Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
a r t i c l e i n f o a b s t r a c t
Article history: Keyhole-induced porosity in remote laser welded Al joints leads to weakened joint strength. In the study,
Received 13 August 2016 the remote laser welding processes are numerically simulated to reveal mechanism of keyhole and
Received in revised form 6 December 2016 keyhole-induced porosity formation. It is found that porosity formation takes three steps: bubble forma-
Accepted 9 December 2016
tion, bubble floating to the back of molten pool and bubble being captured by solidification front. The
Available online 18 December 2016
porosity prevention can be achieved by interrupting one of these three steps. The process simulation
shows that violent melt flow behind the keyhole is the root cause of pore formation. It leads to keyhole
Keywords:
collapsing and resulting in large fluctuation of keyhole depth and bubble formation. The vortex-type melt
Laser welding
Numerical simulation
flow behind the keyhole is also the main cause of the bubbles floating from the keyhole’s bottom into the
Al welding molten pool; and Al’s high thermal conductivity and strong melt flow make the bubbles difficult to
Keyhole dynamics escape. Various porosity prevention approaches are simulated in the study to check their effectiveness
Porosity in terms of interrupting the three steps. Also, the corresponding experimental test are carried out as ver-
Molten pool dynamics ification. The amounts of porosity predicted by the simulations agree very well with what being observed
in the experimental test. The study suggests that high welding speed is helpful in keeping the keyhole
open and not creating strong melt flow; large forward inclination angle also creates quiescent molten
pool flow and hence makes the bubbles difficult to float into the rear molten pool. The findings from
the study provides fundamental insights into the mechanism of porosity formation during laser welding
of Al alloys and guidance in keyhole-induced porosity prevention.
Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction the welding path via rotations of two tiny galvo mirrors. As a
result, the remote laser welding process offers much faster scan-
Aluminum alloys have been extensively adopted for manufac- ning speed than the conventional one [6,7]. The remote laser weld-
turing lightweight structures [1,2] due to their light weight and ing process is autogenous and contactless, also usually in a keyhole
high relative strength [3]. Laser welding, as a process which pro- mode to generate deep penetration. As a result, the weld porosity,
vides high energy density, high welding speed and narrow heat often induced by keyhole instability during deep penetration, is
affected zone, is ideal for joining thin automotive panels and mate- one big challenge in remote laser welding applications [8]. Large
rials with high thermal conductivity such as aluminum alloys [4,5]. porosity strongly affects the strength and fatigue property of
The process also provides great weight reduction opportunity by welded joints [9]. Its irregular shape even causes stress concentra-
being able to work on narrow flanges. There are two types of laser tion and leads to cracking.
welding processes: conventional and remote ones, categorized The porosity induced by keyhole instability is one of the two
based on beam travelling apparatus. The conventional laser weld- types of porosities involved in Al welding. The other porosity is
ing process, where the welding optics operate at joints’ proximity purely due to low boiling-point contents such as hydrogen and
with or without filler wires, relies on the robotic movement of magnesium in the Al materials. This type of pore has perfect spher-
welding optics to move the laser beam along welding path. On ical shape, and often observed in Al welds if hydrogen content is
the other end, the remote laser welding process, which operates not controlled or there are low-boiling elements in the alloys.
at far distance such as 600 mm away, moves the laser beam along The mitigation and prevention of porosity due to low-boiling con-
tents has been well discussed in the literatures and textbooks, will
⇑ Corresponding authors. not be studied here. Irregular shaped pore formation in the
E-mail address: hui-ping.wang@gm.com (H.-p. Wang). keyhole-mode welding process has been studied and determined
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2016.12.019
0017-9310/Ó 2016 Elsevier Ltd. All rights reserved.
R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256 245
to be related to keyhole dynamics [10,11]. Experimental methods and Na [25,30,31] have done a series of work to develop a compu-
such as high-speed CCD filming [12], spectral analysis [13] and tational simulation model of laser keyhole welding, especially on
sound monitoring [14] were used to analyze behaviors of keyhole heat source model and laser-keyhole interaction. Multiple reflec-
and molten pool. Matsunawa and Katayama [15–17] developed a tion was implemented simultaneously with their proposed ray-
high speed optical and X-ray transmission imaging system to tracing technique. The Fresnel absorption model was chosen as
observe the keyhole behavior directly and study mechanism of an energy transfer mechanism from laser beam to workpiece. Their
keyhole and porosity formation. Their study observed that bubbles works greatly advanced integrity of numerical modeling of laser
were formed when the keyhole wall collapsed and trapped shield- welding processes. Methodologies being developed above, such
ing gas or air inside. This was caused when the dynamic pressure of as, heat source model and multiple reflection model, have been
metallic vapor jet from the front side of the keyhole dented the rear implemented into commercial fluid mechanics software, FLOW3D,
wall of the keyhole. By inserting a tungsten particle in the base for laser welding process simulation. Such developed FLOW3D
metal, they also studied the material flow in the molten pool sur- software is used in our study to model keyhole dynamics and pore
rounding the keyhole and observed vortex flow in the molten pool formation.
behind the keyhole. Katayama’s group has also done many exper- In the following sections, effect of weld speed and laser inclina-
imental works to study the effect of weld parameters on porosity tion angle on porosity formation are studied using numerical pro-
formation [18,20,21]. They found high welding speed or low laser cess simulations and validated by experimental test. The welding
power could lead to low-heat-input welding conditions which is process parameters are so designed that adequate and similar pen-
beneficial to suppressing porosity formation. Shielding gases could etration in the overlap joints are ensured for all the test to avoid
also be helpful to reduce porosity except that N2 was found pro- keyhole depth effect. Moreover, the welding processes are modeled
moting porosity formation. Also, they found that when laser beam using the laser module in Flow 3D which considers multi-reflection
pointing forward with an acute inclination angle with the base and Fresnel absorption of laser beam to simulate the remote laser
metal surface, porosities decreased as the inclination angle welding of AA5182. Keyhole dynamics, melt flow and pore forma-
decreased. Li et al. [19] found that large spot size could lead to tion under different welding process parameters are discussed to
low depth-to-width ratio of the keyhole which helped to maintain investigate mechanism of porosity formation. The findings are
the keyhole’s stability. That is, less bubbles were resulted, and summarized in the conclusion section.
rarely pores remained in the weld. Many literatures can be found
on experimental studies of effect of welding process parameters
on porosity formation, such as Seto and Katayama’s work in 2001
[18] on welding speed, Katayama and Mizutani’s work [20] on laser 2. Materials and experimental procedure
power and Seto and Katayama’s work in 2002 [21] on shielding
gas. However, the mechanism of these parameters affecting pore AA5182 material was chosen in this study due to its widespread
commercial usage and its susceptibility to welding porosity. The
formation have not been fully investigated.
In recent years, with the development of the computer technol- elemental concentrations of major alloying additives are listed in
Table 1. The material is much more susceptible to porosity forma-
ogy, researchers have been developing numerical models to simu-
late laser welding processes so as to study mechanism of keyhole tion than 6xxx Al alloys during a keyhole mode welding. This is
because the Mg element in AA5182 has the low vaporization point
formation, melt flow and porosity formation [22–24]. Ki et al.
[22] developed a three-dimensional laser keyhole welding model of 1090 °C.
As we have mentioned before, there are two types of porosity
using the level set method to represent liquid/vapor and solid/liq-
uid interfaces in order to model keyhole free surface evolution. No involved in Al welding. Fig. 1 shows a cross sectional picture of a
typical 5182 Al joint welded by a remote laser welding process
keyhole collapse was considered. Cho and Na [25] modeled multi-
ple reflection and Fresnel absorption of laser beam in the keyhole, with no filler wire and no shielding gas. The pore shown in the
rectangle a is a typical keyhole-induced pore, which has irregular
and their predicted keyhole dimension agreed well with the exper-
imental ones. However, no keyhole collapse was considered either. shape with traces of fluid flow on the inner wall. The keyhole-
induced pores are the focus of the study. On the other hand, the
Moraitis and Labeas [26] used finite element method to build a
local model for keyhole shape prediction based on temperature pore shown in the rectangle b is induced by metallurgical factors.
It has spherical shape with very smooth inner face. The metallurgi-
distribution and a full-scale model for thermal–mechanical process
simulation of laser welding of Al lap joints. Both models were cal factor-induced pores are generated due to low boiling point ele-
based on solid mechanics with no convective fluid flow and no free ments such as H and Mg in Al alloys or due to surface
contaminations of the sheets. It is not a major issue in AA5182
surface formation. Zhou and Tsai [27] modeled the keyhole col-
lapse and bubble formation using a two-dimensional model. Effect welding and will not be studied here. Picture in the rectangle c
shows that the keyhole-induced irregular pore may also serve as
of keyhole’s depth-to-width aspect ratio on bubble formation was
studied using the model. It was found that a small depth-to-width a cracking nucleation point under stress, weakening the joint
strength and fatigue life.
aspect ratio keyhole welding led to no bubble in the final weld. For
a medium or large aspect ratio keyhole welding process, bubble Fig. 2 gives schematics of the remote laser welding process,
which we use to laser stitch AA5182 overlap joints in the study.
was found in the final weld. No validation of the prediction was
presented in the paper. Pang et al. [28] developed a sharp interface The laser beam in the figure is from an IPG 6kw continuum wave
fiber laser system, and the two galvo mirrors are contained in a
method to capture free surface and investigate weld pool dynam-
ics. The mechanism of keyhole stability was studied and it was remote laser welding optics. The material stackup is a 1.0 mm
AA5182 plate on top of a 2.0 mm AA5182 plate. The welding is per-
found that variation of the keyhole depth could qualitatively reflect
porosity level. However, the model did not explicitly model the formed with beam’s focus plane on the top surface of the top sheet,
pore formation. Zhao et al. [29] used volume of fluid (VOF) method
to solve multi-phase problems and built a gas–liquid–solid coupled Table 1
model. But the adopted heat source model was not accurate since Chemical composition of AA5182 Al alloy.
it ignored the phenomenon of multiple reflection in deep penetra-
Element Al Mg Mn
tion. Moreover, the simulation stopped at bubble formation, not
Weight % 95.2 4.5 0.3
explaining the mechanism of subsequent pore formation. Cho
246 R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256
where /(n) is the angle between the nth incident ray and the nth welding if the plume is not suppressed appropriately. The laser
surface normal, and e is a coefficient determined by laser type and type and substrate materials are the decisive factors of generation
material properties. The absorption rate a calculated in Eq. (3) is of plasma plume. In the modeling, the effect of plasma on laser
used in each reflection point to calculate the absorbed energy and energy absorption is implicitly considered in Eq. (3) in which coef-
the lost energy in the process of reflection. That is, the absorbed ficient e is determined by laser type and material properties.
energy after the nth reflection is calculated by qðnÞ ¼ q0ðn1Þ að/ðnÞ Þ,
where q(n) represents the nth absorbed energy and q0ðn1Þ represents 3.2. Evaporation model
the reflected energy from (n 1)th reflection, which is the incident
ray for the nth reflection. By repeating the above steps, multiple- Evaporation is an important phenomenon in laser welding
reflect could be achieved. which involves gas–liquid interface, heat transfer, mass loss and
During laser welding, laser-induced plasma plume plays a cru- recoil pressure. Modeling of gas–liquid interface is challenging in
cial role in the energy transfer of laser beam. It could defocus the the simulation. To simplify the problem, the gas phase is not
laser energy, decrease the weld penetration and even change the explicitly modeled in the study. Nevertheless, the effect of evapo-
welding mode from deep penetration welding to heat conduction ration is fully considered. Heat and mass loss due to evaporation
248 R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256
Tv
PR ¼ A exp B 1 ð8Þ
T
where PR is recoil pressure, A and B are two coefficients related to
material properties and equal to 80,000 and 5, respectively, for Al
alloys. Tv is evaporation temperature, T is temperature of keyhole
wall.
(3) When there is a free surface or fluid interface, the VOF 4. Numerical simulation and result discussions
method would be used to track a new fluid configuration.
(4) Energy equation is solved by means of an explicit option to Our numerical model consists two regions: a fluid region of
achieve the new temperature distribution, and the thermal 30 mm 4 mm 3 mm representing the two sheet stackup and
properties of the melt are updated. a void region on top of the fluid region for tracking free surface
deformation. The transient keyhole evolution, heat transfer
and fluid flow behavior in laser welding are solved in coupled
equations with very small time step (typically around
1.0 107 s–1.0 106 s) determined by the mesh size. Fig. 6
gives the meshes used in the process simulation. Considering the
expensive computational cost of the coupled simulation, the min-
imum mesh size is set to be 0.1 mm. Table 3 lists thermo-physical
properties of AA5182 used in the simulation.
Fig. 8. Process of bubble formation due to keyhole collapsing (P = 2.5 kW, v = 3 m/min): (a) 0.2340 s; (b) 0.2345 s; (c) 0.2350 s; and (d) 0.2355 s.
250 R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256
During welding, the laser beam shines on different locations of bottom of keyhole is close to beam’s focal plane and absorbs large
keyhole’s wall depending on the keyhole profile and laser beam’s amount of the laser heat. As a result, an extreme high temperature
relative position. This causes temperature variations on the key- region formed at the bottom of keyhole which leads to large recoil
hole wall. The recoil pressure and surface tension are both temper- pressures pushing the keyhole bottom downward and forming a
ature dependent. That is, the higher the surface temperature the deep keyhole, shown in Fig. 8(b). The lower rear keyhole wall is
higher the recoil pressure and lower the surface tension. As a now at high temperature and generating recoil pressure pushing
result, the melt flow in high temperature region is much stronger the lower melt behind the keyhole flow backward and causing a
than the melt flow in low temperature region. Fig. 8(a)–(d) displays vortex. When it comes to Fig. 8(c), the vortex melt flow becomes
one typical process of bubble formation under keyhole fluctuation so accelerated that it becomes unstable. Driven by the strong
from our simulation. The arrows in the picture are vectors of melt clockwise flow, the melt behind upper rear keyhole wall flows
velocity. When the keyhole is shallow as shown in Fig. 8(a), the toward the front keyhole wall, causing a keyhole collapsing, shown
B
B
Fig. 9. Process of bubble being brought into the rear part of molten pool under the effect of melt flow (P = 2.5 kW, v = 3 m/min): (a) 0.3340 s; (b) 0.3345 s; (c) 0.3350 s; (d)
0.3355 s; (e) 0.3360 s; (f) 0.3365 s; (g) 0.3370 s; (h) 0.3375 s; and (i) 0.3380 s.
in Fig. 8(d). The gas in the keyhole is trapped by the collapsing and ing solidification front and becomes a pore. Fig. 9(e) shows a sec-
a bubble is formed at the bottom of keyhole. The whole process ond keyhole-induced bubble being formed. Fortunately, before
lasts about 1.5 ms. Formation of a bubble may not lead to porosity this second bubble is floating away from the keyhole bottom, it
as long as the bubble can merge back with keyhole or float out is captured by the reopened keyhole as shown in Fig. 9(f)–(g). Pore
from the free surface. formation is hence prevented.
Fig. 9(a)–(i) shows the process of the formed bubble floating to In summary, the keyhole collapsing due to strong melt flow is
the back of the molten pool. The direction of arrows in the figure the cause of initial bubble formation. The formation of a bubble
show the melt flowing direction and the length of arrow represents does not necessarily lead to a pore as the bubble can be merged
the magnitude of flow velocity. The color represents the tempera- with the re-opened keyhole or float out from the free surface. How-
ture with red being hot and blue being cool. The progressive illus- ever, when the bubble is brought from the bottom of keyhole to the
tration in Fig. 9 clearly shows the bubble is gradually pushed lower back of the molten pool behind the keyhole by the strong
backward by the strong melt flow under the combined effect of vortex flow, it is in great danger of being captured by the solidifi-
vortex flow and surface forces on the keyhole wall. The bubble cation front since Al has high thermal conductivity and thus fast
stays the similar size during the process indicating not much moving solidification front. Also, Al’s melt has strong vortex flow.
change on its temperature. Once the bubble moves to the lower Together they make almost impossible for the bubble sitting in
back of the molten pool, it is easily captured by the fast approach- the lower back of molten pool to escape. When the bubble is cap-
Fig. 11. Distribution of porosity in longitudinal weld cross sections from following welding cases: (a) P = 2.5 kW, v = 3 m/min; (b) P = 5.0 kW, v = 10 m/min; and (c)
P = 6.0 kW, v = 12 m/min.
Fig. 12. Distribution of porosity in longitudinal welding sections from simulations with following sets of parameters: (a) P = 2.5 kW, v = 3 m/min; (b) P = 5.0 kW, v = 10 m/
min; and (c) P = 6.0 kW, v = 12 m/min.
252 R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256
tured by the solidification front, a pore is formed. Fig. 10 summa- dent. Effect of various welding process parameters on these three
rizes the three necessary steps for keyhole-induced porosity for- steps will be discussed next.
mation: bubble forming, bubble being brought into the lower
back of molten pool behind the keyhole and bubble being captured
by solidification front. Prevention of pore formation can be 4.2. Effect of laser power and weld speed
achieved by interrupting any one of these three steps. For example,
keeping the keyhole not collapsing can prevent the occurrence of In general, if we only increase welding speed among all the pro-
the first step and a quiescent melt flow can inhibit the Step 2 inci- cess parameters, the porosity formation decreases in the joints.
This is because the increased welding speed leads to reduced pen-
etration in the material and hence reduced keyhole depth, which is
beneficial for keyhole stability. However, adequate penetration is
crucial to ensure joint strength. Therefore, in the study of effect
of process parameters on pore formation, we increase the laser
power together with the welding speed so that penetration depths
are kept similar among all the test shown in Table 2. Fig. 11(a)–(c)
displays pictures of longitudinal cross sections of physical welds
generated with the three combinations of laser power and welding
speed for studying effect of the welding speed. The three cases lead
to similar weld penetrations as shown in the cross sectional pic-
tures. Fig. 12(a)–(c) gives the corresponding weld cross sections
predicted by the numerical simulations. The predicted penetra-
tions agree with the physical ones very well. The weld in Fig. 11
(a) is welded with low power at a low speed: P = 2.4 kW and
v = 3 m/min. Many large keyhole-induced pores are shown near
the bottom of the weld nugget. The predicted weld cross section
in Fig. 12(a) also shows large pores near the weld bottom. The weld
in Fig. 11(b) is generated with power being 5 kW and speed being
10 m/min. Keyhole-induced pores are not observed. The corre-
sponding predicted result in Fig. 12(b) shows no pore, either. The
weld in Fig. 11(c) is created with power being 6 kW and a high
welding speed being 12 m/min. No keyhole-induced pore is
observed in both physical weld and predicted weld shown in
Fig. 12(c). Fig. 13 plots the porosity area ratio which is ratio of pore
area to the whole weld cross section area calculated from the pro-
Fig. 13. Porosity area ratio at different laser power and weld speed based on
cess simulations for various combinations of power and speed. It
process simulations. clearly shows that with weld speed and laser power increased,
the formation of porosities is significantly suppressed.
Fig. 14. Predicted keyhole depth evolution curves (a) and keyhole width evolution curves (b) under different parameters.
R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256 253
To understand why high welding speed helps to prevent the speed leads to a much longer molten pool behind the keyhole. The
pore formation, we use our process simulations to check the effect long and large molten pool is beneficial for bubbles to escape if
of weld speed and laser power on keyhole dynamics. The variations there is any. Also, the long and large molten pool has large momen-
of keyhole width and depth during the welding process are shown tum to resist any interrupting force from the keyhole wall and
in Fig. 14, where the width is measured at the keyhole opening. The hence less likely to become turbulent. This is confirmed by the
keyhole width curves show that high power and high speed lead to velocity vectors plotted in Fig. 15, where the melt flow of the right
keyholes with larger opening which helps to prevent keyhole col- large molten pool is much more quiescent compared to the left
lapsing. Also, the keyhole depth curves show that high power small pool which is very turbulent. The quiet melt flow also
and high speed leads to much more stable keyholes with little fluc- reduces heat loss due to convective heat change and contributes
tuation in depth. Large fluctuation in keyhole depth is shown in to maintaining large melt pool. Furthermore, Fig. 15 shows the
low speed and low power case although the weld penetrations keyhole’s overall opening from top to bottom is much wider at
are similar in two cases. The large fluctuation in keyhole depth is high power and high speed than at low power and low speed. This
an indication of dynamic change of laser energy on the keyhole is because at high weld speed and high power the fast moving laser
wall, which is caused by violent change of keyhole shape and also beam advances the keyhole’s front wall forward while the large
very likely occurrences of keyhole collapsing. The physical weld and quiet rear molten pool helps to keep the rear wall stable, not
profile in Fig. 11(a) also displays a little higher undulation at the rushing toward the front wall. The laser position relative to key-
weld bottom compared to the smooth weld bottom line in Fig. 11 hole wall also plays important role in the melt dynamics. At high
(c) with high power and high speed. welding speed, the laser beam shines mainly on the front wall of
We then look at melt dynamics to study the cause of keyhole the keyhole and the resulted recoil pressure is mainly applied on
dynamics. Fig. 15(a) and (b) gives predicted velocity fields and the front wall, helping to advance the front wall while not adding
temperature distributions in the molten pools for both low speed large disturbance to the rear melt flow. On the other end, at low
and high speed cases. The position of laser beam relative to keyhole power and low speed, laser beam shines on both front and rear
surfaces are also illustrated. It shows that compared with the case wall. Its irradiation on the rear keyhole wall aggravates turbulent
of low power and low speed, the case of high power and high weld flow of the rear molten pool, which in turn exacerbates the keyhole
Fig. 15. Predicted velocity fields and temperature distribution in the molten pools, and laser’s position relative to keyhole surface for: (a) P = 2.5 kW, v = 3 m/min and (b)
P = 6 kW, v = 12 m/min.
Fig. 16. Pore area percentage at different inclination angle: (a) experiment results and (b) simulation results.
254 R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256
instability and leads to keyhole collapsing, and hence bubble for- tion and consequently keyhole dynamics and pore formation. In
mation. In summary, high speed and high power welding inhibits the proposed test listed in Table 2, inclination angles from 15°
the occurrence of Step 1-bubble formation by maintaining keyhole to 45° are tested to see their effects on pore formation. We have
opening during the process, and also prevents the occurrence of also used the process simulation to study the physics in the pro-
Step 2-bubble floating to the rear melt pool by creating a quiet cess. Laser welding with beam inclination angles being 15°, 0°,
melt flow. In addition, the process inhibits the bubble being cap- 15°, 30° and 45° are modeled to see the angle’s effect on pore for-
tured by solidification line by having a large long molten pool mation. In order to maintain similar weld penetration, the laser
which gives time for the bubble to escape. Pore formations are thus power increases from 2.5 kW to 3.0 kW for the cases with large
prevented. inclination angles of 30° and 45°.
Fig. 16(a) plots the pore area percentages based on the experi-
4.3. Effect of laser beam’s inclination angle mental test results and Fig. 16(b) gives the ones based on the sim-
ulations. It is found that the porosity is suppressed with the
Laser beam’s inclination angle is another important parameter increase of the inclination angle, shown in both experimental
in laser welding. In a conventional laser welding which welds at results and simulation results. The predicted trend in pore percent-
a close distance, to prevent laser light’s back reflection, the laser age is in good agreement with the trend revealed by the physical
beam inclination angle is usually maintained at a value between test. One thing worth to mention is that since we conduct the test
10° and 5°. That is, the beam pointing backward and in a drag- with an optics at a fixed position, laser beam focal plane has a
ging mode. While in remote laser welding, where the optics is far noticeable shift during the process when the inclination angle is
away from the welding surface, a zero inclination angle is permit- large. This is reflected in changes of weld penetration as the shift
ted. During welding, since the laser beam’s movement is achieved of focal plane causes change in laser energy and hence aggravates
by a galvo mirror’s rotation, the inclination angle in the process has keyhole dynamics resulting in collapsing. That is why more pores
a small variation during sweeping. Assuming the optics is 500 mm are observed in the experimental test. Nevertheless, the overall
away from the material surface, for a 25 mm long stitch, the trend is in agreement with the simulations which assumes the
sweeping angle is 2.87°. That is, the inclination angle changes from focal plane stays on the top material surface. This change of focal
0° to 1.5°, which is negligibly small. However, we expect to see a plane can be corrected by programming corresponding optics
large laser beam inclination angle has impact on keyhole orienta- movement. Fig. 17(a)–(e) gives predicted pore distribution in lon-
Fig. 17. Calculated pore distribution in weld cross sections for the following cases: (a) 2.5 kW, 3 m/min and 15°; (b) 2.5 kW, 3 m/min and 0°; (c) 2.5 kW, 3 m/min and 15°;
(d) 3 kW, 3 m/min and 30°; and (e) 3 kW, 3 m/min and 45°.
R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256 255
Fig. 18. Calculated velocity fields and temperature distribution in the molten pools for different inclination angles and laser powers: (a) 2.5 kW, 3 m/min and 15°; (b) 3 kW,
3 m/min and 30°; and (c) 3 kW, 3 m/min and 45°.
gitudinal weld cross sections for inclination angles being 15°, 0°, increased to 45°, the generated rear molten pool has a large area
15°, 30°, 45°, respectively. A clear decrease of pores are displayed of laminar flow going away from the rear keyhole wall. This is
as the angle increases. because at the rear keyhole wall which now has a very shallow
Now let’s examine the melt and keyhole dynamics to under- slope, the directions of gravity and recoil pressure become more
stand why the laser inclination angle affects the porosity ratio. aligned, and the resulted force drives the flow downward and
Fig. 18(a)–(c) gives contour plots of the melt temperature and vec- backward. This type of laminar flow greatly prevents the keyhole
tor plots of the melt flow around the keyhole for the cases with collapsing since the melt has to work against the gravity to reach
inclination angles being 15°, 30° and 45°, respectively. It shows the upper/front keyhole wall. With an adequate laser power, the
that the laser beam inclination angle determines the keyhole’s ori- front wall can be held by the laser beam from collapsing down.
entation, and affects rear molten pool shape and melt flow. At an As shown in the simulations, the keyhole is maintained open in
inclination angle being 15°, the laser beam generates a rear molten the process and pore formation is prevented by inhibiting occur-
pool with a violent vortex flow. The keyhole collapsing is observed rence of Step 1. Also, without a strong vortex flow, Step 2 is inhib-
with bubble formation. The turbulent rear molten pool also has a ited since the bubble is hard to be brought into the rear part of
deep and narrow shape, which makes the bubble easy to be molten pool. In summary, in the case of small laser inclination
brought into the rear molten pool but hard to escape from the fast angle, the dynamic recoil pressure and gravity are driving the melt
approaching solidification line. Pore formation is almost inevitable. moving to different directions and causing strong vortex flow, sus-
At an inclination angle being 30°, the laser beam creates a rear ceptible to keyhole collapsing. In the case of the large laser inclina-
molten pool with two vortex flow. The lower vortex flow is driven tion angle, the dynamic recoil pressure and gravity are driving the
by the recoil pressure and bottom flow coming from the molten melt moving to similar directions and resulting in laminar flow,
pool in front of the keyhole. The upper vortex flow is formed preventing keyhole collapsing.
because of Marangoni effect which leads to a backward flow on
molten pool surface. The keyhole’s rear wall in the case of 30° is
less steep compared to the case of 15°. At this position, the keyhole 5. Conclusions
rear wall is under a composite force pushing the wall backward
and down. As a result, the melt flows away from the rear wall as So far, we have studied the formation mechanism of keyhole-
shown in Fig. 18(b) and two vortex flow are formed at a distance induced pores and effect of process parameters on the pore forma-
behind the rear wall. The composite force helps to maintain the tion. We started our study by modeling the remote laser welding of
keyhole open while the laser beam keeps the front wall advancing AA5182 to investigate the keyhole and molten pool dynamics and
forward. Consequently, the keyhole collapsing is less likely to occur their relationship with keyhole-induced pore formation. We have
and bubbles are less likely to form. When the inclination angle found pore formation is directly reflected in keyhole dynamics
256 R. Lin et al. / International Journal of Heat and Mass Transfer 108 (2017) 244–256
and molten pool dynamics, where violent-melt-flow-induced key- [8] A. Matsunawa, Problems and solutions in deep penetration laser welding, Sci.
Technol. Weld. Join. 6 (2001) 351–354.
hole collapsing is the start of bubble formation and the vortex flow
[9] L. Tsay, Y.P. Shan, Y.H. Chao, W. Shu, The influence of porosity on the fatigue
in the molten pool behind the keyhole is the main reason of a bub- crack growth behavior of Ti–6Al–4V laser welds, J. Mater. Sci. 41 (2006) 7498–
ble entering into the rear molten pool. We concluded that the pore 7505.
formation in a weld requires three necessary steps: Step 1, bubble [10] S. Katayama, N. Seto, M. Mizutani, A. Matsunawa, Formation mechanism of
porosity in high power YAG laser welding, in: Proceedings of the International
formation caused by keyhole collapsing; Step 2, bubble being Congress on Applications of Lasers and Electro-optics, ICALEO, Springer-
brought from the keyhole bottom into rear molten pool; Step 3, Verlag, 2000, p. 200016.
the bubble being captured by the solidification front. Prevention [11] J. Zhou, H.L. Tsai, Effects of electromagnetic force on melt flow and porosity
prevention in pulsed laser keyhole welding, Int. J. Heat Mass Transfer 50
of pore formation can be achieved by interrupting one of the three (2007) 2217–2235.
steps. [12] I. Eriksson, P. Gren, J. Powell, A.F. Kaplan, New high-speed photography
The effect of welding speed and laser beam inclination angle on technique for observation of fluid flow in laser welding, Opt. Eng. 49 (2010),
100503–100503-3.
pore formation were studied next via both physical test and [13] X. Hao, G. Song, Spectral analysis of the plasma in low-power laser/arc hybrid
numerical process simulation. Mechanism of keyhole induced welding of magnesium alloy, IEEE Trans. Plasma Sci. 37 (2009) 76–82.
porosity formation under various process conditions were investi- [14] W. Huang, R. Kovacevic, Feasibility study of using acoustic signals for online
monitoring of the depth of weld in the laser welding of high-strength steels,
gated. Our key findings are summarized as follows: Proc. Inst. Mech. Eng. B-J. Eng. Manuf. 223 (2009) 343–361.
[15] A. Matsunawa, N. Seto, J.D. Kim, M. Mizutani, S. Katayama, Dynamics of
1. At high power and high welding speed, the keyhole has large keyhole and molten pool in high-power CO2 laser welding, in: Advanced High-
Power Lasers and Applications, International Society for Optics and Photonics,
opening and little fluctuation in depth. The laser beam mainly
2000, pp. 34–45.
shines on the front keyhole wall. The molten pool behind the [16] S. Fujinaga, H. Takenaka, T. Narikiyo, S. Katayama, A. Matsunawa, Direct
keyhole is long and large with quiescent flow, which is greatly observation of keyhole behaviour during pulse modulated high-power Nd:
beneficial for maintaining the keyhole open, inhibiting bubbles YAG laser irradiation, J. Phys. D Appl. Phys. 33 (2000) 492.
[17] Y. Naito, S. Katayama, A. Matsunawa, Keyhole Behavior and Liquid Flow in
to be brought into the rear pool and giving the bubbles time to Molten Pool during Laser-arc Hybrid Welding, International Society for Optics
escape. All these features inhibit the occurrence of Step 1, Step 2 and Photonics, 2003.
and Step 3. [18] N. Seto, S. Katayama, A. Matsunawa, Porosity formation mechanism and
suppression procedure in laser welding of aluminium alloys, Weld. Int. 15
2. While at low speed and low power, the keyhole opening is rel- (2001) 191–202.
atively small and the keyhole depth has large fluctuation due to [19] L. Kai, F.G. Lu, S.T. Guo, H.C. Cui, X.H. Tang, Porosity sensitivity of A356 Al alloy
collapsing. The molten pool behind the keyhole is small and has during fiber laser welding, Trans. Nonferr. Metals Soc. China 25 (2015) 2516–
2523.
turbulent flow. The laser beam shines on both front and rear [20] S. Katayama, M. Mizutani, A. Matsunawa, Development of Porosity Prevention
keyhole walls. The laser irradiation on the rear keyhole wall Procedures during Laser Welding, International Society for Optics and
aggravates turbulent flow of molten pool. All these features pro- Photonics, 2003.
[21] N. Seto, S. Katayama, A. Matsunawa, Porosity formation mechanism and
mote the occurrence of Step 1, Step 2 and Step 3. reduction method in CO2 laser welding of stainless steel, Weld. Int. 16 (2002)
3. The keyhole orientation is determined by the laser beam incli- 451–460.
nation angle. At a large inclination angle of 45°, the laminar [22] H. Ki, J. Mazumder, P.S. Mohanty, Modeling of laser keyhole welding: Part I.
Mathematical modeling, numerical methodology, role of recoil pressure,
flow, instead of vortex flow, is observed in the rear part of mol-
multiple reflections, and free surface evolution, Metall. Mater. Trans. A 33
ten pool due to the recoil pressure and gravity are working (2002) 1817–1830.
along the similar directions. The front keyhole wall is held by [23] X.H. Ye, X. Chen, Three-dimensional modelling of heat transfer and fluid flow
the laser beam. These features prevent the occurrence of Step in laser full-penetration welding, J. Phys. D Appl. Phys. 35 (2002) 1049.
[24] R. Rai, J. Elmer, T. Palmer, T. DebRoy, Heat transfer and fluid flow during
1 and Step 2. As a result, pore formation is suppressed. keyhole mode laser welding of tantalum, Ti–6Al–4V, 304L stainless steel and
vanadium, J. Phys. D Appl. Phys. 40 (2007) 5753.
[25] J.H. Cho, S.J. Na, Implementation of real-time multiple reflection and Fresnel
absorption of laser beam in keyhole, J. Phys. D Appl. Phys. 39 (2006) 5372.
References [26] G. Moraitis, G. Labeas, Residual stress and distortion calculation of laser beam
welding for aluminum lap joints, J. Mater. Process. Technol. 198 (2008) 260–
[1] H.J. Kim, G.A. Keoleian, S.J. Skerlos, Economic assessment of greenhouse gas 269.
emissions reduction by vehicle lightweighting using aluminum and high- [27] J. Zhou, H.L. Tsai, Porosity formation and prevention in pulsed laser welding, J.
strength steel, J. Ind. Ecol. 15 (2011) 64–80. Heat Transfer – Trans. ASME 129 (2007) 1014–1024.
[2] J. Long, F. Lan, J. Chen, New technology of lightweight and steel-aluminum [28] S. Pang, L. Chen, J. Zhou, Y. Yin, T. Chen, A three-dimensional sharp interface
hybrid structure car body, Chin. J. Mech. Eng. 6 (2008) 006. model for self-consistent keyhole and weld pool dynamics in deep penetration
[3] T. Kuo, H. Lin, Effects of pulse level of Nd-YAG laser on tensile properties and laser welding, J. Phys. D Appl. Phys. 44 (2010) 025301.
formability of laser weldments in automotive aluminum alloys, Mater. Sci. [29] H. Zhao, W. Niu, B. Zhang, Y. Lei, M. Kodama, T. Ishide, Modelling of keyhole
Eng. A 416 (2006) 281–289. dynamics and porosity formation considering the adaptive keyhole shape and
[4] S. Katayama, Y. Kawahito, M. Mizutani, Elucidation of laser welding three-phase coupling during deep-penetration laser welding, J. Phys. D Appl.
phenomena and factors affecting weld penetration and welding defects, Phys. 44 (2011) 485302.
Phys. Procedia 5 (2010) 9–17. [30] J.H. Cho, S.J. Na, Three-dimensional analysis of molten pool in GMA-laser
[5] J. Zhang, D. Weckman, Y. Zhou, Effects of temporal pulse shaping on cracking hybrid welding, Weld. J. 88 (2009) 35–43.
susceptibility of 6061–T6 aluminum Nd: YAG laser welds, Weld. J. (New York) [31] W.I. Cho, S.J. Na, C. Thomy, F. Vollertsen, Numerical simulation of molten pool
87 (2008) 18. dynamics in high power disk laser welding, J. Mater. Process. Technol. 212
[6] G. Tsoukantas, K. Salonitis, A. Stournaras, P. Stavropoulos, G. Chryssolouris, On (2012) 262–275.
optical design limitations of generalized two-mirror remote beam delivery [32] T. Fuhrich, P. Berger, H. Hügel, Marangoni effect in laser deep penetration
laser systems: the case of remote welding, Int. J. Adv. Manuf. Technol. 32 welding of steel, J. Laser Appl. 13 (2001) 178–186.
(2007) 932–941.
[7] J. Suh, M.Y. Lee, B.H. Jung, M.J. Song, H.S. Kang, J.O. Kim, Remote welding of
automobile components using CO2 laser and scanner, J. Weld. Join. 26 (2008)
74–78.