You are on page 1of 9
Available online at www.sciencedirect.com Physics Procedia 12 (2011) 232–240 LiM 2011 Understanding of Humping Based on

Available online at www.sciencedirect.com

Available online at www.sciencedirect.com Physics Procedia 12 (2011) 232–240 LiM 2011 Understanding of Humping Based on

Physics Procedia 12 (2011) 232–240

LiM 2011

Available online at www.sciencedirect.com Physics Procedia 12 (2011) 232–240 LiM 2011 Understanding of Humping Based on

Understanding of Humping Based on Conservation of Volume Flow

Peter Berger*, Helmut Hügel, Axel Hess, Rudolf Weber, Thomas Graf

Institut für Strahlwerkzeuge (IFSW), Pfaffenwaldring 43, 70569 Stuttgart, Germany

Abstract

A generalised view on the causes leading to humping during high-speed seam welding is proposed. It is based on the fact that, whenever the melt velocity in a welding bead exceeds considerably the welding speed, the weld groove cannot be completely filled by melt at any location and any instant. As a consequence of solidification from the walls of the weld groove, the shape of the free cross section for the melt flow is changed in such a way that the rest melt stream experiences an upwards deflection which initiates the formation of a hump. Conservation of volume in an incompressible fluid requires that in the solidified seam regions of smaller and larger cross sections compared to that of the groove must exist. The initiation of humping is discussed for different situations in laser beam welding characterized by the aspect ratio welding depth over spot diameter. As this figure is typical for the geometry of the melt pool/ weld groove, it appears as a meaningful parameter for correlating the onset of humping.

Keywords: Welding; Quality; Instabilities; Humping; Surface Structure

1. Introduction

The humping phenomenon is generally understood as the formation of drop like piles on the top of a weld bead that alternates with the appearance of undercut between the individual humps. It occurs during seam welding at feed rates above a certain value and is likewise observed in electron beam, arc and laser beam welding. As it imposes a restriction to the attainable welding speed and, hence, limits the productivity of these joining technologies, it has been intensively studied for more than 30 years. To explain the causes of humping several approaches were pursued which are based on quite different conceptions. None of them, however, is capable to fully explain the phenomena observed under a great variety of process parameters; moreover, some of them are questioned in recent literature. Three facts, however, are generally accepted, namely that (1) the humps are finally formed at the end of the molten region, (2) the onset of humping is related to high melt velocities within the weld pool, and (3) the high-velocity melt flow is due to the ablation pressure at the keyhole front in e-beam as well as in laser welding and is produced by electromagnetic forces in the discharge during arc welding. The herein presented view of the causes giving rise to humping should be regarded as an extended and generalized interpretation of experimental evidence so far assessed under similar, yet more special aspects; they are to be found in [1], [2] and [3]. These approaches, and also the occasionally discussed one of a decaying jet, shall be briefly outlined, before in sections 2 and 3 more recent experiments are presented and interpreted with regard to the conservation of mass.

* Corresponding author. Tel.: +49-711-685-66843; Fax: +49-711-685-66841. E-mail address: peter.berger@ifsw.uni-stuttgart.de.

1875-3892 © 2011 Published by Elsevier Ltd.

doi:10.1016/j.phpro.2011.03.030

Peter Berger et al. / Physics Procedia 12 (2011) 232–240

233

From the results in [1] it clearly becomes evident that under conditions leading to humping in e-beam welding of thin metal sheets the picture of a melt groove that is completely filled with melt is not valid. Instead, a high-velocity melt jet starts to solidify from the side-walls (here in the case of through-welds) of the groove thereby narrowing the “free” flow channel. This causes the rest of the flowing melt to be deflected upwards and towards the center of the bead where it finally piles up to humps. Such a mechanism is regarded as a deciding factor for the onset of humping in gas tungsten arc (GTA) welding as well [4]. Quite similar arguments are formulated along these lines in [5] and (6) (although both publications deal with gas metal arc (GMA) welding, where additional momentum is transferred to the melt by the droplets from the wire, the basic physical phenomena are identical): A pinching of the thin metal film moving along the wall of the weld groove and its rapid solidification is determined as a requirement for hump formation. The elongated “wall jet” – a perfect term used in [6] to describe the high-speed melt flow in the thin film – is sensitively exposed to rapid solidification which can choke off it from a growing hump. As a consequence, a “melt bridge” remains for some time between the solidifying hump at the pool end and a new swelling occurring closer to the arc position. Such a “bridging” was observed also in laser beam welding [2] where this fact was interpreted as a periodic switching of melt pool length. That surface tension effects should play an important role during the formation of humps suggests itself and shall be underlined by two examples from e-beam [7] and GTA welding [4]. Most convincing evidence is given in [7] where a higher oxygen content is demonstrated to drastically enhance the tendency to humping. Physically the same effect is documented in [4] by observing that with lower sulfur content the welding speed becomes higher at which humping begins to develop. Furthermore, surface tension is implicit related to the otherwise purely geometrical issues of the Raleigh instability theory as it determines the growth time of a disturbance leading to the break-up of a cylindrical fluid jet [8], [9] and [10]. It is interesting to note, how these authors thought about the characteristic distance between the individual humps. In [8] it is stated that the calculated critical length l c after which a liquid jet breaks up into droplets does not sufficiently agree with the experimentally observed pool length l m (the reported values of l c are up to 30% larger than l m ), whereas in [9] and in [10] l c = l m was taken as an a priori assumption. Implicitly this equating indicates the obvious view of the authors that solidification at the pool end plays an important role and must somehow be related to the time scale of hump formation. In fact, the theoretical treatments in [9] and [10] using that ansatz end up by identifying geometrical features of the solidified bead (ratios of melt pool width to length [9] and, additionally, seam height to width [10]) as characteristic figures to describe humping. – Despite this fact it remains questionable (see also [11]) whether that approach is applicable against the background of experimental evidence (e.g. a “free” jet definitely does not exist, humps my start to grow close behind the keyhole). Other authors [2], [12] calculated the 2D-flow field around a keyhole and regarded the high melt velocity in the center of the pool as responsible for the humps, being formed as a consequence of high stagnation pressures at the rear end of the weld pool. The high velocity values reported in [2] are confirmed by more recent FEM calculations [13]. They can exceed the feed rate by more than an order of magnitude and, during deceleration of the flow, produce high stagnation pressures. Yet another cause for humping is discussed in [3] and [12] where the flow field close to the keyhole is regarded responsible. An upwards directed flow component in the pool behind the keyhole interacts with the flow around the keyhole thereby leading to the formation of a hump. Possible causes for the underlying momentum in the melt may be a momentum transfer via friction and direct impact of the vapor flow onto the rear keyhole wall. In welding with disk and fiber lasers an additional component might arise from the “piston” effect of moving steps at the keyhole front [14]: the melt film accelerated there along the direction of the laser beam axis is deflected at the bottom of the weld groove below the keyhole tip and gives rise to an upwards flow.

2. Experimental and theoretical results substantiating the proposed conception

In addition to the investigations performed in the authors` labs [2], [13], [14], [15] and [16], reliable facts from other publications have been evaluated, among them and in particular those reported in [10], [17], [19] and [20]. The lasers used in the various experiments included all types that were and are employed in industrial applications, namely CO 2 , Nd:YAG, disk and fiber lasers. The most common diagnostic method in all recent studies is high-speed

  • 234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240

photography which allows to observe and distinguish in detail keyhole geometry, melt stream(s) and hump formation (and, of course, spattering in its great variety). The pictures in Figure1 show the appearance of different fluid dynamic states as the travel speed is increased at otherwise constant parameters. In these experiments [15], [16] a single mode fiber laser with a BPP of 0.34 mm mrad was used. The focusing conditions were such that a (theoretical) spot diameter of 50 µm was established at the work-piece surface. The line energy P/v of 600 J/m was kept constant and adjusted in such a way that the welding depth in the metal sheet (stainless steel 1.4310, thickness 0.5 mm) was close to 0.5 mm but no full penetration did occur.

stainless steel; s = 500 m; d f = 50 m; P /v= 600 J/m

234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 photography which allows to observe
234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 photography which allows to observe
  • v = 20 m/min

short weld pool, tiny spatters,

round keyhole

  • v = 40 m/min

longer weld pool, standing melt wave,

elongated keyhole

234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 photography which allows to observe
  • v = 50 m/min

long waves, longer weld pool,

elongated keyhole

234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 photography which allows to observe
  • v = 60 m/min

periodic beaded weld seam,

elongated “keyhole”

234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 photography which allows to observe
  • v = 70 m/min

resize of window display

long region of underfill, hump formation

starting far behind region of energy coupling

234 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 photography which allows to observe
  • v = 80 m/min

distance between humps becomes smaller

Figure 1: Appearance of keyhole and melt pool with increased travel speed (note that scale is changed in the two last pictures), see text.

At “low” values of travel speed, the classical picture of a thermally determined laser weld can be seen, i.e. a more or less circular keyhole orifice and a typical elliptically elongated weld-pool geometry. Tiny spatters leave from the rim of the keyhole. This indicates that a strong vapor flow exerts high shearing strains via friction on the keyhole wall thereby tearing off small droplets. As the speed is increased the keyhole gets elongated under the direct impact of the vapor on its rear wall, see also [17], [19]. At the same time, a melt wave appears behind the keyhole; it is initiated by momentum which stems from vapor impact and from the “piston” effect of moving steps at the keyhole front, see e.g. [14]. For speed values around 60 m/min and above, behind the keyhole there exists a region of “underfill” or “depression”, where the volume flow V , which was molten at the front of the pool, does by far not fill up the weld groove; this phenomenon is seen also in the experiments described in [17], [19] and [20]. The fact that the melt flow is concentrated mainly at the bottom and less at the side-walls is due to an obviously existing velocity component along the front of the keyhole pointing towards the tip of the keyhole. Only recently, it has been directly visualized and measured to be around 8 m/s for the conditions investigated in [20]. It is noted that in situations showing this underfill the term keyhole should not be understood in the sense that a closed “beam trap” does exist, but rather as the geometrical region where energy is coupled in. All together, the appearance of underfill is a clear indication that the velocity of the melt stream by far exceeds the welding speed. Downstream of this region then the melt starts to pile up and form humps. During their growth they are still faster than the welding speed. At the very rear of the pool, where they begin to solidify, they then move with the travel speed, i.e. they are fixed to the frame of

Peter Berger et al. / Physics Procedia 12 (2011) 232–240

235

the work piece. Although the typical aspect ratio, a = s/d f , welding depth s over spot diameter d f , is an order of magnitude larger here as in [17], [19], the observed fluid dynamic phenomena are essentially the same. Driving force for the high-speed melt jet downstream of the region of energy coupling is the high ablation pressure there [19]. Results of 2D-calculations [13], presented in Figure 2, illustrate the fluid dynamic conditions. The underlying assumption

d f = 50 m d f = 50 m d f = 25 m d
d f = 50 m
d f = 50 m
d f = 25 m
d f = 25 m
d f = 50 m
d f = 50 m
d f = 25 m
d f = 25 m
v in m/min
v in m/min
u mx in m/s
p 0 in bar
b in m
in m

Figure 2: Calculated fluid dynamic properties around a keyhole. The diagrammes show the total pressure p 0 at the keyhole front, the melt velocity component u mx in welding direction at both sides of the keyhole, the pool width b there and the melt film thickness at the front.

of a cylindrical keyhole

does not impose restrictions

on the comparability of this data with measurements since the calculations reveal

that practically all energy is deposited at the front. It can

be seen that the melt velocities at the side walls

are about an order of

magnitude higher than the

welding speed. Also, the

theoretical seam widths agree well with the meas- ured ones of 110, 80, and 70 µm corresponding to weld speeds of 15, 50 and 100 m/min. Furthermore, it appears reasonable to

assume that the flow com- ponent along the front is of the same order of magnitude as the velocity u mx on the sides of the keyhole presented in Figure 2. Thus, from both, experiments and theoretical descriptions, the existence of a high-speed melt stream behind the region of energy coupling can be regarded as given fact for process parameters yielding long weld pools. Start and growth development of humps are studied from sequences like those presented in Figure 3 which were obtained at 15000 fps. After a region of underfill where the melt velocity component in travel direction u mx is larger than the travel speed v, the melt begins to rise up. This position l s is interpreted to be the start of hump development. As the hump moves farther backwards, its velocity becomes lower. At the end of the pool l m , when solidification of the hump commences, it moves with v. Under some conditions, however, the previously produced hump is still liquid. This has led to assessments according to which the weld pool length fluctuates [2] respectively that there exist more than one hump within l m [6], [16]. With increased v the distance between two solidified humps d h decreases slightly. This tendency qualitatively agrees with findings in [10] and [18] where a proportionality

  • d h ~

P / v
P
/
v

was deduced from experiments.

(1)

  • 236 Peter Berger et al. / Physics Procedia 12 (2011) 232–240

l s start of hump built up P/ v = 600 J/m l s d f
l s start of hump built up
P/ v = 600 J/m
l
s
d f = 50 m l m length of molten pool
stainless steel
d h distance between humps
0.010318
s
0.010445
s
0.010572
s
l
d
m
h

Figure 3: Development of humps at v = 60 m/min and definition of characteristic dimensions, see text.

The evaluation of pictures like those in Figure 3 yields

a dependence of the characteristic lengths on welding

speed at otherwise constant parameters as presented in Figure 4. The data shows a steadily increasing growth of

both quantities with speed. This is explained as a result of the rapid increasing u mx with v (see Figure 2). The high- velocity, high-temperature melt stream leads to an intensified energy transport to the rear so that

solidification sets in later (farther downstream). The reproducible “discontinuity” at around 60 m/min might be understood as a consequence of a changing keyhole/ melt

flow interaction: Up to this speed value an elongated keyhole exists the rear wall of which is hit by vapor flow from its front. At somewhat increased speed the high- velocity melt flow around the front has a larger

momentum so that the side wall streams merge only at a distance far away from the front; a direct impact of a vapor jet on the melt does no longer occur. Hence, this “jump” would characterize the change from a “keyhole dominated” regime to a ”fluid dynamic dominated” one.

3500 3000 P/v = 600 J/m d f = 50 m l m 2500 stainless steel
3500
3000
P/v = 600 J/m
d f = 50 m
l
m
2500
stainless steel
2000
1500
l
s
1000
500
0
0
10
20
30
40
50
60
70
80
90
100
l s ,
l m in m

v in m/min

Figure 4: Weld-pool length l m and distance ls where flow appears to become deflected and humps start to grow.

3. Conception based on conservation of mass/volume flow

The approach presented in the following tries to understand humping in a most general way as a consequence of mass, respectively volume conservation over the complete time period from melting during energy coupling and acceleration of the molten material there until all the melt`s complete solidification. Accordingly, when somewhere along the weld pool at some time instant the melt velocity considerably exceeds the feed rate, the weld “groove” there cannot be filled completely with melt. As a consequence from the conservation of mass (volume appears a more appropriate term as it accounts also for thermal expansion effects), at some other location material has to be accumulated. In the interaction of fluid dynamics, solidification of the melt (mainly) from the sidewalls of the groove and surface tension effects humps are formed. So, in a completely solidified seam there must exist regions with smaller and larger cross sections as compared to that of the groove. This idea, which explicitly takes into account the conservation of mass/volume, is graphically represented in Figure 5; it altogether follows largely Arata`s [1] interpretation of phenomena observed during e-beam welding of metal sheets.

Peter Berger et al. / Physics Procedia 12 (2011) 232–240

237

A(x) A = V / v groove d h x 1 d h for = const:
A(x)
A
=
V / v
groove
d
h
x
1
d
h
for = const:
A x
(
)
dx
A
groove
d
h
A

Figure 5: Sketch illustrating the consequence of mass/volume conservation on bead cross section. V is the volume being molten per unit of time and A groove is the cross section produced thereby. Volume losses due to spatter and vaporisation as well as volume increase by thermal expansion are neglected in the discussion.

In laser beam welding (what is the main concern of this contribution) humping is observed within a wide field of experimental conditions which, at a first glance, can be characterized by the spot diameter. In the case of large values, say d f about 0.4 mm and larger (figures which are common for most CO 2 and Nd:YAG lasers), a situation is found which shows close resemblance to that in arc welding. In such cases humping occurs in relatively shallow grooves with ratios of weld depth to width of about one. Welding with disk and fiber lasers allowing spot diameters as low as several tens of µm also leads to humping with, however, very narrow weld seams and aspect ratios of the order of 10 and more; a situation similar to e-beam welding of thick sections. Because of sometimes confusing discussions without clearly addressing the particular situation, here the two cases – shallow or deep and narrow beads – will be discussed on the basis of s/d f which appears to be the most appropriate figure for a general description.

AB C D t = t 0 u m v l s l m d h
AB
C
D
t = t 0
u m
v
l
s
l m
d h
t 0 +2 t
A(t 0 +2 t)
B(t 0 +2 t)
t 0 + t
A(t 0 + t) B(t 0 + t) C(t 0 + t)
t
0
A(t 0 )
B(t 0 )
C(t 0 )
D(t 0 )
t 0 t
B(t 0 t) C(t 0 t)
D(t 0 t)
t
u mx » v
u mx v
u mx v

In Figure 6 an attempt is undertaken to illustrate the time and

space dependent development of humps. On top of the figure is

a sketch of a melt pool which is fixed in space. The positions of

the different humps A, B, C, D at various time steps can be

followed along the straight parallel lines v = dx/dt, the slope of

which corresponds to 1/v. Indicated are different flow regimes

distinguished by relating the longitudinal flow component u mx to

v. The region of very high melt flow velocity, u mx >> v, ends where, due to solidification effects, the flow is deflected

upwards and melt is piled up to form a hump. Since the position

l s as well as the geometry there (i.e. the cross section of the melt flow already solidified) depend on the spot diameter – at otherwise constant process parameters – the particular features

will be accounted for in Figs. 7 and 8. In the x-t diagram the high velocity of a fluid element travelling from the front towards l s is indicated by the smallest slope; this region corresponds to

the appearance of underfill. After the hump has begun to grow,

  • x it moves (still growing) towards the end of the pool where it comes to rest relative to the work piece, i.e. u mx = v. The stagnation pressure of the oncoming flow and surface tension

Figure 6: Time dependent development of humps depicted in a x-t diagramme, see text.

finally form the hump A(t o ). A still liquid hump B(t o ) could be assessed as the existence of two humps within the melt pool; the length of which in the notation of Figure 6 would then have a value of l m = l s + d h . The distance d h is established in a

balance of fluid dynamic and thermal effects including cooling and solidification; its value cannot be predicted by the quantitative argumentations developed herein (on the other hand, an equating of d h with l m is neither satisfying from the point of gaining an understanding of physical causes).

s

  • 238 Peter Berger et al. / Physics Procedia 12 (2011) 232–240

l d m h l s a b cd u mx » v u mx v
l
d
m
h
l
s
a
b
cd
u mx » v
u mx v
u mx v
e
e
bottom jet or
wall jet region
deflection region /
start of hump formation
stagnation
region
b
cross section e
cross section d
cross section a
melt stream
solidified wall jet
weld groove
cross section b
cross section c
s

Figure 7: Hump formation in a typical shallow and long pool geometry caracterized by s/d f 1.

l d m h l s u mx v d d a b c s
l
d
m
h
l
s
u mx v
d
d
a
b
c
s
s 238 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 l d m h

a

a

cross section d

s 238 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 l d m h
b
b
s 238 Peter Berger et al. / Physics Procedia 12 (2011) 232–240 l d m h

Here, in addition to melt

deflection by rapid

solidification from side

walls, u mx stemming from

vapor friction in keyhole

(and moving steps at its

front) initiates humping

melt stream

solidified melt

weld groove

cross section a

cross section c

cross section b

Figure 8: Hump formation in a typical deep, slender and short pool geometry characterized by s/d f 10.

The situation in shallow beads s/d f 1 is discussed in more detail along the various sections through the bead as sketched in Figure 7. Here, the region of underfill might extend up to several d f as reported in [10]. The melt flow is concentrated towards the bottom of the groove without wetting the “cold” side walls up to the work-piece`s surface (section a); upon final solidification this effect gives rise to undercut in the region between two humps (section d). Solidifying melt from the wall and the rear causes a narrowing and change of the free cross section so that at some position l s (section e) the local high-speed volume flow is deflected upwards. Viewing many high-speed films, a figure of thumb for l s typically lies between 0.5 l m and l m (the very end of the pool). Thus, the hypothesis presented

Peter Berger et al. / Physics Procedia 12 (2011) 232–240

239

in [2] that the stagnation of a high-velocity melt jet at the pool end causes humping would hold for the case of l s = l m . The two arrows drawn at the hump in its stagnation position (section c) shall indicate that its size might still become larger due to the decelerating oncoming flow as well as by surface tension effects pulling material from the melt bridge between the actual and the preceding hump. A somewhat different flow field will be established in geometries that can be characterized by s/d f = 10, nevertheless, with the same basic mechanisms involved. In very slender and deep weld pools as indicated in Figure 8 the cooling from the side walls is very efficient (note that the cooling rate from a geometrically defined body depends on the ratio of its surface area to volume). This implies that, assuming b ~ d f , the cooling rate increases as d f is made smaller. As a consequence, the melt pool (at least the part below the surface and in the depth) is much shorter in such cases in comparison to its extension for s/d f = 1. The, hence, wedge-like solidification

Peter Berger et al. / Physics Procedia 12 (2011) 232–240 239 in [2] that the stagnation
Peter Berger et al. / Physics Procedia 12 (2011) 232–240 239 in [2] that the stagnation
Peter Berger et al. / Physics Procedia 12 (2011) 232–240 239 in [2] that the stagnation

(a)

front at the rear causes a deflection of the x- component of melt flow to occur far closer to the

keyhole. Thus, the piling up of melt is observed to appear at distances rather near the keyhole [18], [23] defined by the location behind the keyhole at which the weld pool more or less abruptly changes

in depth. This is quite the situation illustrated at the high welding speeds in Figure 4. Thermal expansion [24] can account for an increase of volume flow, which might be a reason that underfill is not so pronounced in these deep and slender welds. In addition to the deflection mechanism, other effects too will contribute to an

(b)

Figure 9: Experimental (left,see text) and theoretical proof of an upwards directed (against the direction of laser beam) flow component for d f = 50 µm, P = 500 W, v = 30 m/min: pool shape (a) without, (b) with taking into account vapor friction in the keyhole.

upwards flow of the melt under these conditions. In particular, the momentum transferred to the keyhole wall by friction from the vapor flow might be of major importance. This can be concluded from many high-speed films and other evidence, examples of the latter are given in Figure 9. The cross section at the left indicates how tracer material (sheet metal without Cr, 50 µm thick, 50 µm below surface) is transported from lower regions of the pool towards its top and, hence, is found in the humps. On the right is demonstrated how by taking into account vapor friction in theoretical calculations this mechanism changes the melt pool shape (in these calculations, not allowing for geometrical changes of the surface, the upwards flow u mz is just deflected close to the (here fixed)

surface giving rise to a longer melt pool there. – Altogether, from these discussions a picture evolves that in essence is quite similar to that given in [3].

4. Conclusions

The way in which the melt flow evolves as function of time and space in a solidifying melt pool is the central aspect that was pursued in the conception presented herein. It is postulated that humps start to develop in a region where the bulk melt flow experiences an upwards deflection which is primarily caused by a change/narrowing of the bead`s cross section due to solidification starting from the wall of the weld groove. As this is a geometrical effect, process parameters yielding the corresponding pool/groove shapes should play a deciding role for the onset of humping. In fact, characterizing typical geometries by the aspect ratio, the description given along these lines can explain the observed phenomena without being in conflict with other approaches, particularly those in [2] and [3]. Although the conception formulated herein takes into account the various phenomena observed in the experiments and is supported by basic physical correlations, it is far away from representing a model allowing quantitative statements. To establish such descriptions it is necessary to obtain more insight into e.g. the solidification under particular conditions, the mechanisms giving rise to the upwards flow close behind a keyhole, the effects of thermal volume expansion (of interest mainly for welding with extreme small spot diameters) and not least, the role of surface tension and other material properties.

  • 240 Peter Berger et al. / Physics Procedia 12 (2011) 232–240

References

[1] Arata, Y.; Nabegata, E.: Tandem electron beam welding. Trans. JWRI, 7 (1978), 101. [2] Beck, M. et al.: Aspects of keyhole/melt interaction in high speed laser welding. SPIE Vol. 1397 (1990), 769. [3] Beyer, E.: Schweißen mit Laser. Springer, 1995. [4] Mendez, P.F.: Penetration and defect formation in high-current arc welding. Welding J. 82 (Oct. 2003), 296-s. [5] Cho, M.H., Farson, D.F.: Understanding bead hump formation using a numerical simulation. Metall.&Mat. Transactions 38B (20 07), 305.

[6] Nguyen, T. C. et al.: The humping phenomenon during high speed gas metal arc welding. Sci. and Techn. of Welding and Joining 10 (2005),

447.

[7] Sievers, E.-R.: Schmelzbadinstabilitäten beim Elektronenstrahlschweissen von Grobblechen. Schweissen und Schneiden 58 (2006), 288. [8] Albright, Ch. E.; Chiang, S.: High speed laser welding discontinuities. J. of Laser Appl.1 (1988), 18. [9] Gratzke, U. et al.: Theoretical approach to the humping phenomenon in welding processes. J. Phys. D: Appl. Phys. 25 (1992), 640. [10] Thomy, C. et al.: The occurance of humping in welding with highest beam qualities. Key Eng. Materials 344 (2007), 731. [11] Kumar, A.; Debroy, T.: Toward a unified model to prevent humping defects in gas tungsten arc welding. Welding Journal 85, (Dec. 2006),

292-s.

[12] Pirch, N. et al.: Die Humping Instabilität beim Schweißen mit Laserstrahlung. Proc. 10th Int. Congress Laser 1991 (1992), 552. [13] Schuster, R.: Melt bath shapes and keyhole vapour pressures. 21st Meeting of Mathematical Modelling of Material Processing with Lasers. Igls/Innsbruck, 2008. [14] Berger, P. et al.: Zur Bedeutung von gleitenden Stufen an der Kapillarfront beim Schweißen und Schneiden mit Laserstrahlen. Accepted for publication in Schneiden und Schweißen, 18. Nov. 2010. [15] Hess, A., Dausinger, F.: Humping mechanisms during high-speed welding with brilliant lasers. 3rd Pacific Int. Conf. on Appl. of Lasers and in Optics, 2008, paper 3 on proceedings CD. [16] Collective of authors: 2nd Intermediate Report on Project SCHARP, FGSW 2007, IFSW University of Stuttgart. [17] Fabbro, R. et al.: Experimental study of the humping process during Nd:YAG laser welding. Proc. 4th Int. WLT Conference La sers in Manufacturing (2007), 277. [18] Neumann, St. et al.: Influence of material and process parameters on the humping effect. Proc. ICALEO 2008, 345. [19] Fabbro, R., Slimani, S.: Melt pool dynamics durig deep penetration cw Nd:YAG laser welding. Proc. 4th Int. WLT Conf. Lasers in Manufacturing (2007), 259. [20] Erikson, I. et al.: High speed camera investigations of the keyhole front. Proc. CD of 13th Int. Workshop on Process Diagnostics ond Control in Laser Beam Welding and Cutting, 2011 (IFSW). [21] Hügel, H., Graf, Th.: Laser in der Fertigung. Vieweg und Teubner (2009). [22] Radaj, D.: Schweissprozesssimulation – Grundlagen und Anwendungen. DVS Vol. 141 (1999). [23] Kawahito, Y. et al.: Elucidation of high-power fibre laser welding phenomena of stainless steel and effect of factors on weld geometry. J. Phys. D: Appl. Phys. 40 (2007), 5854. [24] Sudnik, W. et al.: Numerical simulation of weld pool geometry in laser beam welding. J. Phys. D: Appl. Phys. 33 (2000), 66 2.