Skip to main content
Article Open Access

Study on Dynamic Development of Three-dimensional Weld Pool Surface in Stationary GTAW

  • EMAIL logo , , , and
Published/Copyright: June 16, 2017
Download Article (PDF)
High Temperature Materials and Processes
From the journal High Temperature Materials and Processes Volume 37 Issue 5

Abstract

The weld pool contains abundant information about the welding process. In particular, the type of the weld pool surface shape, i. e., convex or concave, is determined by the weld penetration. To detect it, an innovative laser-vision-based sensing method is employed to observe the weld pool surface of the gas tungsten arc welding (GTAW). A low-power laser dots pattern is projected onto the entire weld pool surface. Its reflection is intercepted by a screen and captured by a camera. Then the dynamic development process of the weld pool surface can be detected. By observing and analyzing, the change of the reflected laser dots reflection pattern, for shape of the weld pool surface shape, was found to closely correlate to the penetration of weld pool in the welding process. A mathematical model was proposed to correlate the incident ray, reflected ray, screen and surface of weld pool based on structured laser specular reflection. The dynamic variation of the weld pool surface and its corresponding dots laser pattern were simulated and analyzed. By combining the experimental data and the mathematical analysis, the results show that the pattern of the reflected laser dots pattern is closely correlated to the development of weld pool, such as the weld penetration. The concavity of the pool surface was found to increase rapidly after the surface shape was changed from convex to concave during the stationary GTAW process.

Keywords: gas tungsten arc welding (GTAW); weld pool surface; laser-vision-based sensing; convex and concave; weld penetration

Introduction

The welding process is widely applied in current manufacturing industries, the trend in high productivity and high-quality welding has been toward process automation, the sensing and control of weld joint penetration is a key issue in the automated welding. Especially, in the case of the root pass welding in thick plate or fillet welding under variable load conditions in gas tungsten arc welding (GTAW) process. However, a particular problem is, it is very difficult to detect the penetration of weld pool in the backside of workpiece. Meanwhile, a feasible approach to obtain the situation of weld pool in topside is important to detect the weld pool penetration [1]. In recent years, different observations and measurement approaches have been developed, such as machine vision [2, 34], ultrasonic [5, 6], X-ray radiation [7, 8], pool oscillation [9, 10], and other methods [11, 12].

Among these methods, noncontact vision-based sensing methods were studied more extensively, they have evolved from two-dimensional (2D) to three-dimensional (3D) domain and are widely studied by many researchers. An important technique for a 2D weld pool boundary measurement is the coaxial viewing of the weld pool, which was first proposed by Richardson et al. [13]. Agapakis and Bolstad [14] presented an innovative vision sensing system that used intense stroboscopic illumination to overpower the arc light in the welding process and produce a clear image with synchronized camera. In the recent study, the 3D shape of weld pool is concerned. Wang [15] established a 3D weld pool sensing system. The results of weld pool reconstruction were presented by the image processing. Based on the reconstructed 3D weld pool geometry, the weld pool geometry was described by an artificial neural network (ANN) model. Zhang and his colleagues established a novel stereo camera system that used only one camera and a biprism placed in front of the lens, which was applied to capture virtual image pair of the weld pool surface, and reconstructed 3D shape of the weld pool surface during base current period of P-GMAW [16]. Also, some special image processing algorithms for welding process was developed to analyze and extract the information of weld pool, for the visual sensing of weld pool [17, 1819]. Although these studies can extract certain 2D geometry or 3D information from the weld pool by vision-base methods, the dynamic specular characteristics of the weld pool and the interference from a strong arc, complicate the observation and image processing and deteriorate the effectiveness of these methods. Due to the poor welding conditions, some methods were proposed, which could get accuracy geometry of the surface of weld pool. Ai Xiaopu [20, 21] and his colleagues proposed a novel method which based on stripes reflection. The phase that changes the phase data was analyzed by comparing the phase data extracted from the deformed grating image and the reference grating image. The 3D surface shape of TIG weld pool was obtained, and achieved satisfying results. In order to avoid arc light interference in the measurement process, Zhang Yuming and his coworkers proposed a novel structured laser machine vision system, which takes advantage of the specular reflected, for reconstructing 3D weld pool surface. In the experiments, a structured laser projected on the weld pool surface, and reflected to the imaging plate. Then the structured laser images on the plate were captured by camera. The reconstructed results proved the effectiveness of the system, and the acquired weld pool was more accurate [22, 23, 2425]. By the same way, our research group members have also done a lot of research about the weld pool surface in GTAW by Laser-Vision-Based, in previous study [9, 26, 27].

In this work, we proposed an improved system to observe the weld pool surface. The experimental process and its corresponding dots laser pattern were captured by CCD. The converting of dots can display the development of the weld pool surface in the welding process. Further, a mathematical model, that contains the incident ray, reflected ray and screen, was built based on the structured laser reflection imaging system. The dynamic variation of weld pool surface and its corresponding dots laser pattern were simulated and analyzed by the Matlab software. It is found that the weld pool surface is like a Gaussian surface. The result of the experiment and analysis is applied to further understanding the dynamic variation process of the TIG welding weld pool surface.

Experiments

The principle of observation system

The principle of observation system is shown in Figure 1. The dots laser pattern was projected onto the weld pool surface. Its reflection that reflected from the specular weld pool surface was intercepted by a screen. Then the deformed dots laser patterns on screen, which were hereafter referred to as reflected patterns, were captured by a CCD camera. The system of position parameters can be determined in the world coordinate system. And the surface reconstruction scheme was proposed to obtain the 3D surface based on the reflection law.

Figure 1: The experimental principle of measurement using dots laser vision.
Figure 1:

The experimental principle of measurement using dots laser vision.

The real experimental measurement system is shown in Figure 2. The weld pool surface observation system of the TIG welding mainly consists of laser vision system and structure laser reflected imaging system. The laser vision system is mainly composed of CCD camera. The CCD camera is fitted with a 20 nm bass-pass filter, centered at 650 nm, to reduce the interference of arc light. The structure laser reflected imaging system mainly includes laser generator and screen. The laser generator was placed on YOZ plane, and the angle with the horizontal of workpiece was 30°. The adopted parameters of laser generator were 12 V, 250 mW, 650 nm. The perpendicular distance between the fixed center of laser and the workpiece was 45 mm, and the horizontal distance between the fixed center of laser and the welding torch was 75 mm. The welding torch was fixed and perpendicular to workpiece. The distance between welding torch and screen was 160 mm. The screen, with the size of 47 cm×29 cm, was placed parallel to the XOZ plane.

Figure 2: Experimental measurement system.
Figure 2:

Experimental measurement system.

Experimental results

In order to reconstruct the dynamic development of TIG welding weld pool surface, the 304 stainless steel plate with the size of 250 mm×100 mm×3 mm, was used to do stationary TIG welding experiment. The welding current of TIG welding experiment was 50A, 60A, 70A, 80A, respectively. The flow rate of Argon gas was 10 L/min. The dots laser reflected from the dynamic weld pool was captured by a CCD camera.

Taking welding current is 70A as an example to elaborate the dynamic development of weld pool. The whole process of initial stage to weld pool depression is illustrated in Figure 3. During the whole welding process, arc starting acts as the time initial point, which assumes the time initial point is t=0. At the welding initial stage, as a result of the small weld pool area, dots laser reflected from weld pool was less as shown in Figure 3(a) and 3(b). From Figure 3(c)–(e), it can be seen that the dots laser reflected from weld pool was increased, and the dots laser was dense with the increased duration of welding process. Due to the laser dots pattern was steady, it indicated that the weld pool was convex based on its pattern. The dots laser began to gather in the middle of weld pool as shown in Figure 3(f)–(j). It shows that the weld pool transits from convex to concave slowly. The laser dots almost gather together, which shows that the weld pool surface is depressed heavily. The results show that, with a certain welding current and welding abidance, the weld pool begins to be depressed. Then, the weld pool surface was altered from convex to concave and its depth was increased. Combining Figures 2 and 3, it shows, during the welding process, the variations of reflected laser dots pattern can represent the dynamic development of weld pool. The simulated results of dots laser pattern accord with the experiment very well. Further, it shows that the weld pool surface changes from convex to concave are at a certain welding current. The dots laser pattern on screen can characterize the convex and concave of weld pool surface, based on the structured laser machine vision.

Figure 3: The dots laser pattern change of the weld pool surface at the welding current 70A. (a) t=2.37 s. (b) t=3.57 s. (c) t=7.18 s. (d) t=13.72 s. (e) t=14.77 s. (f) t=16.89 s. (g) t=17.00 s. (h) t=17.80s. (i) t=18.1 s. (j) t=18.74 s. (k) t=19.77 s. (l) t=23.64 s.
Figure 3:

The dots laser pattern change of the weld pool surface at the welding current 70A. (a) t=2.37 s. (b) t=3.57 s. (c) t=7.18 s. (d) t=13.72 s. (e) t=14.77 s. (f) t=16.89 s. (g) t=17.00 s. (h) t=17.80s. (i) t=18.1 s. (j) t=18.74 s. (k) t=19.77 s. (l) t=23.64 s.

The laser dots projected onto workpiece and formed the dots laser pattern after welding as shown in Figure 4(a). It is shown that the row curvature of dots laser is different with that onto the crude workpiece and the weld pool is concave. The weld penetration maybe is partial penetration or complete penetration. Then combining with Figure 4(b), the cross-section macroscopic feature of weld pool, it shows, under the welding current of 70 A, the weld pool is complete penetration and the depression weld pool is heavily.

Figure 4: Weld pool appearance at the welding current of 70 A. (a) The front feature of weld pool. (b) The cross-section macroscopic feature of weld pool.
Figure 4:

Weld pool appearance at the welding current of 70 A. (a) The front feature of weld pool. (b) The cross-section macroscopic feature of weld pool.

Under different welding currents, the cross section of macroscopic feature was different as shown in Figure 8. The depth of the depression of weld pool was different at different welding currents, which represented different penetration depth and width. From Figure 5(a), the macroscopic feature of weld pool, it is seen that weld pool was convex at the welding current of 50 A. While welding current fixed to 60 A, the weld pool began to be depressed. But the depression has a low depth as shown in Figure 5(b). With the increase of current, the depth and width of weld pool and the amplitude of surface fluctuation were increased as shown in Figure 5(c) and 5(d). When welding current was too high, the workpiece maybe penetrated.

Figure 5: The macroscopic feature of weld pool (at different current (a) 50 A; (b) 60 A; (c) 70 A; (d) 80 A).
Figure 5:

The macroscopic feature of weld pool (at different current (a) 50 A; (b) 60 A; (c) 70 A; (d) 80 A).

Figure 6: Illustration for reflection model.
Figure 6:

Illustration for reflection model.

Figure 7: Simulation results for surface and incident rays at r=5.8. (a) Convex. (b) Concave. (c) Incident rays.
Figure 7:

Simulation results for surface and incident rays at r=5.8. (a) Convex. (b) Concave. (c) Incident rays.

Figure 8: The dynamic development of weld pool surface. (a) r=5.8. (b) r=6.4. (c) r=7.6. (d) r=8.8. (e) r=7. (f) r=6.2. (g) r=5.8.
Figure 8:

The dynamic development of weld pool surface. (a) r=5.8. (b) r=6.4. (c) r=7.6. (d) r=8.8. (e) r=7. (f) r=6.2. (g) r=5.8.

During the welding process, the time of weld pool changing from convex to concave is different with the change of welding parameters, such as welding current. The detail parameters show in Table 1.

Table 1:

The time of weld pool surface change from convex to concave at different currents.

Experiment NO.Welding current I/(A)Time t/(s)
150Always convex
26072.31
37023.64
48015.55

Discussion

Dynamic model of weld pool surface

To reconstruct the dynamic development of the convex and concave of TIG welding weld pool surface, the mathematical model was established to make the dots laser reflect on weld pool surface. The dots laser projected onto the Gaussian surface and its corresponding reflected ray were intercepted by screen. The reflected model is shown in Figure 6. The Matlab software was applied in the simulation and analysis of weld pool surface changing process.

The main analysis procedures are listed below:

Step 1: In order to ascertain the weld pool surface model, Gaussian surface is used as the weld pool surface. The convex and concave are presented in Figure 7(a) and 7(b), respectively. The mathematical equation of Gaussian convex and concave can be expressed as:

(1){G1(x1,y1,z1)=z1+Hexp[−(x1−a)2+(y1−b)24πr4+c],convexG2(x2,y2,z2)=z2−Hexp[−(x2−a)2+(y2−b)24πr4−c],concave

where H is the surface height coefficient, r the radio of Gaussian surface, x∈[−r2,r2]，y∈[b−r2,b+r2],z∈(e,f), (a, b, c) is the center of Gaussian surface. And a, b, c, e, f are constants.

Step 2: Compute the incident rays

In the second step, the direction vector of the incident ray in the world coordinate can be computed by the two points, which are introduced and calibrated on the incident ray. It was Ai,jxA,yA,zA, and Bi,jxB,yB,zB, respectively, as shown in Figure 1. I⃗ and Ii,j are described as eqs (2) and 3, respectively. The 9×41 rays act as incident rays of the simulation process, were determined based on incident rays relationship. The incident rays are showed in Figure 7(c).

(2)I⃗=Ai,jBi,j→Ai,jBi,j
(3)Ii,j=Ai,j+I⃗t1,t1〉0

where Ii,j represents any point on the incident ray in the world coordinate, and t1 is the time step.

Combining eqs 1 and (3), the coordinate of projected point, Oi,j(x0,y0,z0), can be solved.

Step 3: It is easy to obtain the surface normal at Oi,j(xo, yo, zo), denoted as n⃗i,j. The normal line n⃗i,j of projected point should be deduced, depending on the weld pool surface projected point Oi,j. And this point on the assumed surface G(x, y, z) of weld pool in first and second step. The n⃗i,j can be computed as eq. (4).

(4)n⃗i,j=∂G(xo,yo,zo)∂xo,∂G(xo,yo,zo)∂yo,∂G(xo,yo,zo)∂zo
(5)N⃗=n⃗i,jni,j

At the above of reflection model, relationship between incident vector, reflection vector and normal vector can be expressed as

(6)R⃗=I⃗+2N⃗

where R⃗ denotes the reflection vector, I⃗ the incident vector, N⃗ the normal vector. R⃗, I⃗ and N⃗ are all unit vectors.

The expression of reflected laser points on the reflected ray can be expressed using Oi,j and R⃗ as

(7)Ri,j=Oi,j+R⃗t2,t2〉0

where t2 is the time step, Ri,j is any point on the reflected way in the world coordinate.

Step 4: The screen was placed parallel to the XOZ plane, expressed as eq. (8).

(8)y−80=0

Combining eqs (7) and 8, the points where reflection occurs on the imaging plane, is defined as Ci,j(xc, yc, zc), and can be computed.

Analysis and verification

Based on above description, the determined mathematical relationship shows the convex and concave of surface. Its corresponding reflected rays were intercepted by screen, and formed corresponding dots laser pattern. The convex weld pool surface and its corresponding dots laser pattern are shown in Figure 8(a)–(c). With the height of convex decreased, r was changed from 5.8 to 7.6, respectively, the line curvature became smaller. Then, with the depression of the weld pool surface increased, the reflected ray was gathering together gradually. The results indicated that the weld pool surface was concave, and r was changed from 8.8 to 5.8, respectively. The results are shown in Figure 8(d)–(g). From Figure 8, it is seen that the different convex and concave of weld pool and its corresponding reflected laser dots pattern are different changing with the weld pool surface. It is also revealed that the dots laser pattern can represent the convex and concave of weld pool surface. However, there is also some deficiency in this system, the welding process is stationary which is not applicable for industrialization. The relationship between the weld pool surface depression and dots laser pattern should be verified by means of experiment to further analyze.

Conclusion

  1. During the stationary TIG welding process, the convex and concave of weld pool surface were closely correlated to welding current and welding time. While current in low range has negligible effects on surface depression, the weld pool was always convex. With the higher current and sufficiently long welding time, the weld pool changes from convex to concave.

  2. Deep penetration was mainly produced by high current, and sufficient welding duration was needed for deep penetration. The pool geometry was also greatly influenced.

  3. The weld pool depression value was potentially useful for controlling weld penetration. The weld pool depression was correlated with dots laser pattern. The simulated dots laser variation process was in agreement with the experimental results.

Funding statement: This study was financially supported by the National Natural Science Foundation of China (No. 51465031).

References

[1] X. Wang, Int. J. Adv. Manuf Technol., 72 (2014) 333–345.10.1007/s00170-014-5659-0Search in Google Scholar

[2] Y.M. Zhang, R. Kovacevic and L. Li, IEEE Trans. Control Syst. Technol., 4 (1996) 394–403.10.1109/87.508887Search in Google Scholar

[3] C.S. Wu, J.Q. Gao, X.F. Liu and Y.H. Zhao, Proc. Inst. Mech. Eng., Part B, 217 (2003) 879–882.10.1243/09544050360673279Search in Google Scholar

[4] G. Saeed, Int. J. Model. Ident. Control, 1 (2006) 84–93.10.1504/IJMIC.2006.010103Search in Google Scholar

[5] D. Hardt and J. Katz, Weld. J., 63 (1984) 273–281.10.1016/0022-3093(84)90406-XSearch in Google Scholar

[6] N. Carlson and J. Johnson, Weld. J. (Miami, FL, U. S.), 67 (1988) 239–246.Search in Google Scholar

[7] A. Guu and S. Rokhlin, Mater. Eval., 47 (1989) 1204–1210.Search in Google Scholar

[8] S. Rokhlin and A. Guu, Weld. J (Miami, FL, U. S.), 69 (1990) 83–97.Search in Google Scholar

[9] Y. Shi, G. Zhang, X.J. Ma, Y.F. Gu, J.K. Huang and D. Fan, Weld. J., 94 (2015) 176–187.Search in Google Scholar

[10] Y.H. Xiao and G. Den Ouden, Weld. J., 69 (1990) 289.Search in Google Scholar

[11] S. Chokkalingham, M. Vasudevan, S. Sudarsan and N. Chandrasekhar, Insight, 54 (2012) 272–277.10.1784/insi.2012.54.5.272Search in Google Scholar

[12] Y.M. Zhang, R. Kovacevic and S. Ruan, J. Eng. Ind., 117 (1995) 210–222.10.1115/1.2803297Search in Google Scholar

[13] R.W. Richardson and D.A. Gutow, Weld. J., 63 (1984) 43–50.Search in Google Scholar

[14] J.E. Agapakis and J.O. Bolstad, International society for optics and photonics, in Fibers’ 91, Boston, MA (1991 32–38.10.1117/12.25346Search in Google Scholar

[15] X.W. Wang, Proceedings of 2013 Chinese intelligent automation conference. Springer Berlin Heidelberg, (2013 741–748.10.1007/978-3-642-38460-8_82Search in Google Scholar

[16] Y.D. Zhang, Z.M. Liang, D.L. Wang and J. Wang, Trans. Tech. Publ., 236 (2012) 469–473.10.4028/www.scientific.net/AMM.236-237.469Search in Google Scholar

[17] G.J. Zhang, Z.H. Yan and L. Wu, Meas. Sci. Technol., 17 (2006) 1877.10.1088/0957-0233/17/7/028Search in Google Scholar

[18] J.J. Wang, T. Lin and S.B. Chen, Int. J. Adv. Manuf. Tech., 26 (2005) 219–227.10.1007/s00170-003-1548-7Search in Google Scholar

[19] C. Fan, F. Lv and S. Chen, Int. J. Adv. Manuf. Tech., 42 (2009) 126–137.10.1007/s00170-008-1587-1Search in Google Scholar

[20] X. Ai, N. Liu, Y. Wei, X. Hu, S. Wei and X.R. Liu, Int. Soc. Opt. Photonics, 7506 (2009) 750628.10.1117/12.838387Search in Google Scholar

[21] Y. Wei, N. Liu, X. Hu and X. Ai, Opt. Eng., 50 (2011) 057209–057209 –6.10.1117/1.3579519Search in Google Scholar

[22] W.J. Zhang, Y.K. Liu, X. Wang and Y.M. Zhang, Weld. J., 91 (2012) 195–203.Search in Google Scholar

[23] X.J. Ma and Y.M. Zhang, J. Manuf. Sci. Eng., 135 (2013) 021002.10.1115/1.4023374Search in Google Scholar

[24] H.S. Song and Y.M. Zhang, Weld. J., 87 (2008) 85.10.1002/bate.200890005Search in Google Scholar

[25] Z.J. Wang, G.J. Zhang, Y.M. Zhang and L. Wu, Chin. J. Mech. Eng., 40 (2008) 300–303.10.3901/JME.2008.10.300Search in Google Scholar

[26] G. Zhang, Y. Shi, C.K. Li, J.K. Huang and D. Fan, Acta Metall. Sin., 50 (2014) 995–1002.Search in Google Scholar

[27] G. Zhang, Y. Shi, J.K. Huang, W.J. Zhang and D. Fan, Chin. J. Mech. Eng., 50 (2014) 10–16.Search in Google Scholar

Received: 2016-12-8
Accepted: 2017-3-26
Published Online: 2017-6-16
Published in Print: 2018-4-25

© 2018 Walter de Gruyter GmbH, Berlin/Boston

This article is distributed under the terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Articles in the same Issue

  1. Frontmatter
  2. Research Article
  3. Effect of Trailing Intensive Cooling on Residual Stress and Welding Distortion of Friction Stir Welded 2060 Al-Li Alloy
  4. Short Communication
  5. Study on the Growth Mechanism of K2Ti4O9 Crystal
  6. Research Articles
  7. Artificial Neural Network-Based Three-dimensional Continuous Response Relationship Construction of 3Cr20Ni10W2 Heat-Resisting Alloy and Its Application in Finite Element Simulation
  8. Influence of Thermal Ageing on Microstructure and Tensile Properties of P92 Steel
  9. A Novel Process for Joining Ti Alloy and Al Alloy using Two-Stage Sintering Powder Metallurgy
  10. Modeling and Finite Element Analysis for the Dynamic Recrystallization Behavior of Ti-5Al-5Mo-5V-3Cr-1Zr Near β Titanium Alloy During Hot Deformation
  11. Study on Dynamic Development of Three-dimensional Weld Pool Surface in Stationary GTAW
  12. Influence of Heat Treatment on Fracture Toughness and Wear Resistance of Nicral-Zro2 Multilayered Thermal Barrier Coating
  13. Kinetic Study on Phosphate Enrichment Behavior in CaO–SiO2–FeO–Fe2O3–P2O5 Steelmaking Slags
  14. Effect of Prestrain on Precipitation Behaviors of Ti-2.5Cu Alloy
  15. Study on Gamma Prime and Carbides of Alloy A286 by Traditional Thermodynamic Calculation
https://doi.org/10.1515/htmp-2016-0224
Keywords for this article
gas tungsten arc welding (GTAW); weld pool surface; laser-vision-based sensing; convex and concave; weld penetration
Creative Commons
BY-NC-ND 3.0

Articles in the same Issue

  1. Frontmatter
  2. Research Article
  3. Effect of Trailing Intensive Cooling on Residual Stress and Welding Distortion of Friction Stir Welded 2060 Al-Li Alloy
  4. Short Communication
  5. Study on the Growth Mechanism of K2Ti4O9 Crystal
  6. Research Articles
  7. Artificial Neural Network-Based Three-dimensional Continuous Response Relationship Construction of 3Cr20Ni10W2 Heat-Resisting Alloy and Its Application in Finite Element Simulation
  8. Influence of Thermal Ageing on Microstructure and Tensile Properties of P92 Steel
  9. A Novel Process for Joining Ti Alloy and Al Alloy using Two-Stage Sintering Powder Metallurgy
  10. Modeling and Finite Element Analysis for the Dynamic Recrystallization Behavior of Ti-5Al-5Mo-5V-3Cr-1Zr Near β Titanium Alloy During Hot Deformation
  11. Study on Dynamic Development of Three-dimensional Weld Pool Surface in Stationary GTAW
  12. Influence of Heat Treatment on Fracture Toughness and Wear Resistance of Nicral-Zro2 Multilayered Thermal Barrier Coating
  13. Kinetic Study on Phosphate Enrichment Behavior in CaO–SiO2–FeO–Fe2O3–P2O5 Steelmaking Slags
  14. Effect of Prestrain on Precipitation Behaviors of Ti-2.5Cu Alloy
  15. Study on Gamma Prime and Carbides of Alloy A286 by Traditional Thermodynamic Calculation
Downloaded on 24.4.2026 from https://www.degruyterbrill.com/document/doi/10.1515/htmp-2016-0224/html?lang=en
Scroll to top button