- Email: [email protected]
Keyhole-induced porosity formation during laser welding
Journal of Materials Processing Tech. 252 (2018) 720–727
Contents lists available at ScienceDirect
Journal of Materials Processing Tech. journal homepage: www.elsevier.com/locate/jmatprotec
Keyhole-induced porosity formation during laser welding Jiajun Xu
a,b
, Youmin Rong
a,b
a,⁎
a
MARK b
, Yu Huang , Pingjiang Wang , Chunming Wang
a
State Key Lab of Digital Manufacturing Equipment and Technology, School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, China b School of Material Science and Engineering, Huazhong University of Science and Technology, Wuhan, China
A R T I C L E I N F O
A B S T R A C T
Keywords: Laser beam welding Porosity formation Bubble development Keyhole dynamics Molten pool dynamics
Through laser welding experiment with glass, the dynamics of keyhole and molten pool could be directly observed to reveal the mechanism of porosity formation using high speed camera. The high speed images showed that large fluctuation of keyhole was responsible for the increased bubble formation, and bubble merging resulted in the formation of large pore. The porosity of the welded samples was characterized qualitatively and quantitatively by combining super depth microscope and image processing. Porosity characteristics such as pore number, pore diameter (maximum and mean), porosity volume and porosity ratio varied with changes in welding speeds and laser powers. Increased energy density led to a decrease in the number but an increase in the other characteristics, indicating that the higher the energy density the more violent the bubble merging and giving rise to an increased possibility to form large pores.
1. Introduction Laser beam welding, due to its high energy density, high welding speed and narrow heat affected zone, has been widely applied in automobile, shipbuilding, aerospace and energy. Porosity is one of the most common and undesirable defects, which heavily degrades the properties of laser welds such as strength and fatigue. To suppress the defect, extensive studies have been performed aimed at understanding the formation of porosity and finding its driving forces. Lin et al. (2017) reported pores could be divided into two types, one was keyhole-induced porosity, the other was metallurgical factor-induced porosity. The metallurgical factor-induced pores are caused by low boiling point elements (H and N) in alloys or contaminations on the surfaces being welded. The porosity is largely induced by keyhole in 316L stainless steel welding and the metallurgical factor-induced type are not studied in this paper. Both in-process detections and post-process inspections are conducted to investigate the extremely complex dynamics of keyhole and molten pool during welding, which are the main driving forces for porosity formation. The in-process detections are performed by high speed camera, x-ray transmission imaging system and spectrometer. Seto et al. (2000) found that porosity formation was resulting from keyhole instability using a high speed optical and X-ray transmission imaging system. The images displayed that bubbles were formed when the unstable keyhole collapsed and the dynamics of keyhole and plasma had a close relationship, suggesting that the plasma dynamics correlate
⁎
with porosity formation. To overcome the problem that only the surface of samples can be observed when using high speed imaging, welding experiments with transparent materials such as glass, water and ice have been performed to observe the keyhole dynamics with low cost, high speed and high resolution. Berger et al. (2011) found that the keyhole as well as the bubble were partly filled with shielding gas and partly with metal vapor during the experiments with water and ice. The post-process inspections are performed to define the relationship between porosity defect and welding parameters by means of metallographic sectional analysis, X-ray radiography and 3D X-ray tomography. Laser power and welding speed play a large role in porosity formation. Porosity ratio decreased with the increased welding speed according to Madison and Aagesen (2012), while increased with the increased laser power according to Yu et al. (2010). The characteristics of molten pool, which are dependent upon laser power and welding speed, correlate with porosity formation. Zhang et al. (2017) proposed a volume characteristic coefficient to predict porosity area ratio. Norris et al. (2011) indicated that the average pore size increased linearly with the area of weld cross-sectional during partial penetration laser welding with 304L stainless steel. As the main portion of porosity, shielding gas are strongly associated with porosity formation. Mazar Atabaki et al. (2015) reported that porosity ratio could be best mitigated by applying the side shielding gas of 92%Argon–8%CO in hybrid laser/arc welds of advanced high strength steel. Elmer (2015) found the porosity still existed in A36 and 304L weld when using Ar but very low or no porosity when using N2.
Corresponding author. E-mail addresses: [email protected] (J. Xu), [email protected] (P. Wang).
http://dx.doi.org/10.1016/j.jmatprotec.2017.10.038 Received 19 July 2017; Received in revised form 11 October 2017; Accepted 24 October 2017 Available online 06 November 2017 0924-0136/ © 2017 Elsevier B.V. All rights reserved.
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Table 3 Welding parameters adopted in the experimentation.
Table 1 Physical properties of the adopted materials in this study.
−3
Density, (kg m ) Thermal expansion coefficient, (K−1) Thermal conductivity, (Wm−1 K−1) Softening point, (°C) Melting point, (°C)
Quartz glass
316L
2200 5.7 × 10−7 1.4 1680 –
7960 1.893 × 10−5 14.15 – 1375–1450
Value
Laser power, (kW) Welding speed, (m min−1) Shielding gas flow, (m3/h) Defocusing distance, (mm) Distance between laser beam and glass (DLG), (mm)
2.4, 2.7, 3.0 1.0, 1.3, 1.6 1.5 −3 0.25, 0.5
band-pass filter with a transmission band of 808 nm. A laser illuminating system (CAVITAR CAVILUX™ Smart) was employed to illuminate the welding zone, synchronizing with the high speed camera. This employs a pulse laser with a wavelength of 808 nm, an output power of 500W and a frequency ranging from 100 Hz to 100000 Hz. The images were taken at 2000 frames/s. The materials were 316L austenitic stainless steel in the dimension of 100 × 50 × 10 mm, and heat resistant quartz glass in the dimension of 100 × 5 × 20 mm. The target chemical compositions in weight percentage and physical properties of the stainless steel and glass are shown in Tables 1 and 2. The laser beam was focused on the stainless steel with a distance (DLG) to the edge of quartz glass. Pang et al. (2016) reported that keyhole had a diameter of 0.33 mm at the neck and a maximum temperature of approximately 2800 °C on the wall during the laser beam welding simulation of stainless steel using 1500W laser power and 3 m/min welding speed. This indicates that molten pool behind keyhole cannot be observed because the glass is softened by the high temperature keyhole wall when the DLG is too small, whereas keyhole cannot be observed when the DSG is large enough to avoid glass softening due to the limited keyhole size. To directly observe the dynamic behaviors of weld pool and keyhole, two DLG’s of 0.25 mm and 0.50 mm were employed. The welding parameters are shown in Table 3. High speed images were processed to investigate keyhole dynamics using Matlab. The height of keyhole was considered as the characteristic parameter, the process is shown in Fig. 2. Otsu’s method reported by N. Otsu (1979) was used to segment the images automatically. To improve the precision of threshold segmentation, a moving window was used to cut the images. A digital super depth microscope (KEYENCE VHX-1000C) was used to observe the microstructure and characterize the porosity of the 316L welds. Hough circle transform improved by Roushdy (2007) was carried out to detect the pores of the microstructure images in 50 magnifications. The transform was improved in this study, with which the diameter of every pore can be detected with high efficiency and high spatial resolution. The resolution can be as fine as 4.12 μm. Porosity characteristics such as pore number, pore diameter, porosity volume and porosity ratio in each laser welded weld were calculated.
Fig. 1. Experiment setup schematic, where the angle between laser beam and work piece surface, α is 80°, and the distance of laser beam center to the edge of quartz glass, DLG has two setups of 0.25 mm and 0.5 mm, respectively.
Material
Welding parameters
On the other hand, computational fluid dynamics (CFD) methods which can be considered as the combination of in-process detections and post-process inspections, have recently been applied not only to understand the keyhole dynamics and porosity formation mechanism, but also to give a more comprehensive guidance for welding parameter optimization to achieve a better quality weld. Lin et al. (2017) simulated porosity formation during laser welding, the results revealed that porosity formation experiences three steps: keyhole-induced bubble formation, bubble floating in the molten pool and pore formation with the solidifying molten pool. Panwisawas et al. (2017) developed a physics-based model to simulate keyhole and porosity formation, and summarized the effect of the welding parameters on maximum size of pore in a normalized processing diagram. However, the mechanism of porosity formation has not been fully investigated. Bubble development induced by molten pool dynamics, especially the development from a small to a large when bubble is floating in the molten pool, have rarely been studied. In this paper, laser welding experiments with glasses were performed to directly observe the dynamics of molten pool and keyhole using high speed camera. Keyhole-induced bubble (KIB) formation and molten pool-induced bubble development were investigated aiming to reveal the mechanism of porosity formation.
3. Results and discussion
2. Experimental procedure
3.1. Bubble formation mechanism
The experiment setup is illustrated in Fig. 1. Laser welding experiments were conducted using a fiber laser (IPG YLS-4000) with a maximum output power of 4 kW, a wavelength of 1060 nm and a beam spot size of 0.35 mm in diameter. High speed imaging was performed on the quartz glass using a high speed camera (Phantom V611) together with a
Wang et al. (2012) reported that plasma during laser beam welding could be divided into keyhole plasma and floating plasma both with a fluctuation in size. Keyhole plasma, the size of which can be seen as the size of keyhole, is the main portion of plasma with the most brightness, whereas floating plasma is with little brightness above the surface of
Table 2 Chemical compositions of the adopted materials (wt-%). Element
SiO2
C
Si
Mn
P
S
Cr
Mo
Ni
316L Glass
– 99.97–99.99
0.021 –
0.77 –
1.019 –
0.039 –
0.001 –
16.92 –
2.03 –
12.16 –
721
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Fig. 2. Flow chart of image processing.
Fig. 3. (a) High speed images showing the process of keyhole formation and (b) keyhole depth evolution. The red arrow shows the laser welding direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4. Relationship between keyhole instability and porosity formation: (a) keyhole depth evolution of 3600 ms, (b) corresponding porosity in longitudinal section, (c–f) four typical 3D views of (b).
722
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Fig. 5. High speed images showing the formation processes of three bubbles. Note that the three pores of pore 1, pore 2 and pore 3 in Fig. 4(d) were formed by the three bubbles, respectively. The red arrow shows the laser welding direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 6. Schematic diagram showing bubble formation mechanism, (a–b) keyhole formation, (c) keyhole collapse and (d) bubble formation. Note that P v is vapor pressure in keyhole and P m is the pressure of molten pool.
Fig. 7. (a) High speed images showing longitudinal molten pool, (b) pool depth and (c) pool length of the 0.5 mm DLG welding using different welding parameters.
723
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Fig. 8. Bubble development and pore formation when using the parameters of 3000 W laser power and 1 m/min welding speed.
sample. Fig. 3(a) shows the process of keyhole formation during the 0.25 mm DLG welding with the parameters of 3000 W laser power and 1 m/min welding speed. Floating plasma was firstly formed at 0.5 ms due to the melting and evaporation of metal at the surface. Keyhole plasma as well as keyhole could be seen at 1.5 ms. Fig. 3(b) shows the keyhole depth evolution during the keyhole formation process. The depth was increasing linearly at an average speed of 0.356 mm/ms until 8.5 ms. The depth increased with a fluctuation after 8.5 ms and a quasisteady keyhole was formed at 30 ms in a depth of 4.524 mm. To investigate the keyhole dynamics during welding, a total of 7201 high speed images were calculated by image processing. The time interval of each image is 0.5 ms. Fig. 4(a) shows the keyhole depth evolution of 3600 ms during the welding. The keyhole was extremely unstable and the fluctuation changed with time. This instability is caused by the floating plasma, in which part of laser is absorbed by inverse bremsstrahlung according to Lacroix and Boudot (1998). The
Fig. 9. Schematic of bubble after development, where m denotes the number of merged bubble (MB) and n denotes the number of keyhole-induced bubble (KIB). Note that the merged bubble is composed of several KIB’s.
Fig. 10. Characterization of the porosity formed in Fig. 7: (a) pore height measurement, (b) 3D topography and (c) pore diameter extraction.
724
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
images showing the process of bubble formation were analyzed. Fig. 5 shows the formation process of three bubbles which are the initial states of the three pores in Fig. 4(d). The high speed images display that a bubble (KIB) is formed due to the collapse of keyhole wall when the fluctuation is large enough. Pang et al. (2014) simulated the formation of bubble and found the vapor pressure of keyhole decreased before the formation of bubble. Fig. 6 shows the schematic diagram of bubble formation mechanism. The process can be detailed as follow: in a keyhole collapse cycle, the depth of keyhole firstly increases shown as Fig. 6(a–b), accompanying with a decrease of Pv and an increase in the size of floating plasma (the vapor pressure in keyhole P v is larger than the pressure of molten pool Ps, Pv > Ps). The decreased Pv causes a shielding gas flow from the keyhole opening to the keyhole tip, suggesting that the keyhole is composed of metallic vapor and shielding gas. The increased floating plasma leads to a decrease in intensity of the incident laser. When the intensity reaches a certain limit as Fig. 6(c), the incident laser could not result enough metal vapor to maintain the keyhole and the keyhole begins to collapse (the vapor pressure in keyhole P v is smaller than the pressure of molten pool Ps, Pv < Ps). Note that the time taken for keyhole collapse is very short that the keyhole is cut into two parts by molten pool, e.g. three time of 0.5 ms, 3.5 ms and 4 ms can be observed according to the red dotted line in Fig. 5. The lower part is rapidly cooled and compressed by the surrounding molten pool. The cooling gives rise to the condensation of metallic vapor in the lower part and a spherical bubble composed of shielding gas is formed at the tip of keyhole and floats with thermal fluid flow shown as Fig. 6(d). Meanwhile, the lack of metal vapor lead to a decrease in the size of floating, which in turn give rise to an increase in intensity of the incident laser. When the floating plasma decrease to a certain limit, the keyhole collapse cycle will restart from Fig. 6(a)–(d). The glass softening caused by the high temperature keyhole had a bad effect on the visualization of the features during the in-situ welding experiment. This can be verified by the unclear images (Fig. 5), the longitudinal section which are coated with a yellow layer (Fig. 4(b)) and the rough 3D topographies (Fig. 4(c–f)). The interactions of the softened glass and molten metal would affect the formation of part pores and thus only the mechanism of bubble formation was investigated during the 0.25 mm DLG welding.
Table 4 Characterization results of the eight pores. No.
Height (μm)
Radius (μm)
H/R ratio
1 2 3 4 5 6 7 8
578 409.3 325.9 250.9 236.9 233.2 175.2 141.6
892.28 644.10 481.65 433.72 416.37 353.74 322.44 228.48
0.65 0.64 0.68 0.58 0.57 0.66 0.54 0.62
Fig. 11. (a) Longitudinal sections and (b) pore numbers in different size ranges of the 0.5 mm DLG welds using different welding parameters.
3.2. Bubble development
absorption has the similar variation rule with the size of floating plasma and thus gives rise to an unstable incident laser with fluctuated intensity during the whole welding process. The size of keyhole is dependent upon the intensity of the incident laser. Fig. 4(b) shows the porosity in the longitudinal section formed within the 3600 ms. In contrast to the keyhole depth evolution (Fig. 4(a)), the porosity formation is shown to be due to the keyhole instability. The pores seem to be formed when the fluctuation is large enough. The size and number of pores depend on the magnitude of fluctuation, i.e. only two small pores can be seen when the magnitudes are on the small side (Fig. 4(c) and Fig. 4(e)), whereas four or more large pores can be seen when the magnitudes are larger (Fig. 4(d) and (f)). To further understand the mechanism of porosity formation, the
To avoid the glass softening, the 0.5 mm DLG welding experiment was performed in this section. The molten pool-induced bubble development could be observed clearly. Due to the limited keyhole size, the keyhole could not be observed. Fig. 7(a) shows the longitudinal molten pools resulting from different welding parameters. The longitudinal view of molten pool can be clearly observed according to the solid/liquid interfaces from the images. The size of molten pool varies obviously with changes in laser power and welding speed such that the lower the power or higher the speed the smaller the size. To further analyze the variations, the pool depths and lengths of different parameters were measured as shown in Fig. 7(b) and (c). The depth variations caused by laser power and
Table 5 Pore analysis results. Laser power (w)
2400 2700 3000 3000 3000
Welding speed (m/min)
1.0 1.0 1.0 1.3 1.6
H/R ratio
0.54 0.58 0.63 0.55 0.56
No. of pores
Volume (μm3)
2.65 × 109 4.42 × 109 5.45 × 109 2.71 × 109 1.63 × 109
266 207 175 224 226
725
Porosity ratio (%)
6.12 6.81 7.38 5.45 4.00
Diameter (μm) Maximum
Mean
1309.69 1502.46 1784.57 1477.33 1079.15
288.42 316.56 366.01 285.66 249.30
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
of each pore, respectively.
welding speed are similar, in respective ranges from 4.48 mm to 3.77 mm and from 4.48 mm to 3.62 mm. There is a big difference between the length variations, in ranges from 12.40 mm to 10.19 mm and from 12.40 mm to 11.8 mm, respectively. The reason is that a longer region is fused within the same time when a higher speed is used. Fig. 8 shows the development processes of two bubbles when using the welding parameters of 3000 W power and 1 m/min speed. The transition from a bubble to a pore is extremely complex, including bubble floating with the thermal fluid flow, bubble merging and pore formation with the solidifying molten pool. The detailed processes could be described as follow: Step 1: A KIB has been formed in the middle part of weld pool front at 2293.5 ms. The KIB was the initial state of bubble 1. From 2360 ms to 2377.5 ms, the KIB merged with a small KIB floated from the bottom of molten pool. Step 2: Three small KIB’s have been formed at 2418 ms and then they merged with each other to be one. The merged bubble was moving to the bubble formed in step 1, forming a larger bubble at 2430.5 ms. Step 3: This step is similar with step 1 except that the KIB floated from the bottom of molten pool was larger. The development of bubble 1 have been completed in this step. Bubble 1 could be observed at 2470.5 ms. Step 4: The development of bubble 2 could be observed in this step. It is similar with step 2. In contrast to bubble 1, bubble 2 is much smaller in size. Step 5: Two pores were formed with the solidifying molten pool from 2782 ms to 2897.5 ms. The time taken for the two developments were 177 ms and 103.5 ms, respectively. The bubble after development is composed of n KIB’s and m merged bubbles as shown in Fig. 9. The development processes indicate that the transition from a small bubble to a large bubble is caused by molten pool dynamics, in which whether the KIB’s merge with each other or not depends on the speed and surface tension of the thermal fluid flow.
4. Discussion The current research demonstrates that the molten pool dynamics has a significant effect on bubble development. The dynamics is the direct result of laser power and welding speed. To further investigate the molten pool-induced bubble development, the porosity of the 0.5 mm DLG welds using five different parameters were analyzed. Fig. 11(a) shows the longitudinal sections of the five welds. Not only the number but also the size of pores varies with welding parameters. The porosity was qualitatively and quantitatively characterized using the method performed in Fig. 10. Pores with the diameters more than 100 μm were extracted by image processing and the weighted average of H/R ratio in each weld was calculated by Eq. (1). The results are shown in Fig. 11(b) and Table 5. Fig. 11(b) shows the pore numbers in different size ranges of different welding parameters. The variations can be clearly observed that the lower the power or higher the speed, the larger the pore numbers in the size ranges of 100–300 μm and 300–500 μm, but the smaller the pore numbers in the size ranges of 700–900 μm and 900–1800 μm. Table 5 shows that porosity characteristics such as pore number, pore diameter (maximum and mean), porosity volume and porosity ratio vary obviously with changes in laser power and welding speed. The variations on the number and the other characteristics are opposite, suggesting the degree of bubble merging is strongly associated with molten pool dynamics. In the weld using the 3000 W laser power and 1 m/min welding speed, a maximum porosity volume of 5.45 × 109 μm3, a maximum pore diameter of 1784.57 μm, but a minimum pore number of 175 were calculated. The higher laser power and lower welding speed used for the 0.5 mm DLG welds has deteriorated the porosity defect and has given rise to an increased number of large pores. This is because the energy density would be more and subsequently not only the occurrence of keyhole collapse is more frequent but also the molten pool is rather more unstable with an increased possibility for bubble merging to form big pores.
3.3. Porosity characterization 5. Conclusions The porosity formed in Fig. 8 were successfully characterized by super depth microscope and image processing. Both the diameters and heights of the pores were measured as shown in Fig. 10. The pores are mostly of spherical segmental in shape and there are 37 pores in total, with diameters ranging from 70.65 μm to 1784.57 μm. The heights, radiuses and height/radius (H/R) ratios of the 8 pores in Fig. 10(c) are tabulated in Table 4. The ratios maintain within a narrow range from 0.54 to 0.68, suggesting that the H/R ratios of pores formed using the same welding parameters are similar. To calculate the volume of each pore in the weld, a new method which is hypothesized that the pores formed using the same welding parameters have the same H/R ratio is proposed. In the case of the welding using the 3000 W power and 1 m/ min speed, the H/R ratio of each pore is hypothesized to be equal to the weighted average (m) of the eight H/R ratios in Table 4. The m is defined by Eq. (1). 8
m=
∑ i=1
1. Bubble formation and bubble development could be directly observed during laser welding experiments with glasses using high speed camera. The experiments could provide an advance in understanding the porosity formation. 2. Large fluctuation of keyhole was responsible for increased bubble formation and bubble merging resulted in the formation of large pore. 3. The characterization method detailed in this study could analyze the micrometer-scale porosity qualitatively and quantitatively. Pores characterized in the welds were mostly of spherical segment in shape, with diameters of 70.65–1784.57 μm and H/R ratios of 0.54–0.63. 4. Increased laser power and decreased welding speed gave rise to an increased possibility in the formation of large pore.
hi ri3 ri ∑8j = 1 r j 3
Acknowledgements This research is supported by the National Basic Research Program of China (973 Program, NO.2014CB046703), mega project of science research of Hubei province (2016AAA070), the National Natural Science Foundation of China (NSFC) (51421062), the Fundamental Research Funds for the Central Universities (HUST: 2016YXMS271).
where hi and ri (rj) are the heights and radiuses of the 8 pores, respectively. The magnitude of m is 0.63. The total porosity volume (V) of the weld is calculated as Eq. (2). n
V=
∑ k=1
n
=
∑ k=1
1 πhk (3rk 2 + hk 2) 6
References
1 πmrk 3 (3 + m2) 6
Berger, P., Hügel, H., Graf, T., 2011. Understanding pore formation in laser beam welding. Phys. Procedia 12, 241–247. Elmer, J.W., 2015. The Effect of Ar and N2 shielding gas on laser weld porosity in steel
where, n is the total number of pores, hk and rk are the height and radius 726
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Pang, S., Shao, X., Li, W., Chen, X., Gong, S., 2016. Dynamic characteristics and mechanisms of compressible metallic vapor plume behaviors in transient keyhole during deep penetration fiber laser welding. Appl. Phys. A Mater. Sci. Process. 122. Panwisawas, C., Perumal, B., Ward, R.M., Turner, N., Turner, R.P., Brooks, J.W., Basoalto, H.C., 2017. Keyhole formation and thermal fluid flow-induced porosity during laser fusion welding in titanium alloys. Exp. Modell. Acta Mater. 126, 251–263. Roushdy, M., 2007. Detecting coins with different radii based on hough transform in noisy and deformed image. Proc. GVIP J. 7, 25–29. Seto, N., Katayama, S., Matsunawa, A., 2000. High-speed simultaneous observation of plasma and keyhole behavior during high power CO2 laser welding: effect of shielding gas on porosity formation. J. Laser Appl. 12, 245. Wang, J., Wang, C., Meng, X., Hu, X., Yu, Y., Yu, S., 2012. Study on the periodic oscillation of plasma/vapour induced during high power fibre laser penetration welding. Opt. Laser Technol. 44, 67–70. Yu, Y., Wang, C., Hu, X., Wang, J., Yu, S., 2010. Porosity in fiber laser formation of 5A06 aluminum alloy. J. Mech. Sci. Technol. 24, 1077–1082. Zhang, C., Gao, M., Wang, D., Yin, J., Zeng, X., 2017. Relationship between pool characteristic and weld porosity in laser arc hybrid welding of AA6082 aluminum alloy. J. Mater. Process. Technol. 240, 217–222.
stainless steels, and nickel. Weld. J. 94, 313–325. Lacroix, D., Boudot, G.J., 1998. Spectroscopic characterization of laser-induced plasma created during welding with a pulsed Nd: YAG laser. J. Appl. Phys. 81 (10). Lin, R., Wang, H., Lu, F., Solomon, J., Carlson, B.E., 2017. Numerical study of keyhole dynamics and keyhole-induced porosity formation in remote laser welding of Al alloys. Int. J. Heat Mass Transf. 108, 244–256. Madison, J.D., Aagesen, L.K., 2012. Quantitative characterization of porosity in laser welds of stainless steel. Scr. Mater. 67, 783–786. Mazar Atabaki, M., Ma, J., Liu, W., Kovacevic, R., 2015. Pore formation and its mitigation during hybrid laser/arc welding of advanced high strength steel. Mater. Des. 67, 509–521. Norris, J., Robino, C., Hirschfeld, D.a., Perricone, M.J., 2011. Effects of laser parameters on porosity formation: investigating millimeter scale continuous wave Nd: YAG laser welds. Weld. J. 90, 198s–203s. Otsu, N., 1979. A tlreshold selection method from gray-level histograms. IEEE Trans. Syst. Man Cybern. 62–66. Pang, S., Chen, W., Wang, W., 2014. A quantitative model of keyhole instability induced porosity in laser welding of titanium alloy. Metall. Mater. Trans. A. 45 (6), 2808–2818.
727
Contents lists available at ScienceDirect
Journal of Materials Processing Tech. journal homepage: www.elsevier.com/locate/jmatprotec
Keyhole-induced porosity formation during laser welding Jiajun Xu
a,b
, Youmin Rong
a,b
a,⁎
a
MARK b
, Yu Huang , Pingjiang Wang , Chunming Wang
a
State Key Lab of Digital Manufacturing Equipment and Technology, School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, China b School of Material Science and Engineering, Huazhong University of Science and Technology, Wuhan, China
A R T I C L E I N F O
A B S T R A C T
Keywords: Laser beam welding Porosity formation Bubble development Keyhole dynamics Molten pool dynamics
Through laser welding experiment with glass, the dynamics of keyhole and molten pool could be directly observed to reveal the mechanism of porosity formation using high speed camera. The high speed images showed that large fluctuation of keyhole was responsible for the increased bubble formation, and bubble merging resulted in the formation of large pore. The porosity of the welded samples was characterized qualitatively and quantitatively by combining super depth microscope and image processing. Porosity characteristics such as pore number, pore diameter (maximum and mean), porosity volume and porosity ratio varied with changes in welding speeds and laser powers. Increased energy density led to a decrease in the number but an increase in the other characteristics, indicating that the higher the energy density the more violent the bubble merging and giving rise to an increased possibility to form large pores.
1. Introduction Laser beam welding, due to its high energy density, high welding speed and narrow heat affected zone, has been widely applied in automobile, shipbuilding, aerospace and energy. Porosity is one of the most common and undesirable defects, which heavily degrades the properties of laser welds such as strength and fatigue. To suppress the defect, extensive studies have been performed aimed at understanding the formation of porosity and finding its driving forces. Lin et al. (2017) reported pores could be divided into two types, one was keyhole-induced porosity, the other was metallurgical factor-induced porosity. The metallurgical factor-induced pores are caused by low boiling point elements (H and N) in alloys or contaminations on the surfaces being welded. The porosity is largely induced by keyhole in 316L stainless steel welding and the metallurgical factor-induced type are not studied in this paper. Both in-process detections and post-process inspections are conducted to investigate the extremely complex dynamics of keyhole and molten pool during welding, which are the main driving forces for porosity formation. The in-process detections are performed by high speed camera, x-ray transmission imaging system and spectrometer. Seto et al. (2000) found that porosity formation was resulting from keyhole instability using a high speed optical and X-ray transmission imaging system. The images displayed that bubbles were formed when the unstable keyhole collapsed and the dynamics of keyhole and plasma had a close relationship, suggesting that the plasma dynamics correlate
⁎
with porosity formation. To overcome the problem that only the surface of samples can be observed when using high speed imaging, welding experiments with transparent materials such as glass, water and ice have been performed to observe the keyhole dynamics with low cost, high speed and high resolution. Berger et al. (2011) found that the keyhole as well as the bubble were partly filled with shielding gas and partly with metal vapor during the experiments with water and ice. The post-process inspections are performed to define the relationship between porosity defect and welding parameters by means of metallographic sectional analysis, X-ray radiography and 3D X-ray tomography. Laser power and welding speed play a large role in porosity formation. Porosity ratio decreased with the increased welding speed according to Madison and Aagesen (2012), while increased with the increased laser power according to Yu et al. (2010). The characteristics of molten pool, which are dependent upon laser power and welding speed, correlate with porosity formation. Zhang et al. (2017) proposed a volume characteristic coefficient to predict porosity area ratio. Norris et al. (2011) indicated that the average pore size increased linearly with the area of weld cross-sectional during partial penetration laser welding with 304L stainless steel. As the main portion of porosity, shielding gas are strongly associated with porosity formation. Mazar Atabaki et al. (2015) reported that porosity ratio could be best mitigated by applying the side shielding gas of 92%Argon–8%CO in hybrid laser/arc welds of advanced high strength steel. Elmer (2015) found the porosity still existed in A36 and 304L weld when using Ar but very low or no porosity when using N2.
Corresponding author. E-mail addresses: [email protected] (J. Xu), [email protected] (P. Wang).
http://dx.doi.org/10.1016/j.jmatprotec.2017.10.038 Received 19 July 2017; Received in revised form 11 October 2017; Accepted 24 October 2017 Available online 06 November 2017 0924-0136/ © 2017 Elsevier B.V. All rights reserved.
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Table 3 Welding parameters adopted in the experimentation.
Table 1 Physical properties of the adopted materials in this study.
−3
Density, (kg m ) Thermal expansion coefficient, (K−1) Thermal conductivity, (Wm−1 K−1) Softening point, (°C) Melting point, (°C)
Quartz glass
316L
2200 5.7 × 10−7 1.4 1680 –
7960 1.893 × 10−5 14.15 – 1375–1450
Value
Laser power, (kW) Welding speed, (m min−1) Shielding gas flow, (m3/h) Defocusing distance, (mm) Distance between laser beam and glass (DLG), (mm)
2.4, 2.7, 3.0 1.0, 1.3, 1.6 1.5 −3 0.25, 0.5
band-pass filter with a transmission band of 808 nm. A laser illuminating system (CAVITAR CAVILUX™ Smart) was employed to illuminate the welding zone, synchronizing with the high speed camera. This employs a pulse laser with a wavelength of 808 nm, an output power of 500W and a frequency ranging from 100 Hz to 100000 Hz. The images were taken at 2000 frames/s. The materials were 316L austenitic stainless steel in the dimension of 100 × 50 × 10 mm, and heat resistant quartz glass in the dimension of 100 × 5 × 20 mm. The target chemical compositions in weight percentage and physical properties of the stainless steel and glass are shown in Tables 1 and 2. The laser beam was focused on the stainless steel with a distance (DLG) to the edge of quartz glass. Pang et al. (2016) reported that keyhole had a diameter of 0.33 mm at the neck and a maximum temperature of approximately 2800 °C on the wall during the laser beam welding simulation of stainless steel using 1500W laser power and 3 m/min welding speed. This indicates that molten pool behind keyhole cannot be observed because the glass is softened by the high temperature keyhole wall when the DLG is too small, whereas keyhole cannot be observed when the DSG is large enough to avoid glass softening due to the limited keyhole size. To directly observe the dynamic behaviors of weld pool and keyhole, two DLG’s of 0.25 mm and 0.50 mm were employed. The welding parameters are shown in Table 3. High speed images were processed to investigate keyhole dynamics using Matlab. The height of keyhole was considered as the characteristic parameter, the process is shown in Fig. 2. Otsu’s method reported by N. Otsu (1979) was used to segment the images automatically. To improve the precision of threshold segmentation, a moving window was used to cut the images. A digital super depth microscope (KEYENCE VHX-1000C) was used to observe the microstructure and characterize the porosity of the 316L welds. Hough circle transform improved by Roushdy (2007) was carried out to detect the pores of the microstructure images in 50 magnifications. The transform was improved in this study, with which the diameter of every pore can be detected with high efficiency and high spatial resolution. The resolution can be as fine as 4.12 μm. Porosity characteristics such as pore number, pore diameter, porosity volume and porosity ratio in each laser welded weld were calculated.
Fig. 1. Experiment setup schematic, where the angle between laser beam and work piece surface, α is 80°, and the distance of laser beam center to the edge of quartz glass, DLG has two setups of 0.25 mm and 0.5 mm, respectively.
Material
Welding parameters
On the other hand, computational fluid dynamics (CFD) methods which can be considered as the combination of in-process detections and post-process inspections, have recently been applied not only to understand the keyhole dynamics and porosity formation mechanism, but also to give a more comprehensive guidance for welding parameter optimization to achieve a better quality weld. Lin et al. (2017) simulated porosity formation during laser welding, the results revealed that porosity formation experiences three steps: keyhole-induced bubble formation, bubble floating in the molten pool and pore formation with the solidifying molten pool. Panwisawas et al. (2017) developed a physics-based model to simulate keyhole and porosity formation, and summarized the effect of the welding parameters on maximum size of pore in a normalized processing diagram. However, the mechanism of porosity formation has not been fully investigated. Bubble development induced by molten pool dynamics, especially the development from a small to a large when bubble is floating in the molten pool, have rarely been studied. In this paper, laser welding experiments with glasses were performed to directly observe the dynamics of molten pool and keyhole using high speed camera. Keyhole-induced bubble (KIB) formation and molten pool-induced bubble development were investigated aiming to reveal the mechanism of porosity formation.
3. Results and discussion
2. Experimental procedure
3.1. Bubble formation mechanism
The experiment setup is illustrated in Fig. 1. Laser welding experiments were conducted using a fiber laser (IPG YLS-4000) with a maximum output power of 4 kW, a wavelength of 1060 nm and a beam spot size of 0.35 mm in diameter. High speed imaging was performed on the quartz glass using a high speed camera (Phantom V611) together with a
Wang et al. (2012) reported that plasma during laser beam welding could be divided into keyhole plasma and floating plasma both with a fluctuation in size. Keyhole plasma, the size of which can be seen as the size of keyhole, is the main portion of plasma with the most brightness, whereas floating plasma is with little brightness above the surface of
Table 2 Chemical compositions of the adopted materials (wt-%). Element
SiO2
C
Si
Mn
P
S
Cr
Mo
Ni
316L Glass
– 99.97–99.99
0.021 –
0.77 –
1.019 –
0.039 –
0.001 –
16.92 –
2.03 –
12.16 –
721
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Fig. 2. Flow chart of image processing.
Fig. 3. (a) High speed images showing the process of keyhole formation and (b) keyhole depth evolution. The red arrow shows the laser welding direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4. Relationship between keyhole instability and porosity formation: (a) keyhole depth evolution of 3600 ms, (b) corresponding porosity in longitudinal section, (c–f) four typical 3D views of (b).
722
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Fig. 5. High speed images showing the formation processes of three bubbles. Note that the three pores of pore 1, pore 2 and pore 3 in Fig. 4(d) were formed by the three bubbles, respectively. The red arrow shows the laser welding direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 6. Schematic diagram showing bubble formation mechanism, (a–b) keyhole formation, (c) keyhole collapse and (d) bubble formation. Note that P v is vapor pressure in keyhole and P m is the pressure of molten pool.
Fig. 7. (a) High speed images showing longitudinal molten pool, (b) pool depth and (c) pool length of the 0.5 mm DLG welding using different welding parameters.
723
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Fig. 8. Bubble development and pore formation when using the parameters of 3000 W laser power and 1 m/min welding speed.
sample. Fig. 3(a) shows the process of keyhole formation during the 0.25 mm DLG welding with the parameters of 3000 W laser power and 1 m/min welding speed. Floating plasma was firstly formed at 0.5 ms due to the melting and evaporation of metal at the surface. Keyhole plasma as well as keyhole could be seen at 1.5 ms. Fig. 3(b) shows the keyhole depth evolution during the keyhole formation process. The depth was increasing linearly at an average speed of 0.356 mm/ms until 8.5 ms. The depth increased with a fluctuation after 8.5 ms and a quasisteady keyhole was formed at 30 ms in a depth of 4.524 mm. To investigate the keyhole dynamics during welding, a total of 7201 high speed images were calculated by image processing. The time interval of each image is 0.5 ms. Fig. 4(a) shows the keyhole depth evolution of 3600 ms during the welding. The keyhole was extremely unstable and the fluctuation changed with time. This instability is caused by the floating plasma, in which part of laser is absorbed by inverse bremsstrahlung according to Lacroix and Boudot (1998). The
Fig. 9. Schematic of bubble after development, where m denotes the number of merged bubble (MB) and n denotes the number of keyhole-induced bubble (KIB). Note that the merged bubble is composed of several KIB’s.
Fig. 10. Characterization of the porosity formed in Fig. 7: (a) pore height measurement, (b) 3D topography and (c) pore diameter extraction.
724
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
images showing the process of bubble formation were analyzed. Fig. 5 shows the formation process of three bubbles which are the initial states of the three pores in Fig. 4(d). The high speed images display that a bubble (KIB) is formed due to the collapse of keyhole wall when the fluctuation is large enough. Pang et al. (2014) simulated the formation of bubble and found the vapor pressure of keyhole decreased before the formation of bubble. Fig. 6 shows the schematic diagram of bubble formation mechanism. The process can be detailed as follow: in a keyhole collapse cycle, the depth of keyhole firstly increases shown as Fig. 6(a–b), accompanying with a decrease of Pv and an increase in the size of floating plasma (the vapor pressure in keyhole P v is larger than the pressure of molten pool Ps, Pv > Ps). The decreased Pv causes a shielding gas flow from the keyhole opening to the keyhole tip, suggesting that the keyhole is composed of metallic vapor and shielding gas. The increased floating plasma leads to a decrease in intensity of the incident laser. When the intensity reaches a certain limit as Fig. 6(c), the incident laser could not result enough metal vapor to maintain the keyhole and the keyhole begins to collapse (the vapor pressure in keyhole P v is smaller than the pressure of molten pool Ps, Pv < Ps). Note that the time taken for keyhole collapse is very short that the keyhole is cut into two parts by molten pool, e.g. three time of 0.5 ms, 3.5 ms and 4 ms can be observed according to the red dotted line in Fig. 5. The lower part is rapidly cooled and compressed by the surrounding molten pool. The cooling gives rise to the condensation of metallic vapor in the lower part and a spherical bubble composed of shielding gas is formed at the tip of keyhole and floats with thermal fluid flow shown as Fig. 6(d). Meanwhile, the lack of metal vapor lead to a decrease in the size of floating, which in turn give rise to an increase in intensity of the incident laser. When the floating plasma decrease to a certain limit, the keyhole collapse cycle will restart from Fig. 6(a)–(d). The glass softening caused by the high temperature keyhole had a bad effect on the visualization of the features during the in-situ welding experiment. This can be verified by the unclear images (Fig. 5), the longitudinal section which are coated with a yellow layer (Fig. 4(b)) and the rough 3D topographies (Fig. 4(c–f)). The interactions of the softened glass and molten metal would affect the formation of part pores and thus only the mechanism of bubble formation was investigated during the 0.25 mm DLG welding.
Table 4 Characterization results of the eight pores. No.
Height (μm)
Radius (μm)
H/R ratio
1 2 3 4 5 6 7 8
578 409.3 325.9 250.9 236.9 233.2 175.2 141.6
892.28 644.10 481.65 433.72 416.37 353.74 322.44 228.48
0.65 0.64 0.68 0.58 0.57 0.66 0.54 0.62
Fig. 11. (a) Longitudinal sections and (b) pore numbers in different size ranges of the 0.5 mm DLG welds using different welding parameters.
3.2. Bubble development
absorption has the similar variation rule with the size of floating plasma and thus gives rise to an unstable incident laser with fluctuated intensity during the whole welding process. The size of keyhole is dependent upon the intensity of the incident laser. Fig. 4(b) shows the porosity in the longitudinal section formed within the 3600 ms. In contrast to the keyhole depth evolution (Fig. 4(a)), the porosity formation is shown to be due to the keyhole instability. The pores seem to be formed when the fluctuation is large enough. The size and number of pores depend on the magnitude of fluctuation, i.e. only two small pores can be seen when the magnitudes are on the small side (Fig. 4(c) and Fig. 4(e)), whereas four or more large pores can be seen when the magnitudes are larger (Fig. 4(d) and (f)). To further understand the mechanism of porosity formation, the
To avoid the glass softening, the 0.5 mm DLG welding experiment was performed in this section. The molten pool-induced bubble development could be observed clearly. Due to the limited keyhole size, the keyhole could not be observed. Fig. 7(a) shows the longitudinal molten pools resulting from different welding parameters. The longitudinal view of molten pool can be clearly observed according to the solid/liquid interfaces from the images. The size of molten pool varies obviously with changes in laser power and welding speed such that the lower the power or higher the speed the smaller the size. To further analyze the variations, the pool depths and lengths of different parameters were measured as shown in Fig. 7(b) and (c). The depth variations caused by laser power and
Table 5 Pore analysis results. Laser power (w)
2400 2700 3000 3000 3000
Welding speed (m/min)
1.0 1.0 1.0 1.3 1.6
H/R ratio
0.54 0.58 0.63 0.55 0.56
No. of pores
Volume (μm3)
2.65 × 109 4.42 × 109 5.45 × 109 2.71 × 109 1.63 × 109
266 207 175 224 226
725
Porosity ratio (%)
6.12 6.81 7.38 5.45 4.00
Diameter (μm) Maximum
Mean
1309.69 1502.46 1784.57 1477.33 1079.15
288.42 316.56 366.01 285.66 249.30
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
of each pore, respectively.
welding speed are similar, in respective ranges from 4.48 mm to 3.77 mm and from 4.48 mm to 3.62 mm. There is a big difference between the length variations, in ranges from 12.40 mm to 10.19 mm and from 12.40 mm to 11.8 mm, respectively. The reason is that a longer region is fused within the same time when a higher speed is used. Fig. 8 shows the development processes of two bubbles when using the welding parameters of 3000 W power and 1 m/min speed. The transition from a bubble to a pore is extremely complex, including bubble floating with the thermal fluid flow, bubble merging and pore formation with the solidifying molten pool. The detailed processes could be described as follow: Step 1: A KIB has been formed in the middle part of weld pool front at 2293.5 ms. The KIB was the initial state of bubble 1. From 2360 ms to 2377.5 ms, the KIB merged with a small KIB floated from the bottom of molten pool. Step 2: Three small KIB’s have been formed at 2418 ms and then they merged with each other to be one. The merged bubble was moving to the bubble formed in step 1, forming a larger bubble at 2430.5 ms. Step 3: This step is similar with step 1 except that the KIB floated from the bottom of molten pool was larger. The development of bubble 1 have been completed in this step. Bubble 1 could be observed at 2470.5 ms. Step 4: The development of bubble 2 could be observed in this step. It is similar with step 2. In contrast to bubble 1, bubble 2 is much smaller in size. Step 5: Two pores were formed with the solidifying molten pool from 2782 ms to 2897.5 ms. The time taken for the two developments were 177 ms and 103.5 ms, respectively. The bubble after development is composed of n KIB’s and m merged bubbles as shown in Fig. 9. The development processes indicate that the transition from a small bubble to a large bubble is caused by molten pool dynamics, in which whether the KIB’s merge with each other or not depends on the speed and surface tension of the thermal fluid flow.
4. Discussion The current research demonstrates that the molten pool dynamics has a significant effect on bubble development. The dynamics is the direct result of laser power and welding speed. To further investigate the molten pool-induced bubble development, the porosity of the 0.5 mm DLG welds using five different parameters were analyzed. Fig. 11(a) shows the longitudinal sections of the five welds. Not only the number but also the size of pores varies with welding parameters. The porosity was qualitatively and quantitatively characterized using the method performed in Fig. 10. Pores with the diameters more than 100 μm were extracted by image processing and the weighted average of H/R ratio in each weld was calculated by Eq. (1). The results are shown in Fig. 11(b) and Table 5. Fig. 11(b) shows the pore numbers in different size ranges of different welding parameters. The variations can be clearly observed that the lower the power or higher the speed, the larger the pore numbers in the size ranges of 100–300 μm and 300–500 μm, but the smaller the pore numbers in the size ranges of 700–900 μm and 900–1800 μm. Table 5 shows that porosity characteristics such as pore number, pore diameter (maximum and mean), porosity volume and porosity ratio vary obviously with changes in laser power and welding speed. The variations on the number and the other characteristics are opposite, suggesting the degree of bubble merging is strongly associated with molten pool dynamics. In the weld using the 3000 W laser power and 1 m/min welding speed, a maximum porosity volume of 5.45 × 109 μm3, a maximum pore diameter of 1784.57 μm, but a minimum pore number of 175 were calculated. The higher laser power and lower welding speed used for the 0.5 mm DLG welds has deteriorated the porosity defect and has given rise to an increased number of large pores. This is because the energy density would be more and subsequently not only the occurrence of keyhole collapse is more frequent but also the molten pool is rather more unstable with an increased possibility for bubble merging to form big pores.
3.3. Porosity characterization 5. Conclusions The porosity formed in Fig. 8 were successfully characterized by super depth microscope and image processing. Both the diameters and heights of the pores were measured as shown in Fig. 10. The pores are mostly of spherical segmental in shape and there are 37 pores in total, with diameters ranging from 70.65 μm to 1784.57 μm. The heights, radiuses and height/radius (H/R) ratios of the 8 pores in Fig. 10(c) are tabulated in Table 4. The ratios maintain within a narrow range from 0.54 to 0.68, suggesting that the H/R ratios of pores formed using the same welding parameters are similar. To calculate the volume of each pore in the weld, a new method which is hypothesized that the pores formed using the same welding parameters have the same H/R ratio is proposed. In the case of the welding using the 3000 W power and 1 m/ min speed, the H/R ratio of each pore is hypothesized to be equal to the weighted average (m) of the eight H/R ratios in Table 4. The m is defined by Eq. (1). 8
m=
∑ i=1
1. Bubble formation and bubble development could be directly observed during laser welding experiments with glasses using high speed camera. The experiments could provide an advance in understanding the porosity formation. 2. Large fluctuation of keyhole was responsible for increased bubble formation and bubble merging resulted in the formation of large pore. 3. The characterization method detailed in this study could analyze the micrometer-scale porosity qualitatively and quantitatively. Pores characterized in the welds were mostly of spherical segment in shape, with diameters of 70.65–1784.57 μm and H/R ratios of 0.54–0.63. 4. Increased laser power and decreased welding speed gave rise to an increased possibility in the formation of large pore.
hi ri3 ri ∑8j = 1 r j 3
Acknowledgements This research is supported by the National Basic Research Program of China (973 Program, NO.2014CB046703), mega project of science research of Hubei province (2016AAA070), the National Natural Science Foundation of China (NSFC) (51421062), the Fundamental Research Funds for the Central Universities (HUST: 2016YXMS271).
where hi and ri (rj) are the heights and radiuses of the 8 pores, respectively. The magnitude of m is 0.63. The total porosity volume (V) of the weld is calculated as Eq. (2). n
V=
∑ k=1
n
=
∑ k=1
1 πhk (3rk 2 + hk 2) 6
References
1 πmrk 3 (3 + m2) 6
Berger, P., Hügel, H., Graf, T., 2011. Understanding pore formation in laser beam welding. Phys. Procedia 12, 241–247. Elmer, J.W., 2015. The Effect of Ar and N2 shielding gas on laser weld porosity in steel
where, n is the total number of pores, hk and rk are the height and radius 726
Journal of Materials Processing Tech. 252 (2018) 720–727
J. Xu et al.
Pang, S., Shao, X., Li, W., Chen, X., Gong, S., 2016. Dynamic characteristics and mechanisms of compressible metallic vapor plume behaviors in transient keyhole during deep penetration fiber laser welding. Appl. Phys. A Mater. Sci. Process. 122. Panwisawas, C., Perumal, B., Ward, R.M., Turner, N., Turner, R.P., Brooks, J.W., Basoalto, H.C., 2017. Keyhole formation and thermal fluid flow-induced porosity during laser fusion welding in titanium alloys. Exp. Modell. Acta Mater. 126, 251–263. Roushdy, M., 2007. Detecting coins with different radii based on hough transform in noisy and deformed image. Proc. GVIP J. 7, 25–29. Seto, N., Katayama, S., Matsunawa, A., 2000. High-speed simultaneous observation of plasma and keyhole behavior during high power CO2 laser welding: effect of shielding gas on porosity formation. J. Laser Appl. 12, 245. Wang, J., Wang, C., Meng, X., Hu, X., Yu, Y., Yu, S., 2012. Study on the periodic oscillation of plasma/vapour induced during high power fibre laser penetration welding. Opt. Laser Technol. 44, 67–70. Yu, Y., Wang, C., Hu, X., Wang, J., Yu, S., 2010. Porosity in fiber laser formation of 5A06 aluminum alloy. J. Mech. Sci. Technol. 24, 1077–1082. Zhang, C., Gao, M., Wang, D., Yin, J., Zeng, X., 2017. Relationship between pool characteristic and weld porosity in laser arc hybrid welding of AA6082 aluminum alloy. J. Mater. Process. Technol. 240, 217–222.
stainless steels, and nickel. Weld. J. 94, 313–325. Lacroix, D., Boudot, G.J., 1998. Spectroscopic characterization of laser-induced plasma created during welding with a pulsed Nd: YAG laser. J. Appl. Phys. 81 (10). Lin, R., Wang, H., Lu, F., Solomon, J., Carlson, B.E., 2017. Numerical study of keyhole dynamics and keyhole-induced porosity formation in remote laser welding of Al alloys. Int. J. Heat Mass Transf. 108, 244–256. Madison, J.D., Aagesen, L.K., 2012. Quantitative characterization of porosity in laser welds of stainless steel. Scr. Mater. 67, 783–786. Mazar Atabaki, M., Ma, J., Liu, W., Kovacevic, R., 2015. Pore formation and its mitigation during hybrid laser/arc welding of advanced high strength steel. Mater. Des. 67, 509–521. Norris, J., Robino, C., Hirschfeld, D.a., Perricone, M.J., 2011. Effects of laser parameters on porosity formation: investigating millimeter scale continuous wave Nd: YAG laser welds. Weld. J. 90, 198s–203s. Otsu, N., 1979. A tlreshold selection method from gray-level histograms. IEEE Trans. Syst. Man Cybern. 62–66. Pang, S., Chen, W., Wang, W., 2014. A quantitative model of keyhole instability induced porosity in laser welding of titanium alloy. Metall. Mater. Trans. A. 45 (6), 2808–2818.
727