Factors affecting weld root morphology in laser keyhole welding

Optics and Lasers in Engineering 101 (2018) 89–98

Contents lists available at ScienceDirect

Optics and Lasers in Engineering journal homepage: www.elsevier.com/locate/optlaseng

Factors affecting weld root morphology in laser keyhole welding Jan Frostevarg Luleå University of Technology, Div. of Product and Production Development Luleå SE - 971 87, Sweden

a r t i c l e

i n f o

Keywords: Laser welding Hybrid welding Weld root Humping Weld quality

a b s t r a c t Welding production efficiency is usually optimised if full penetration can be achieved in a single pass. Techniques such as electron and laser beam welding offer deep high speed keyhole welding, especially since multi-kilowatt lasers became available. However, there are limitations for these techniques when considering weld imperfections such as weld cap undercuts, interior porosity or humps at the root. The thickness of sheets during full penetration welding is practically limited by these root humps. The mechanisms behind root morphology formation are not yet satisfactory understood. In this paper root humping is studied by reviewing previous studies and findings and also by sample examination and process observation by high speed imaging. Different process regimes governing root quality are presented, categorized and explained. Even though this study mainly covers laser beam and laser arc hybrid welding, the presented findings can generally be applied full penetration welding in medium to thick sheets, especially the discussion of surface tension effects. As a final result of this analysis, a map of methods to optimise weld root topology is presented. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction Welding is used in a wide range of industries for joining metallic components [1–3]. For production purposes, single pass welding is often desired, which can be accomplished for thin sheets by a variety of methods without concerns about root quality. For thicker sheets, high power Electron Beam Welding (EBW), Laser Beam Welding (LBW) and Laser-Arc Hybrid Welding (LAHW) [4–9] can be applied, offering deep penetration depths at high welding speeds in a single pass. These techniques offer lower heat inputs and faster cooling rates than traditional arc welding, which often requires multiple pass welding [1,3,10]. 1.1. Weld humps Deep penetration welding processes can give rise to quality problems [11], such as the imperfections illustrated in Fig. 1(a), including; undercut [12], inhomogeneous material mixing [13] (which can depend on process orientation [14,15] and the presence of a gap [13]), porosity [6,16,17] and centreline cracking [18]. For single pass full penetration in sheets thicker than 10–12 mm [19], LAHW is often associated with excessive root penetration, with corresponding limitations stated in the Standard ISO 12932:2013 [20]. This imperfection can be divided into continuous root sagging or the intermittent [21] formation of droplets, known as root humping [6,22,23] as shown in Fig. 1(b) and (c). This particular imperfection has also been called dropping [24], dropout [25], drop through [26], burn through hump formation [27] or chain of pearls [28], but root humping is the most common term. Excessive root

penetration is associated with weld cap underfill [6] (material flows to the root) and root humping is also linked to porosity [24,29] and lack of fusion [21] which reduce the fatigue life of the welded component [30,31]. Root sagging on the other hand, is scarcely mentioned in the literature, and is considered a less severe problem. Melt humping at the root is believed to have similarities to weld bead humping on the weld cap, which is normally associated with high welding speeds. Based on the conservation of mass, Berger et al. [32] investigated humping for both wide/shallow and narrow/deep welds. They observed that a thin melt pool that moves at high speed is sensitive to rapid solidification, potentially choking the melt flow and consequently initiating swelling of the melt. As the heat source moves away from the growing hump, the melt pool extends and can again get choked by melt pool solidification, creating a new swelling closer to the heat source. It was also observed that humps tend to form when the melt stream velocity far exceeds the welding speed. Surface tension is also important and increased oxygen content and melt speeds drastically increase the tendency for humping. Although hump formation at the root has similarities with hump formation on the weld cap, internal weld melt flows and gravitational forces have a large effect on root humping. Blecher et al. [29] have surveyed occurrences of root humping for LBW and found that in almost all cases the process was due to having too high a heat input. Though not included in the survey, it was predicted that if LAHW was used the additional heat source could increase the chances of root humping.

E-mail address: [email protected] https://doi.org/10.1016/j.optlaseng.2017.10.005 Received 26 May 2017; Received in revised form 19 September 2017; Accepted 9 October 2017 0143-8166/© 2017 Elsevier Ltd. All rights reserved.

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

Fig. 1. LAHW (a) upside down illustration showing root imperfections. (b) cross section and (c) weld cap and root appearance [25]. (d) shows good root quality.

1.2. Observations of root humping

Haug et al. [35] also studied the influence of laser wavelength, (1 μm and 10 μm) on deep penetration welding. They found that process behaviour differs, as does the resulting robustness and seam quality. The 1 μm wavelength laser has a narrower process window in terms of power output and weld speed, but the 10 μm wavelength laser generates more spatter. Pan et al. [22] used X-ray photography with tracers of tungsten for one case when welding with and one case without hump formation. When having root hump formation, two time-series are shown. In the first, the hump forms through an opening which is not directly beneath the keyhole, as also described by Haug et al. [35]. In the second time series, the tracer exits the keyhole at the rear and is moved by a flow along the root, below the sheet bottom edge, to the already formed hump, matching the observations of Ilar et al. [21]. When there are no hump formation, the tracer follows a flow just above the sheet bottom edge and later moves back to the rear of the keyhole in a backwards current. HSI was also made by the present author, observing root humping [25]. As shown in Fig. 2, melt is shown to be pushed out near the keyhole and then flows beneath the sheet bottom edge. It was shown that surface tension can keep large volume droplets from escaping the melt and that a new hump will start to form close to the process exit after the melt flow to the previous one has solidified.

Studies of root humping have been made using High Speed Imaging (HSI) [33] and X-ray technologies. For LBW, Ilar et al. [21] studied the formation of root humps and found no correlation between hump size and distance, indicating instabilities in supply of melt flow to the root, which can be explained by keyhole instabilities. They also studied the flow dynamics of the weld cap and root simultaneously [23]. It was found that the top structure does not correlate with the appearance of the root and that the melt pool is wider at the top than at the root. Hump formation was found to be initiated at the end of the long melt pool tail, similar to the formation of humps on the weld cap. These findings are explained in more detail by Powell et al. [34]. Process windows for root humping have been found for LBW by Haug et al. [35] and for LAHW by Petring et al. [26] and Pan et al. [22]. For both processes, there are power input regimes categorized in order of increasing power as: (1). Insufficient Penetration (IP), (2). Root Humping (RH), (3). Good Result (GR) and (4). Over Penetration (OP), with root humping, occasional spatter and open pores at the root (a.k.a. root concavity or Shrinkage groove [20]). Haug et al. [35] studied hump formation and process stability in the RH regime using a 1 μm laser on 12 mm steel. They found that the humps form when full penetration is not achieved directly by the laser but by conductive heating just behind the keyhole, and melt flows out to form the droplets. Additionally, the penetration depth of the keyhole was found to vary. Weld stability and intermittent penetration by 1 μm lasers has also been studied by others, e.g. using HSI [36]. Ohnishi et al. [27] looked at LBW with hot and cold wire with different shielding gas setups. They found that if the shielding gas is too effective, the oxygen content in the melt is reduced which leads to lower viscosity and less penetration, causing humps in the same manner as observed by Haug et al. [35] when the power input is too low for full penetration. They also observed the same trend for the power input from the wire in the process. Root humping occurred with cold wire and was reduced when welding with hot wire (due to the higher heat input). The front of the keyhole has been observed by Eriksson et al. [37] who found that the flow down the keyhole front is wave driven for 1 μm (fibre laser) laser radiation. Absorption based models depending on laser wavelength were later made by Kaplan [38,39]. Locally, the waves absorb most light on the wave shoulders for 1 μm light, and in areas where the keyhole wall is very steep for 10 μm light (CO2 laser), which tends to create a smoother melt surface on the keyhole wall in the case of CO2 laser welding [38]. It has also been shown by simulation that the keyhole stability is higher when using 10 μm lasers [40]. Globally, the front is wavier for 1 μm and the melt gets more thrust downwards. When using 1 μm lasers at high enough power densities, the welding process can fail and transform into a process called vapour pressure fusion cutting [41].

1.3. Process theory Besides these experimental observations, there are also theoretical models which examine the pressure in the keyhole. These models assume that the melt above the opening at the root needs to be contained by surface tension, otherwise melt will escape. Petring et al. [26] developed a 2D pressure balance equation for maximum root width at any location along the root length, which takes into account the downwards momentum of the melt. Blecher et al. [29] uses a force balance equation and assumes a circular keyhole exit with surface tension acting as an upward force that prevents the melt from exiting. A static gravitational force from the mass of the melt pool acts downwards. Frostevarg et al. [25] reasons that surface tension is responsible for redirecting the downward flow, that would otherwise be ejected as spatter. Depending on the width of the process zone exit, the surface tension becomes weaker for wider diameters. Bachmann et al. [42] successfully applied electromagnetic backing to suppress root humping in stainless steel and aluminium [43], this technique was also used for duplex steel by Avilov et al. [44]. An inductive contactless electromagnetic force is applied to counter the hydrostatic pressure at the root, preventing the melt from exiting the process zone. The force needed is based on a 2D static pressure model, considering contact angles for a spherical exit. The forces included are gravity on the mass above the exit, surface tension, the Laplace pressure and electromagnetic pressure. For partial penetration, it is shown in a 3D simulation model of LBW by Pang et al. [45], that the solidified fusion zone (FZ) will be both deeper and wider than the 90

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

Fig. 2. HSI of LAHW on the root side, having root humping and occasional spatter [25]. Table 1 Material composition (wt %) of the steel plates and filler wire.

∗

Name

C

Si

Mn

P

S

Al

Nb

V

Ti

Fe∗

S420 MC AWS A5.18

0.10 0.08

0.03 0.85

1.50 1.70

0.025

0.01

0.015

0.09

0.20

0.15

98.01 97.37

Displayed values are maximum, except for Fe, which is the minimum

Table 2 Samples and corresponding variables used in the experiments. Sample#

Average laser power (kW)PL

Welding speed (m/min)v

Gap width (mm)d

Process

Wire feed rate (m/min)wv

Arc power (kW)PA

1 2 3 4 5 6 7

11 10 10 – 12 10 – 12 8 – 10 8 – 10 5 – 12

1.8 1.8 1.8 1.8 1.8 1.8 1

0.1 0–0.3 0.1 0.1 0.2 0.2 0.1

LAHW LAHW LBW LAHW LBW LAHW LAHW

8 8 – 8 – 8 4

5.7 5.7 – 5.7 – 5.7 5.7

actual keyhole depth and width at the root. The simulation shows that melt will form behind and below the keyhole bottom and in some cases full penetration is not achieved fully by the keyhole, but conducive heat could melt through to the bottom of the sheet behind the keyhole. 1.4. Current paper Knowledge about weld root shape geometry is increasing but limited. Existing models do not take into account internal melt flow and the surface tension effects behind the keyhole exit. Also, there are no comparative observations of keyhole exit. In this paper, direct observations of root formation, including semi-flat roots, humping and root sagging, are presented along with an analytical model and a summary of methods which can counteract root humping. Fig. 3. Setup of high speed imaging to observe the weld root with the illumination laser in-line with the camera.

2. Methodology To be able to observe the weld root, the sample sheets where positioned and clamped so that one sheet was not fully supported and the root could be seen with a camera. A side view of the experimental setup for welding and HSI is shown in Fig. 3. The HSI equipment (including camera and illumination laser) was placed beside the sheets, looking downwards at a mirror reflecting the light, so that the mechanics of weld root formation could be observed perpendicularly from the side at a 30° angle. The welding process was filmed at 1000 or 1500 frames per second, with each frame illuminated with 500 W for 5 μs (average 10 W). A process map and a survey of root humping together with possible prevention methods was thereby developed.

12 mm thick welds where produced using both LBW and LAHW, 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) in CW mode. The GMA torch was applied in a tilted leading (30˚) position, at a 3 mm distance from the laser irradiation zone. 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 applied shielding gas was Mison18 (82% Ar, 18% CO2 ) at a flow rate of 20 l/min through the weld torch nozzle. The LBW welds were carried out without shielding gas. The 12 mm mild steel sheets were welded with a milled I-gap in a butt joint configuration, with a controlled gap varying between 0 and 0.3 mm. Surface oxides were removed by grinding. Table 1 shows the material properties of the base material and the filler wire. The laser was focused 4 mm below the surface by 300 mm focal length optics to a spot diameter of 800 μm (Rayleigh length ± 4 mm) with an output power which was varied as listed in Table 2. To prevent back reflections damaging the optical fibre, a slight tilting (−7˚) of the laser was applied.

3. Welding results and observation 3.1. Sample examination The appearance of the top and root surfaces for all the samples listed in Table 2 can be seen in Fig. 4, with fusion zone (FZ) diameters mea91

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

Fig. 4. Weld cap and root appearance. Lines indicate where FZ width measurements are made.

Fig. 5. Frames from HSI observing the root for LAHW (a)-(c) samples 1–2 respectively and (d) sample 7.

Table 3 Fusion zone width of samples at cross cut positions at the sheet surfaces (i.e. above the weld cap). Sample

1

2

3

4

5

6

7i

7ii

Cap Root

3.68 1.90

4.01 2.15

2.02 1.10

4.06 1.32

1.58 1.16

2.97 2.09

3.47 2.33

3.86 3.15

to humping, followed by root sagging. However, as LBW has a thinner root, the humping regime has a shorter window of operation and both the humps and root sagging are reduced. For wider gaps, penetration efficiency (power needed to penetrate a certain depth) is increased and the LBW process (sample 5) does not form humps, but eventually reaches the over penetration (OP) regime, with root spatter and concavity. For the LAHW process the humping region has a smaller operating window when the air gap is increased (sample 6 vs Sample 4). These trends imply that both the penetration efficiency and the width of the keyhole exit strongly affect the appearance of the root. Sample 7 was welded at reduced speed and thus had longer heating times, resulting in increased weld width. Also for this case, as power increases, the humps are elongated and more separated, until root sagging develops. In the humping regime the humps are connected by a narrower weld bead. This intermittent weld bead gradually widens when power is increased, until root sagging is achieved. Seemingly, the average weld bead volume remains similar for both the humping and sagging regimes.

sured in cross sections at the melt cap in Table. 3. Sample 1 is treated as reference weld, where the root humps form regularly with a weld seam between the humps, with underfill at the weld cap. Sample 2, which had a controlled, widening gap, begins with barely full penetration which gradually turns into humping and later, as the humps become elongated, into root sagging. For samples 3–6, it is possible to compare the effects of using LBW or LAHW for different gaps and increasing laser power. For both the LBW and LAHW processes (samples 3–4) increasing power resulted in a gradual change from insufficient penetration 92

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

Fig. 6. Frames from HSI observing the root for LBW and LAHW. (a) and (c) is LBW at 0.1 and 0.2 mm gap respectively. (b) and (d) is LAHW at 0.1 and 0.2 mm gap respectively.

Fig. 7. (a) Dimensional description for surface tensional forces for a free lower exit of a liquid filled tube (b) illustration of root shape for two opening diameters and forces. (c) Upwards force by surface tension for different angles and (d) equilibrium angles for different pressures, depending on exit diameters.

93

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

Fig. 8. (a) Illustration of melt flow in thick sheet keyhole welding including pressures and melt flows, viewed (a) in cross section and (b) from below. (c) Surface tensional melt to surface contact areas in front of the keyhole and behind.

Based on these observations, the root regime follows the following pattern as input power is increased: 1. 2. 3. 4. 5.

Insufficient penetration (IP) Root humping (RH) Good result (GR) Root sagging (RS) Over penetration (OP), root humping with occasional spatter or open pores at the root, or root concavity

3.2. Direct observation of root melt flow HSI frames of the melt flow are shown in Figs. 5 and 6. In all cases spatter only occurs in the vicinity of the keyhole exit. Melt is either ejected or redirected and retained in the melt pool. Melt flows backwards from the process exit and the resulting shape of the solidified root is a result of surface and internal melt flows which are governed by heat input (width and volume of melt), surface tension and cooling (solidification). When the laser power is gradually increased the trend of root bead appearances is: IP < RH < RS < OP as shown by Haug et al. [35], Petring et al. [26] and Pan et al. [22] this trend is confirmed here at two welding speeds, different power levels, differing gaps and for both LBW and LAHW. In cases where the process is only just fully penetrating, a fluctuation of penetration depth can be seen (Figs. 5(b),(c),i and 6(a),(c),(d) i) and humps can form near the process exit area. When penetration is increased, melt flows from the process exit at high speed and eventually reaches the end of the melt pool (stagnation area). In a thin melt pool, Fig. 6(c), the melt solidifies gradually in a wedge-like shape. When more melt flows from the exit it may form a hump. However, as seen in all cases in Figs 5 and 6, not all melt flowing to the stagnation area forms humps as some melt can be seen to flow back from the melt channel/hump. The hump eventually solidifies and a new hump can start to form. With increasing laser power the distance between the humps increases, eventually producing root sagging. If even more laser power is applied, melt ejections increase (spatter) and the weld root bead becomes more irregular. As weld pool width increases, Figs. 5(b),(c) and 6(b),(d), more over penetration is required to change root humping into root hang. It is evident that a solidified flat weld root bead cannot be achieved when the weld width is above a certain size. Wide weld roots tend to take on a semi-circular cross section. When the gap width is increased,

Fig. 9. Melt flow behaviour for different cases; (a) non-full penetration to illustrate melt redirection upwards, (b) quasi full penetration or insufficient penetration, (c)-(e) full penetration where; c) has a thin root, (d)-(e) has wider root and (e) has more heat input than (d).

Fig. 6(a) and (b) compared to 6(c) and (d) respectively, penetration efficiency is also increased and a narrower root melt pool is produced. Welding at higher travel rates also produces narrower melt pools so that flatter weld root beads are formed, sample 7 in Fig. 5(c) (slower) compared to sample 4 in Fig. 6(b) (faster). 94

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

𝜎 = 1750 mN/m. If the downward pressure decreases, the surface tension would project a force upwards, pushing melt upwards inside the tube until equilibrium is achieved. With a melt height of h = 3 mm or 12 mm, for example, the downwards gravitational pressure would be 206 or 826 N/m2 respectively. The required angles for equilibrium for these, and other, pressures are shown in Fig. 7(d). For small diameters and low pressures, the equilibrium angle will be close to zero, having an almost flat liquid surface, while at increased pressures and exit diameters the threshold of 90° can be breached so that the liquid detaches.

4. Analysis Based on the observation of samples and HSI, the size of the melt region at the root, keyhole stability and melt flow, are all factors affecting the final weld geometry. Surface tension seems to play a crucial role in redirecting the material at the process zone exit, as well as resisting gravity and retaining the melt in the weld zone. The melt around the keyhole experiences high downward pressures due to surface evaporation. The surface tension forces redirecting the melt mass flow at the process zone exit largely depend on; (i). the geometrical shape and size of the exit zone (governed by; laser parameters, welding speed, sheet thickness), (ii). material properties (viscosity, surface tension, conductivity, melting and evaporation temperatures), (iii). ambient atmosphere (shielding gas, pressure).

4.2. Surface tension effects on weld root Fig. 8. illustrates the various melt flows in the melt pool. Zone i; all melt flowing down to the process exit needs to be diverted and redirected by surface tensional forces, otherwise it will be ejected as spatter. From the HSI in Figs. 5 and 6, it is observed that spatter only comes from the area directly around the keyhole exit, Fig. 8(a) and (b), i. The spatter probably comes from fluctuations in the vapour pressure inside the keyhole and is driven by the downward thrust of the laser on the keyhole front, especially in the case of 1 μm wavelength lasers [39]. There is no clear trend of increasing or decreasing amounts of spatter depending on process exit diameters, and there are far fewer large spatter droplets than small ones. The surface tension produced pressure in front of the keyhole is, by adapting Eq. (5), better described as (geometry as in Fig. 8(c), i),

4.1. Dimensionally dependent surface tension To better understand the influence of surface tension at the root, a mathematical model, previously discussed in [25], may be applied. Fig. 7(a) depicts a cross sectional tube with height h and width d, representing a weld root. The melt experiences a gravitational pressure 𝑃𝑚 = 𝜌𝑔ℎ downwards, where 𝜌 is melt density and g is gravity. This downward pressure is counteracted by the surface tension 𝜎, acting upwards at the edges of the open tube, so there is an equilibrium between downward and upward forces. The model is simplified as any remaining upwards forces from surface tension at the top of the tube and extra forces downwards (e.g. pressure on the melt from keyhole vapour) are accounted for by the parameter PP . The pressures above and below this tube are also assumed to be equal, but could also be accounted for in PP . The total downward pressure is then 𝑃𝑡𝑜𝑡 = 𝑃𝑚 + 𝑃𝑃 ,

4𝜎 𝑃𝜎 = sin 𝛼 ( ). 𝑑𝑝 − 𝑑𝑘

Root humps form at the trailing end of the solidification front, to which molten metal flows in a long channel, zones ii-iv, the cross sectional shape of which is determined by surface tension and the rate of solidification. This melt flow channel only marginally widens by heat conductivity melting at the sides. In this region, the surface tension produces a pressure that will also push back the melt, which also is drawn up into the melt pool by internal melt flows (and surrounding pressures). Root concavity (together with spatter) may be caused by upward pressures (part of PP ). Zone ii; when the melt moves behind the process exit zone, the surface tension will impose high pressures on the melt upwards if there is molten metal beneath the sheet edges, according to (geometry as in Fig. 8(c) ii);

(1)

where the downward force FP is Ptot projected on a circular area 𝜋𝑑 2 . (2) 4 The force produced by surface tension acts along the solid-liquid interface to make the exit melt surface flat and thereby imposes a reactive force upwards, keeping the melt from falling out of the weld zone. The magnitude of this reactive force is angle dependent and expressed as 𝐹𝑃 = 𝑃𝑡𝑜𝑡 𝐴 = 𝑃𝑡𝑜𝑡

𝐹𝜎 = 𝐶𝜎 sin 𝛼 = 𝜋𝑑𝜎 sin 𝛼.

(6)

(3)

2𝜎 . (7) 𝑑 Any metal flowing beneath the edges will thereby be pushed into the weld pool if there is molten metal above, but will otherwise keep its momentum. With thin melt widths, this upward flow may even cause the melt to rise above the sheet edges and even solidify in that position. Zone iii; when previously molten metal has solidified above the still liquid and flowing melt, surface tension still produces upwards pressure along the length of the melt pool flowing towards zone iv. The pressure imposed by surface tension compresses the melt (as it needs to produce as low angle as possible), Fig. 8(a) and (b) iii, producing an inner flow beneath the solidified liquid, back towards the process region and above zone ii. Zone iv; if the volume of melt is too high, humps may form. It is possible for root humps to have more than 90° contact angle to the surface, since there is solidified material above them and the droplets have low mass. If the heat input to the root is increased, the root width increases and zone ii gets longer, forming a curvature that can fill the melt volume at the root (consequently decreasing the upwards pressure by surface tension). Thin welds will thereby have a flat root surface while wider welds have higher volume and rounded geometry (root sagging). For thin sheet full penetration welding, root humping or root sagging will probably not occur, as lower pressures and narrower weld pools are involved. For the thinnest sheets, surface tension will force both the weld cap and root surfaces to be flat. As sheet thickness increases, downward

When the upward and downward forces are in equilibrium, 𝐹𝜎 = 𝐹𝑝 , the angle of the solid-liquid interface is ( ) 𝑑 𝛼 = sin−1 𝑃𝑡𝑜𝑡 (4) 4𝜎 while the resulting angle dependent upward pressure is

𝑃𝜎 = sin 𝛼

4𝜎 . (5) 𝑑 According to Eq. (4), the resulting equilibrium angle increases when the downward pressure increases. This angle also increases when the exit diameter rises. For any diameter, the minimum upwards pressure is when 𝛼 = 0◦ and the maximum when 𝛼 = 90◦ , which would lead to drop detachment. Fig. 7(b) illustrates the required angle (and size) differences for force equilibrium between two gap sizes with the same downward pressure. The surface tension 𝜎 depends on material properties, temperature, and the surrounding atmosphere (e.g. oxygen produces FeO surfactants). With the inclusion of alloying elements the average surface tension of iron could be decreased by 40%, or even 50% if exposed to oxygen. Contrary to iron, that has decreased 𝜎 at elevated temperatures, the surface tension of iron oxides increases slightly with temperature (although the values are still lower than those for unoxidised iron at elevated temperatures) [46]. Fig. 7(c) shows the upwards force depending on the protruded angle of droplet beneath the tube with different contact angles, using Eq. (5) and typical values for iron: 𝜌 = 7.015kg/m3 , 𝑃𝜎 = sin 𝛼

95

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

Fig. 10. Charts of (a) root humping prevention methods and (b) corresponding explanation, where roman numbers link between (a) and (b).

pressures and weld pool widths will increase. It is also reasonable to suggest that surface tension effects are highly relevant when welding upside down. Fig. 9 shows melt flow models for the different welding regimes discussed. Fig. 9(a) illustrates the redirection of downward flow upwards for incomplete penetration. Root humps form when having quasi-full

penetration, where conductive melting can widen the keyhole exit and the humps are fed directly from the melt, Fig. 9(b), or when melt flows via a melt channel towards the hump, Fig. 9(d). For quasi-full penetration, the melt at the rear of the melt pool can melt through the sheet (by conduction rather that laser irradiation) and some melt will flow out and form a hump. An unstable keyhole can alternate between in96

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

complete penetration, conduction melt through and full penetration. For Fig. 9(c)–(e), as long as there is some open melt above the melt pool, surface tension will apply upwards pressure on the melt, making some of the melt to go back towards the keyhole area, Fig. 8(b) ii. Depending on the width of the melt at the root, the solidified surface will have different curvatures, which are correlated to the width and pressures creating the resulting curvature. As long as melt is redirected upwards sufficiently so any melt does not flow backwards to form humps, and the upward flow and melt solidification behind the keyhole is not high enough to form a root concavity a flat root will be formed, Fig. 9(c). A wider melt width allows more melt to flow after the keyhole exit, from zone ii to zone iii, allowing inappropriate amounts of melt to flow to the end of the melt pool so that humps may form, Fig. 9(d). If even more energy is added, the melt pool width increases, allowing for a larger in volume root bead, as well as extending the melt region above the flowing melt behind the process exit, zone ii, forming root sagging instead of humping, Fig. 9(e).

surface roughness can improve the penetration efficiency for ‘zero gap’ welding, e.g. when using laser cut edges [49,51]. Another way of increasing penetration efficiency is to increase the beam quality so that the beam focus is narrower (laser or electron beam, or focusing an arc), to produce a thinner root exit melt pool. It is also possible to increase penetration efficiency by adjusting the focal position, as reported in [52] for LBW and [6] for LAHW. Horizontal welding can also increase penetration efficiency as well as reducing the gravitational pressure towards the root [26,53–55]. The ambient pressure also affects penetration efficiency, as explained by e.g. Pang et al. [56] and low ambient pressure can supress root humping [49]. Another technique to minimise weld root problems is to increase the upwards surface tension pressure by adding electromagnetic backing [42–44]. However, such backing adds to process complexity and is not always feasible due to constraints in accessibility. All these techniques are options for increasing the operating window for achieving a desired weld root surface. If problems persist, physical prevention techniques might be required, e.g. metal strip, powder, [57] or ceramic [58] backing. It is also an option to not fully penetrate, or to use multi-pass welding [59], which is the most common alternative. Post welding techniques are also available, such as grinding [24] or laser re-melting [60].

4.3. Root hump suppression methodology Root humps can be prevented by adding more input energy, as demonstrated here and in other studies [7,22,35]. When welding thicker material, simply adding more laser power will increase penetration, but penetration efficiency will decrease and extra heat is required to penetrate, causing the exit to get wider. Another consequence is that adding more laser power increases the pressure on the root, possibly causing spatter. It can be reasoned that there is a limit to weld thickness for full penetration welding without melt ejection (spatter), whilst having a sound weld root without loss of material at the weld cap, as identified by Haug et al. [35]. The operating window for good quality welding can be increased by using a 10 μm (CO2 ) instead of 1 μm (fibre) laser, since the pressure on the melt is lower and the surface tension can successfully redirect the downward melt, thus reducing spatter. When another heat source is added, e.g. LAHW, more process heat is provided to the root, increasing the width of the weld pool. When using LAHW with the laser leading, melt is pushed by the arc towards the rear of the keyhole, creating a flow of melt downwards to the keyhole exit. Therefore, it is recommended for full penetration single pass welding to use LAHW in the arc-leading configuration [15,22,47]. The laser-arc inter-distance, dLA also affects the penetration depth [47]. When using LAHW, the penetration depth is increased, but the penetration efficiency also decreases (the total heat input is increased compared to LBW with the same penetration depth). As predicted by Blecher et al. [29], it has been verified here that excessively high energy input may be a cause for root imperfections to begin with as the surplus energy produces a wider rather than a deeper keyhole. Therefore, penetration efficiency is a very important factor to reduce root hump formation and in order to increase the maximum single-pass weld thicknesses, special techniques and methods need to be considered and applied. A summary of root humping prevention methods that can be used to extend the operating window, without root humps, is shown in Fig. 10. As the main force redirecting the melt at the root, surface tension needs to be kept as high as possible. The chemical composition of the melt is therefore important and high levels of oxidation of the melt on the root [46] should be avoided. For example, mill scale present on the root side in the weld zone provokes root humping [29], probably as a result of oxidization lowering the surface tension [48]. Therefore, applying shielding gas to the root as well as the weld cap is recommended [22]. However, a limited amount of oxidation in the melt could increase penetration efficiency by reducing melt viscosity and therefore producing a narrower process exit. Then there can be a net gain in force produced by surface tension, even though surface tension decreases (smaller rk or d compared to smaller 𝜎) [7,27], according to Eq. (7). One way of increasing penetration efficiency is by introducing a gap in the weld joint, as observed in the present study and reported by others [19,22,49,50]. It has also been reported that moderately high edge

5. Conclusions The formation of weld root surface qualities has been studied by extended literature review and thorough experimental observation. Even though this study considers only laser beam welding (LBW) and laserarc hybrid welding (LAHW), the mechanisms are valid for all keyhole welding (possibly for all full penetration welding). •

•

•

•

•

Different root topology regimes have been categorized and explained. The force exerted by surface tension has been identified as the key factor in redirecting melt flow at the root. Limiting the size of the process exit minimises spatter, while the melt flow, melt width and pressures in the keyhole are responsible for the final weld geometry. Both root humping and root sagging are decreased when the process exit and melt flow width are reduced. A mapping of techniques for preventing these imperfections has also been presented, along with corresponding explanations.

Acknowledgement The author gratefully acknowledges funding by the European Commission, programme FP7-RFCS, project HYBRO, no. RFS-CR-12024. References [1] Wetzel G, Neubert J, Kranz B. Investigations of properties in hybrid welds at 40-mm sheet thickness. Weld World 2014;58 883–91. [2] Miki C, Homma K, Tominaga T. High strength and high performance steels and their use in bridge structures. J Constr Steel Res 2002;58:3–20. [3] Liu C, Zhang JX, Xue CB. Numerical investigation on residual stress distribution and evolution during multipass narrow gap welding of thick-walled stainless steel pipes. Fusion Eng Des 2011;86 288–95. [4] Guo W, Li L, Crowther D, Dong S, Francis JA, Thompson A. Laser welding of high strength steels (S960 and S700) with medium thickness. J Laser Appl 2016;28:22425. [5] Kaplan A. A model of deep penetration laser welding based on calculation of the keyhole profile. J Phys D Appl Phys 1994;27:1805. [6] Zhang M, Chen G, Zhou Y, Liao S. Optimization of deep penetration laser welding of thick stainless steel with a 10 kW fiber laser. Mater Des 2014;53 568–76. [7] Pan Q, Mizutani M, Kawahito Y, Katayama S. Effect of shielding gas on laser–MAG arc hybrid welding results of thick high-tensile-strength steel plates. Weld World 2016;60:1–12. [8] Frostevarg J, Kaplan AFH, Lamas J. Comparison of CMT with other arc modes for laser-arc hybrid welding of steel. Weld World 2014;58 649–60. [9] Kah P. Overview of the exploration status of laser-arc hybrid welding processes. Rev Adv Mater Sci 2012;30 112–32. 97

J. Frostevarg

Optics and Lasers in Engineering 101 (2018) 89–98

[10] Ribic B, Palmer TA, DebRoy T. Problems and issues in laser-arc hybrid welding. Int Mater Rev 2009;54 223–44. [11] Nielsen SE. High power laser hybrid welding–challenges and perspectives. Phys Procedia 2015;78:24–34. [12] Frostevarg J, Kaplan AFH. Undercuts in laser arc hybrid welding. Phys Procedia 2014;56 663–72. [13] Frostevarg J. Comparison of three different arc modes for laser-arc hybrid welding steel. J Laser Appl 2016;28:22407. [14] Victor B, Nagy B, Ream S, Farson D. High brightness hybrid welding of steel. In: ICALEO 2009 congr. proc., 102; 2009. p. 79–88. [15] Zhao L, Sugino T, Arakane G, Tsukamoto S. Influence of welding parameters on distribution of wire feeding elements in CO2 laser GMA hybrid welding. Sci Technol Weld Join 2009;14 457–67. [16] Tucker JD, Nolan TK, Martin AJ, Young GA. Effect of travel speed and beam focus on porosity in alloy 690 laser welds. JOM 2012;64 1409–17. [17] Lu F, Li X, Li Z, Tang X, Cui H. Formation and influence mechanism of keyhole-induced porosity in deep-penetration laser welding based on 3D transient modeling. Int J Heat Mass Transf 2015;90 1143–52. [18] Wiklund G, Akselsen O, Sørgjerd AJ, Kaplan AFH. Geometrical aspects of hot cracks in laser-arc hybrid welding. J Laser Appl 2014;26:12003. [19] Vollertsen F, Grünenwald S. Defects and process tolerances in welding of thick plates. In: Proc. laser mater. process. conf. ICALEO; 2008. p. 489–97. [20] Standard 12932:2013 EN ISO. Welding - laser-arc hybrid welding of steels, nickel and nickel alloys - quality levels for imperfections; 2013. [21] Ilar T, Eriksson I, Kaplan AFH. Simultaneous top and root high speed imaging on droplet formation in laser welding. In: 32nd int. congr. appl. lasers electro-optics, Miami, FL, USA; 2013. p. 283–8. [22] Pan Q, Mizutani M, Kawahito Y, Katayama S. High power disk laser-metal active gas arc hybrid welding of thick high tensile strength steel plates. J Laser Appl 2016;28:12004. [23] Ilar T, Eriksson I, Powell J, Kaplan A. Root humping in laser welding–an investigation based on high speed imaging. Phys Procedia 2012;39:27–32. [24] Chaussé F, Paillard P, Bertrand E, Rückert G. Hybrid laser-metal active gas welding of S460ML steel plates: weldability and consequences of dropping on weld quality. J Laser Appl 2016;28:22416. [25] Frostevarg J, Haeussermann T. Dropout formation in thick steel plates during laser welding. IIW int. conf., Helsinki, Finland; 2015. [26] Petring D, Fuhrmann C, Wolf N, Poprawe R. Progress in laser-MAG hybrid welding of highstrength steels up to 30 mm thickness. In: Proc. int. conf. ICALEO, Forida, USA, 2007; 2007 300–7. [27] Ohnishi T, Kawahito Y, Mizutani M, Katayama S. Butt welding of thick, high strength steel plate with a high power laser and hot wire to improve tolerance to gap variance and control weld metal oxygen content. Sci Technol Weld Join 2013;18 314–22. [28] Salminen A, Piili H, Purtonen T. The characteristics of high power fibre laser welding. Proc Inst Mech Eng Part C J Mech Eng Sci 2010;224 1019–29. [29] Blecher JJ, Palmer TA, DebRoy T. Mitigation of root defect in laser and hybrid laserArc welding. Weld J 2015;94:73–82. [30] Ganesh P, Kaul R, Paul CP, Tiwari P, Rai SK, Prasad RC, et al. Fatigue and fracture toughness characteristics of laser rapid manufactured Inconel 625 structures. Mater Sci Eng A 2010;527 7490–7. [31] Otegui JL, Kerr HW, Burns DJ, Mohaupt UH. Fatigue crack initiation from defects at weld toes in steel. Int J Press Vessel Pip 1989;38:385–417. [32] Berger P, Hügel H, Hess A, Weber R, Graf T. Understanding of humping based on conservation of volume flow. Phys Procedia 2011;12 232–40. [33] Eriksson I, Gren P, Powell J, Kaplan AFH. New high-speed photography technique for observation of fluid flow in laser welding. Opt Eng 2010;49. [34] Powell J, Ilar T, Frostevarg J, Torkamany MJ, Na S-J, Petring D, et al. Weld root instabilities in fiber laser welding. J Laser Appl 2015;27:S29008. [35] Haug P, Rominger V, Speker N, Weber R, Graf T, Weigl M, et al. Influence of laser wavelength on melt bath dynamics and resulting seam quality at welding of thick plates. Phys Procedia 2013;41:49–58. [36] Li S, Chen G, Zhang M, Zhou Y, Zhang Y. Dynamic keyhole profile during high-power deep-penetration laser welding. J Mater Process Technol 2014;214 565–70.

[37] Eriksson I, Powell J, Kaplan AFH. Measurements of fluid flow on keyhole front during laser welding. Sci Technol Weld Join 2011;16 636–41. [38] Kaplan AFH. Fresnel absorption of 1μm-and 10μm-laser beams at the keyhole wall during laser beam welding: comparison between smooth and wavy surfaces. Appl Surf Sci 2012;258 3354–63. [39] Kaplan AFH. Absorption homogenization at wavy melt films by CO 2-lasers in contrast to 1μm-wavelength lasers. Appl Surf Sci 2015;328 229–34. [40] Koch H, Leitz K-H, Otto A, Schmidt M. Laser deep penetration welding simulation based on a wavelength dependent absorption model. Laser assist net shape Eng 6, proc LANE; 2010. Part 2 2010;5, Part B309–15 http://dx.doi.org/10.1016/j.phpro.2010.08.057. [41] Schäfer P, Olschowsky P. Vapor-pressure fusion-cutting, a new remote 3D-cutting technology. Proc Stuttgarter Lasertage; 2010. [42] Bachmann M, Avilov V, Gumenyuk A, Rethmeier M. Experimental and numerical investigation of an electromagnetic weld pool support system for high power laser beam welding of austenitic stainless steel. J Mater Process Technol 2014;214 578–91. [43] Bachmann M, Avilov V, Gumenyuk A, Rethmeier M. Numerical simulation of full-penetration laser beam welding of thick aluminium plates with inductive support. J Phys D Appl Phys 2011;45:35201. [44] Avilov V, Fritzsche A, Bachmann M, Gumenyuk A, Rethmeier M. Full penetration laser beam welding of thick duplex steel plates with electromagnetic weld pool support. In: 34th int. congr. appl. lasers electro-optics, Atlanta, GA, USA; 2015. p. 571–9. [45] Pang S, Chen L, Zhou J, Yin Y, Chen T. A three-dimensional sharp interface model for self-consistent keyhole and weld pool dynamics in deep penetration laser welding. J Phys D Appl Phys 2010;44:25301. [46] Keene BJ. Review of data for the surface tension of iron and its binary alloys. Int Mater Rev 1988;33:1–37. [47] Kah P, Salminen A, Martikainen J. The effect of the relative location of laser beam with arc in different hybrid welding processes. Mechanika 2010;3:68–74. [48] Karlsson J, Norman P, Kaplan AFH, Rubin P, Lamas J, Yañez A. Observation of the mechanisms causing two kinds of undercut during laser hybrid arc welding. Appl Surf Sci 2011;257 7501–6. doi:10.1016/j.apsusc.2011.03.068. [49] Sokolov M, Salminen A, Katayama S, Kawahito Y. Reduced pressure laser welding of thick section structural steel. J Mater Process Technol 2015;219 278–85. [50] Kim Y-P, Alam N, Bang H-S, Bang H-S. Observation of hybrid (cw Nd:YAG laser + MIG) welding phenomenon in AA 5083 butt joints with different gap condition. Sci Technol Weld Join 2006;11:295–307. doi:10.1179/174329306X107674. [51] Sokolov M, Salminen A, Somonov V, Kaplan AFH. Laser welding of structural steels: influence of the edge roughness level. Opt Laser Technol 2012;44. 2064–71 http://dx.doi.org/10.1016/j.optlastec.2012.03.025 . [52] Karlsson J, Markmann C, Alam MdM, Kaplan FH. A. Parameter influence on the laser weld geometry documented by the matrix flow chart. In: 6th LANE, 5. Elsevier B.V.; 2010. p. 183–92. doi:10.1016/j.phpro.2010.08.043. [53] Seffer O, Lindner J, Springer A, Kaierle S, Westlin V, Haferkramp H. Laser GMA hybrid welding for thick wall applications of pipeline steel with the grade X70. In: 31st ICALEO conf., Anaheim, California, USA; 2012. p. 494–504. [54] Nilsson K, Flinkfeldt J, Nilsson T, Skirfors A, Gustavsson B. Laser hybrid welding of high strength steels, n.d. [55] Kristensen JK, Webster S, Petring D. Hybrid laser welding of thick section steels – The Hyblas project. In: Proc. 12th nord. laser mater. process. conf. NOLAMP, Anaheim, California, USA. Technical University of Denmark; 2009. [56] Pang S, Hirano K, Fabbro R, Jiang T. Explanation of penetration depth variation during laser welding under variable ambient pressure. J Laser Appl 2015;27:22007. [57] Seffer O, Lahdo R, Springer A, Kaierle S. Laser-GMA hybrid welding of API 5 L X70 with 23 mm plate thickness using 16 kW disk laser and two GMA welding power sources. J Laser Appl 2014;26:42005. [58] Wahba M, Mizutani M, Katayama S. Single pass hybrid laser-arc welding of 25 mm thick square groove butt joints. Mater Des 2016;97:1–6. [59] Gook S, Gumenyuk A, Rethmeier M. Hybrid laser arc welding of X80 and X120 steel grade. Sci Technol Weld Join 2014;19:15–24. [60] Frostevarg J, Torkamany MJ, Powell J, Kaplan AFH. Improving weld quality by laser re-melting. J Laser Appl 2014;26:41502.

98