IIW International Conference
High-Strength Materials - Challenges and Applications 2-3July 2015, Helsinki, Finland
IIW2015-15/….
Dropout formation in thick steel plates during laser welding
J. Frostevarg
1,a, T. Haeussermann
1,2,b1 Luleå University of Technology, Luleå, Sweden
2 University of Stuttgart, Faculty of Engineering Design, Production Engineering and Automotive Engineering, Stuttgart, Germany
ajan.frostevarg@ltu.se, bthilo.haeussermann@gmail.com
Abstract
Laser welding is a promising technique for welding thick metal sheets, which is usually used for achieving full penetration in a single weld pass. However, among the imperfections that can occur, dropout formation becomes an increasingly larger problem when the sheets to be welded get thicker. When the sheets are 15 mm or thicker it is a challenge to suppress and countermeasures to suppress dropouts get less impact. If full penetration is not achieved, the dropout formation cannot be formed and the melt flows to form better weld caps. Therefore a typical method when welding 15 mm and thicker is to use partial penetration from both sides of the joint and thereby achieving full penetration, at a cost of increasing plate handling complexities and time losses.
If the mechanisms behind the dropout formation can be understood, countermeasures may be developed and applied in order to be able to laser weld thicker plates with full penetration single pass welding. In order to understand the mechanics of the formation of dropout during full penetration welding in 15 mm thick plates, experiments have been conducted using laser hybrid arc welding (LAHW) using the laser in CW mode and also applying power modulation. During the experiments, the root has been observed with High Speed Imaging (HSI) to observe the mechanisms behind the formation. It is determined that the downward flow from the keyhole along with the surface tensional forces of the molten steel and its cooling rate play the most significant roles.
Keywords: Dropout, Thick sheet, Laser welding, Laser- arc hybrid welding, High speed imaging, Melt flow
1. Introduction
Laser beam welding (LBW) and laser-arc hybrid welding (LAHW) have been shown to give excellent results in many applications and materials. However, successful application these techniques can be held back by quality concerns, such as undercuts and varying weld root penetration [1]-[3]. Undercuts is a common quality problem that can be suppressed by various means [2].
Undercuts reduces the mechanical properties of welds and has a particularly severe effect on fatigue life [4]-[8].
Recently, a technique for reducing undercuts on the weld
cap was introduced; laser re-melting [9].
In thicker sheets, the melt inside the keyhole gets a downward thrust, pushing it out of the keyhole, forming humps on the root side of the weld [10], [11]. This is especially true for 1 µm (fiber-, disc-, YAG-) lasers since humps on the keyhole wall are pushed downwards due to evaporation of their top surfaces [11]. Sometimes the drops on the root are instead evenly distributed to form roothang instead. These imperfections, Fig. 1, especially root drops, will besides not being aesthetic also affect both mechanical and fatigue performances of the weld [12]. Dropouts act as both stress raisers and crack initiators. This downward thrust makes it rather difficult to find setup and parameter settings that give full penetration without ejection of the melt from the bottom of the weld [1]. If there is too much material going through the root to form droplets or roothang (continuous excess of material), the weld cap will also suffer from underfill. The underfill on the weld cap in Fig. 1b is continuous as the material flow downwards is even along the length of the weld.
The laser re-melting technique can also be applied to the root, but if the amount of dropout of material is too high, this technique is no longer feasible. A common technique for suppressing dropout formation is to increase the laser power, from the point where full penetration is achieved, shown in Figure 2 for 12 mm mild steel [13]. Dropout formation at lower laser powers decreases when the power is increased. If the power is further increased, spatter starts to form. But this method is limited when increasingly thick sheets are welded (t≥12 mm), probably due to increased mass flows and instabilities of the inner mechanics in the keyhole.
Figure 1. a) upside-down illustration of LAHW imperfection designations. b) cross-section of a droplet in a 15 mm steel plate
There is a need for being able to weld thick sheets with full penetration in a single pass. In order to develop techniques for suppressing dropout, the formation mechanics need to be understood. Excellent studies of the dropout (or droplet) formation have been performed [1], [13]-[15], but the mechanics are not yet fully understood. The models for dropout differ somewhat between those studies. Figure 3a shows mechanics of the formed dropout, including underfill on the weld cap.
When laser power is increased so that the keyhole exits at the bottom of the welded sheets, the melt is illustrated to go around the keyhole exit and then go up into the melt pool again [13], [14]. A more recent study, Fig. 4, using both simulation and observation takes into concern the thermal conductivity and ends up with a model that includes a longer region of melted material at the weld root and growing droplet [1] The creation of a droplet, is encouraged by a combination of the geometry of the melt/solid interface and surface tension which tends to favour the creation of hemispherical droplets.
Figure 2. Laser welding with different laser powers in 12 mm mild steel [13]
Figure 3. Model description of dropout formation when a) the keyhole does not penetrate fully and b) when the laser power is
increased so that the keyhole fully penetrates [13]
Figure 4. Droplet growth with melt/solid interface and surface tension [1]
In this study, the formation of droplets are further investigated by means of High Speed Imaging (HSI) [16]
applied at the root. The aim is to make an improved model for the dropout formation. The welds are made with LAHW by joining 15 mm thick sheets.
The laser power is also modulated (pulsed) in order to get more clues for dropout or root-hang formation.
Hypothetical benefits of modulating the laser power is also to suppress dropout formation by decreasing the average laser power, thereby also the downward push on the melt, while maintaining the same penetration depth and keeping the keyhole exit constantly open.
2. Experimental setup
The welds where produced with an LAHW setup, using a 15 kW Yb:fibre laser (IPG Laser GmbH, type YLR- 15000 (fibre core diameter: 400 µm, beam parameter product: 10.3 mm·mrad, wavelength: 1070 nm). The laser was focused 4 mm below the surface by 300 mm focal length optics to a spot size of 800 µm diameter (Rayleigh length ±4 mm) with an output power varied as listed in Table 1, using both the continuous and the pulsed wave mode. To achieve high enough average laser power, the peak pulse power was always 15kW and usually had longer pulse duration, while the low pulse power varied. To prevent back reflections damaging the optical fiber, a slight tilting (-7˚) of the laser was applied.
The GMA torch was applied in a tilted leading (30˚) position, with a 3 mm process interdistance with the laser. The GMA welding equipment used was a Fronius GMA power source TPS4000 VMT Remote used in Pulsed arc mode with a 1.2 mm diameter iron based filler wire. The shielding gas was Mison18 (82% Ar, 18% CO2), at a flow rate of 20 l/min applied by the weld torch. Table 2 shows the material properties for the base material and the filler wire. The 15 mm mild steel sheets were welded with a milled I-gap in butt joint
IIW International Conference
High-Strength Materials - Challenges and Applications 2-3July 2015, Helsinki, Finland
IIW2015-15/….
configuration, with a forced gap of 0.3 mm and oxides on the surface mildly grinded off. The experimental setup for welding and the setup of HSI are shown in Figs. 2-3 respectively. The HSI (including camera and illumination laser) is placed besides the sheets, looking at a mirror reflecting the light so that the mechanics for weld root formation can be observed.
Figure 2. Experimental setup where the metal sheets, the torch, and the high speed camera can be seen
Figure 3. Setup of high speed imaging to observe the weld root.
The illumination laser is in-line with the camera
3. Results
The resulting appearances of the joint after welding are shown by photographs in Fig. 4. When the CW mode is used (weld #1), the droplets are usually larger than when using the PW mode (welds #2-4). It appears that for a certain frequency, 100 Hz, the droplet formation nearly no longer occur. However, this comes with the serious drawback of large amounts of spatter. The excess material at the root is ejected instead of remaining at the root.
Figure 4. Photos of the resulting welds, cap and root for each Figure 5 shows a HSI frame with designations, from weld #1 in Table 1, using the CW mode. Sometimes due to internal mechanics in the keyhole, the material has momentarily higher momentum and is partly ejected and forms a “hose” from the process zone exit. This “hose”
is often pulled back due to surface tensional effects.
Sometimes the downward flowing material has enough Table 1. Parameters used for the experiments.
Weld settings Laser settings
Sample number Welding speed Average High power - duration Low power - duration Pulse frequency
[# (weld no)] (m/min) (kW) (kW) (ms) (kW) (ms) (Hz)
1 1.3 13 - - - - -
4 1.3 11.33 15 12 4 6 50
5 1.3 11.33 15 6 4 3 111
6 1.3 11.33 15 2 4 1 333
Table 2. Material contents for work pieces and filler wire in wt-%, balance Fe.
Name C Si (max) Mn P (max) S (max) Cr Mo Al
S355J+N 0.22 0.55 1.6 0.035 0.035 0.3 0.08 0.02
Lincoln SupraMIG Ultra (ER70S-6) 0.08 0.85 1.7 - - - - -
momentum to break free, forming a spatter droplet.
Usually, the surface tension seems to counteract gravitational forces and redirects the momentum of the melt. The material getting out at the end of the process zone forms a melt pool at the root. The melt flows and accumulates at the end of the melt pool. The accumulation of material continues until the melt flow channel has solidified, where the accumulated molten material does not get to grow further and is now a cooling droplet. If the solidification is fast enough roothang is formed instead. The keyhole exit and the molten rim around it are hereby referred to as the process zone exit.
Figure 5. High speed imaging frame when using the CW laser mode, showing droplet formation and important designations
Figure 6. High speed imaging frames using the PW laser mode with different frequencies; a) 50 Hz, b) 100 Hz and c) 333 Hz Figure 6 is also HSI frames, but showing frames from welds #2-4, using the PW mode. When the modulated laser is in the high power amplitude and the duration is long enough, the molten material seems to be momentarily accelerated downwards and sometimes has high enough momentum to break free from the surface tensional forces, forming spatter. When the frequency is comparatively low at 50 Hz, Fig. 6a the process is unstable and formation of droplets and spatter is uneven.
When the frequency is increased to 100 Hz, Fig. 6b, the material is more evenly momentarily accelerated and ejected as spatter. This ejection of material also discontinues the melt channel at the process zone exit.
Here, the droplet formation is different; sometimes the ejected material is caught by surface tensional forces and ends up as solidified spatter at the weld root. That the flow at the root can be interrupted indicates that the melted region behind the process zone exit does not extend more than some millimetres behind the keyhole opening and that the root melt pool is very shallow. The melt flow visible in Fig. 5 and 6a is held by surface tension and the redirected melt in the root melt pool seems to keep the original momentum from exiting the
process zone. When the laser frequency is further increased to 333 Hz, Fig. 6c, the behaviour gets more like the CW laser mode, but the droplets are smaller and there is slightly more spatter.
4. Discussion
4.1. Theory for dropout
There are a few aspects to take into account for determining the mechanics behind the root dropout formation. It is known from experience that thicker sheet welding gets increasingly difficult to weld without root imperfections. When the sheets get thicker, the melted material around the keyhole is increasingly accelerated downwards by the laser. Especially 1 µm lasers are believed accelerate the molten material downwards at the process exit than the 10 µm (CO2) lasers due to the difference in absorption angles of the different wavelengths [17]. It is also confirmed that CO2 laser welds are less susceptible for root dropout [13]. Figure 7 illustrates how a hump on the melt in the keyhole absorbs light, evaporates and drags the melt behind it downwards [11]. The angle of absorption and consequently angle of pressure depend on the laser wavelength, where 1 µm lasers promote a more downward pressure than 10 µm lasers. The downward momentum will continue as long as the laser hits the melt front. The difference in momentum of the melt through the exit caused by the different wavelengths was also observed by [13], [14]. A vital requirement for dropout or roothang formation is that the material gets through the process exit and forms a melt pool at the root. When looking at results from [13], [14] and HSI in Figs. 5-6, gravity do not seem to have a great impact for getting the melt through the process zone exit. The forces counteracting dropout needs to redirect the initial downward momentum, as illustrated in Fig. 8a. In the case of welding and having dropout or roothang, part of the melt is redirected and another part flows through the process zone exit, illustrated in Fig. 8b. If no material is redirected and all material is ejected, the resulting weld would more likely resemble a cut.
Figure 7. Illustration of a hump, accelerating the melt downwards by evaporation pressure when absorbing process laser light [11]
IIW International Conference
High-Strength Materials - Challenges and Applications 2-3July 2015, Helsinki, Finland
IIW2015-15/….
Figure 8. a) redirected melt flow at the process zone exit, producing a good weld. b) part of the melt flows through, forming
dropout 4.2. Model theory
The surface tension seems to play a crucial role in redirecting the material at the process zone exit, as well as keeping the melt flow at the root from falling down off the sheets due to gravity. The surface tensional forces redirecting the melt mass flow at the process zone exit largely depends on; i. the geometrical size of the zone (laser wavelength, shape and power, welding speed, sheet thickness), ii. material properties (viscosity, surface tension, conductivity, melting and evaporation temperatures), iii. ambient atmosphere (shielding gas, pressure). It is reasonable that the ambient pressure also adds pressure to redirect the flow, but the magnitudes of those effects are not yet investigated. In thick sheet welding, the downwards momentum and the process zone exit size gets larger, thereby worsening the conditions for melt flow redirection.
In the HSI from the experiments, there are some instances of droplets detaching from the ejected material, Fig. 7. These “hoses” pulled back due to surface tension are sometimes retracted so quickly that a portion of the material is left mid-air with no downward or upward momentum. Rough estimates of surface tension can be calculated by using measured values. Here, the surface tension seems to be a factor six stronger than gravity (9.4 mN and 1.6 mN respectively).
Figure 10a shows the mechanics acting at the weld process zone exit, including the direction of the initial mass flow. The surface tensional forces redirecting the downwards melt flow from the keyhole is believed to be stronger if the keyhole is open rather than if it is filled.
The surface tensional forces needs to be strong enough for the downwards melt flow to be redirected (upwards).
Simple equations for surface tension forces can be derived the two cases; filled keyhole and open keyhole, Figs. 10b-c respectively.
Figure 9. a) HSI with material ejection from the keyhole forming a
“hose”. b) detached droplet (spatter)
Figure 10. a) Illustration of the forces acting at the weld process zone exit. b) A geometrical description for surface tensional forces
when having a spherical droplet and c) a torus (ring) shaped droplet at the root process zone exit
Without taking into account for, ambient pressures, melt flows (e.g. marangoni effects) or keyhole pressures, a surface tensional force upwards is deduced for a static spherical exit to be according to Eq. (1) with geometrical designations shown in Fig. 10b. For a ring (torus or half- doughnut) shaped drop the force is deduced to be Eq. (2), with corresponding geometric designations shown in Fig. 10c. The surface tension σ depends on material properties, melt-to-surface interphases and surrounding atmosphere. It is known that oxidation of iron, forming iron oxides (Fe2O3, FeO, Fe3O4), lowers the viscosity and surface tension [18]. Different ambient atmospheres can prevent oxidation but also increase the surface tension of melted iron.
𝐹𝑐𝑙𝑜𝑠𝑒𝑑 𝑘𝑒𝑦ℎ𝑜𝑙𝑒 𝑒𝑥𝑖𝑡= 𝜋𝜎 sin 𝛼 𝑑𝑝 (1) 𝐹𝑜𝑝𝑒𝑛 𝑘𝑒𝑦ℎ𝑜𝑙𝑒 𝑒𝑥𝑖𝑡 = 𝜋𝜎�sin 𝛼 𝑑𝑝+ sin 𝛽 𝑑𝑘� (2) Based on only the static isolated surface tensional forces at the process exit in for a spherical exit, the highest values in Eq. (1) are achieved by having an angle of
°
=90
α and no force at all if α=0°.
It can be imagined that sphere on the root side grows when pressure is added in order for the surface tension to be stronger and still maintain equilibrium. Equations for determining the equilibrium angles are deduced to
Eq. (3) and Eq. (4) for the closed exit and the ring shaped exit, respectively.
𝛼𝑐𝑙𝑜𝑠𝑒𝑑 𝑘𝑒𝑦ℎ𝑜𝑙𝑒 𝑒𝑥𝑖𝑡= sin−1��𝜌𝑔ℎ + 𝑝𝑝�𝑑4𝜎𝑝� (3)
𝛼𝑜𝑝𝑒𝑛 𝑘𝑒𝑦ℎ𝑜𝑙𝑒 𝑒𝑥𝑖𝑡= sin−1��𝜌𝑔ℎ + 𝑝𝑝��𝑑𝑝4𝜎− 𝑑𝑘�� (4) To indicatively compare the two forms, the graph in Fig. 11 can be derived for increasing process zone exit sizes. For angle equilibrium the angle needs to be
°
<90
α , otherwise the surface tension cannot add any more force and hold back against the push from above and melt will get through. The following values where used for making the graphs; a h = 3mm high melt pillar above the process zone exit, melt density 𝜌 = 7.015 ∙ kg/m3 and surface tension 𝜎 = 1.5 N/m (torus shape angles are equal on both sides with an inner opening of 0.5 mm). When only gravity acts as downward force, the diameter of the exit can get quite wide. However this will only be true for an undisturbed hanging droplet. When additional force is added the equilibrium angle of °90 can be reached. Comparing these two forms, it seems that the open keyhole exit can sustain a wider process zone exit.
Figure 11. Angles needed for equilibrium of gravitational and added forces against surface tensional forces showing five cases; i.
gravity, ii. gravity + 500 n/m2, iii. gravity + 1000 n/m2, iv. gravity + 2000 n/m2, v. gravity + 4000 n/m2
When the laser power is sufficient for full penetration, it can be seen in Fig. 2 that the process exit is covered by a close-to-spherical “bubble” of flowing melt. This bubble exerts an upwards force strong enough to redirect the downward flowing melt. It increases in size when the laser power is increased because the surface tension acts to keep the form intact. When the force is too large, the bubble breaks and melt is ejected as spatter. In Figs. 5-6, the keyhole seems to vary between being open and closed. When having the keyhole open, the static surface tensional force depends on the angles to the surface, but also the thickness of the rim. If the thickness is small, higher angles should be attained and thereby higher forces are attained. In Fig. 2 the keyhole reaches the root
and the melt is thin and flows around the exit, almost spherically.
This concludes that the keyhole exit should be fully penetrated by the laser (preferably with an open keyhole exit) with a spherical molten rim. Thereby sufficiently high surface tensional forces are achieved to redirect the downward melt flow, preventing material to get out of the process exit at the root.
4.3. Guidelines for preventing dropout formation As root dropout needs to be prevented, the following guidelines are deduced:
• Use sufficient laser power for full penetration by the laser
o rather than letting heat conduction melt through to the exit
• Use sufficiently high welding speeds to prevent excessive process exit sizes
o e.g. high enough laser power at sufficiently high welding speed for a given material/thickness o a small gap could help the laser to fully
penetrate the sheets at higher sheet thickness
• Increase the upwards acting forces at the root o Prevent decrease of surface tension from
oxidation, by applying proper shielding gas at the root
o Increasing the ambient pressure at the root
• Decrease the velocity of the downward flow by e.g;
o using 10μm laser instead of 1μm laser
o decreasing gravitational effects by welding horizontally
o weld with a gap to decrease the laser power needed, lowering the downward “push” on the melt by the laser
5. Acknowledgements
The authors gratefully acknowledge funding by the European Commission; programme FP7-RFCS, project HYBRO, no. RFS-CR-12024.
6. Conclusions
Root dropout and roothang are imperfections that need to be suppressed. They can only form when the upward forces at the root are too low to redirect the momentum of the downward flow of melt around the keyhole. By following the presented guidelines and successfully increasing the upwards forces and decreasing the downward forces, successful welding in one pass with full penetration in thicker sheets may be achieved.
Laser power modulation can suppress imperfections at the weld root, but this is due to that the material is ejected at the root as spatter, rather than stabilizing the flow as intended.
A side effect discovered by laser power modulation is
IIW International Conference
High-Strength Materials - Challenges and Applications 2-3July 2015, Helsinki, Finland
IIW2015-15/….
that it seems to momentarily increase the push of the downwards flow, especially when using 1 µm wavelength lasers. This could be useful for remote laser cutting applications. Higher cutting efficiency and quality (less dross) might be achieved.
References
[1] Powell, J., Ilar, T., Frostevarg, J., Torkamany, M. J., Na, S., Petring, D., Zhang, L. and Kaplan, A. F., "Weld root instabilities in fiber laser welding," Journal of laser applications. 27, Issue S2, (2015) pp S29008.
[2] Frostevarg, J., Kaplan, A. F., "Undercuts in Laser Arc Hybrid Welding," Physics Procedia. 56, (2014) pp 663-672.
[3] Blug, A., Abt, F., Nicolosi, L., Heider, A., Weber, R., Carl, D., Höfler, H. and Tetzlaff, R., "The full penetration hole as a stochastic process: controlling penetration depth in keyhole laser- welding processes," Applied Physics B. 108, Issue 1, (2012) pp 97-107.
[4] Alam, M. M., Barsoum, Z., Jonsén, P., Kaplan, A. F. H. and Häggblad, H. Å, "The influence of surface geometry and topography on the fatigue cracking behaviour of laser hybrid welded eccentric fillet joints," Applied Surface Science. 256, Issue 6, (2010) pp 1936-1945.
[5] Bell, R., Vosikovsky, O. and Bain, S., "The significance of weld toe undercuts in the fatigue of steel plate T-joints," International Journal of Fatigue. 11, Issue 1, (1989) pp 3-11.
[6] Nguyen, N. T., Wahab, M. A., "The effect of undercut and residual stresses on fatigue behaviour of misaligned butt joints,"
Engineering Fracture Mechanics. 55, Issue 3, (1996) pp 453- 469.
[7] Nguyen, T. N., Wahab, M. A., "The effect of weld geometry and residual stresses on the fatigue of welded joints under combined loading," Journal of Materials Processing Technology. 77, Issue 1, (1998) pp 201-208.
[8] Otegui, J. L., Kerr, H. W., Burns, D. J. and Mohaupt, U. H.,
"Fatigue crack initiation from defects at weld toes in steel,"
International Journal of Pressure Vessels and Piping. 38, Issue 5, (1989) pp 385-417.
[9] Frostevarg, J., Torkamany, M. J., Powell, J. and Kaplan, A. F.,
"Improving weld quality by laser re-melting," Journal of laser applications. 26, Issue 4, (2014) pp 041502.
[10] Eriksson, I., Powell, J. and Kaplan, A., "Measurements of fluid flow on keyhole front during laser welding," Science and Technology of Welding and Joining. 16, Issue 7, (2011) pp 636- 641.
[11] Kaplan, A. F., "Local flashing events at the keyhole front in laser welding," Optics and Lasers in Engineering. 68, (2015) pp 35- [12] Alam, M., Karlsson, J. and Kaplan, A., "Generalising fatigue 41.
stress analysis of different laser weld geometries," Materials &
Design. 32, Issue 4, (2011) pp 1814-1823.
[13] Haug, P., Rominger, V., Speker, N., Weber, R., Graf, T., Weigl, M. and Schmidt, M., "Influence of laser wavelength on melt bath dynamics and resulting seam quality at welding of thick plates,"
Physics Procedia. 41, (2013) pp 49-58.
[14] Romminger, V., Haug, P., Speker, N. and Holzer, M., "High‐
power Full Penetration Welding Behavior," Laser Technik Journal. 10, Issue 3, (2013) pp 36-40.
[15] Ilar, T., Eriksson, I., Powell, J. and Kaplan, A., "Root humping in laser welding–an investigation based on high speed imaging,"
Physics Procedia. 39, (2012) pp 27-32.
[16] Eriksson, I., "High Speed Imaging Analysis of Laser Welding,"
PhD thesis, Luleå University of Technology. (2013).
[17] Kaplan, A. F., "Fresnel absorption of 1μm-and 10μm-laser beams at the keyhole wall during laser beam welding:
Comparison between smooth and wavy surfaces," Applied Surface Science. 258, Issue 8, (2012) pp 3354-3363.
[18] Keene, B., "Review of data for the surface tension of iron and its binary alloys," International Materials Reviews. 33, Issue 1, (1988) pp 1-37.
