and skin (AA2024).
(4) The ultimate tensile load in the pull-out tests is affected by the laser
(1) The weld geometry is mainly affected by the laser power, laser power, laser beam diameter and incident beam position but, the
beam diameter and welding speed; fracture mode in this test is influenced by porosity and liquation
159
P.I. Oliveira et al. Journal of Manufacturing Processes 35 (2018) 149–160
160