You are on page 1of 6

Journal of Manufacturing Processes 40 (2019) 122–127

Contents lists available at ScienceDirect

Journal of Manufacturing Processes


journal homepage: www.elsevier.com/locate/manpro

Comprehensive analysis of spatter loss in wet FCAW considering interactions T


of bubbles, droplets and arc – Part 1: Measurement and improvement
Chuanbao Jia , Yong Zhang, Junfei Wu, Changjian Xing, Bo Zhao, Chuansong Wu

MOE Key Lab for Liquid-Solid Structure Evolution and Materials Processing, Institute of Materials Joining, Shandong University, Jinan, 250061, China

ARTICLE INFO ABSTRACT

Keywords: The spatter loss under different welding parameters and in different surroundings was measured and compared.
Spatter The spatter loss coefficients were calculated via considering the net weight of metal from flux-cored wire.
Underwater welding Experiments indicated that the spatters mainly congregated near the weld seam in 5 mm. The congregated
Bubble spatters seriously influenced the evenness of weld seam and caused weld bead shrinkage and overlap. According
Measurement
to the measured values, the spatter loss coefficients significantly decreased along with the increasing welding
Improvement
current. It was deduced that spatter loss can be effectively improved by elevating the welding current level, in
which case droplet transfer mode, droplet transfer resistance and bubble shielding could contribute.

1. Introduction repelled spatter were two main factors [8].


For underwater welding, few papers have been found focusing on
Underwater wet flux-cored arc welding (FCAW) has been developed spatter loss. Based on X-ray images in wet FCAW, Guo et al. (2015)
for automatic and semi-automatic welding processes [1]. Because this classified the spatter generation modes into three, i.e. droplet repelled
technology could be used conveniently and inexpensively, its devel- spatter, explosive spatter and molten pool shock spatter [9]. However,
opments and promotion are expected to promote much wider applica- the corresponding spatter loss distribution on the substrate was not
tion in the fields like marine engineering, salvage and offshore structure found; the exact influences from bubble evolution, arc behaviors and
construction etc [2,3]. However, because the flux-cored wire is directly surrounding water on spatter generation were not observed. In fact, it
exposed in the water, the complex aqueous environment makes the has been verified that both welding arc and bubbles have a significant
welding process be difficult to predict and control. Spatter loss is one of impact on the generation and distribution of spatters. Chu and Tung
the most important issues that require further investigation to improve (2005) identified that the stability of arc had a huge influence on
the welding quality and stability. spatter generation rate [10]. Zhang et al. (2016) preliminarily revealed
Spatter loss is recognized as a byproduct of droplet transfer. Due to the effect of generated bubbles on the stability of arc and droplet
the special features of the flux-core wire, research results about spatter transfer [11]. By means of common high-speed cameras and special
loss during air FCAW are of significant reference values. Cheng et al. optical design, Jia et al. (2016) captured clear images of the arc, bub-
(2017) found that when the mean current ranged from 140 A to 170 A, bles, droplets and spatters in the same time [12].
there was a very small amount of spatters [4]. Zhu et al. (2016) pro- According to current research results, the complex and less pre-
posed that the long duration of the short-circuit procedure resulted in dictable spatter loss in wet FCAW seriously deteriorates the weld seam
energy concentration and generated more gas explosive forces [5]. Li appearance, welding stability and quality. The spatters features and
et al studied the “droplet rebounded spatter” in dry hyperbaric GMAW, mechanisms have not been revealed considering the complex interac-
which was always generated after the cathode root burning on the tions between bubbles, droplets, arc and spatters. In the part 1 of this
droplet contacted to the workpiece; the electromagnetic force was report, experiments were designed and carried out. The spatter ap-
considered as the main driving force [6]. De et al used deposition ef- pearance and features were compared and analyzed. Measuring method
ficiency to evaluate the metal transfer stability and spattering level; no for spatter loss coefficient in wet FCAW was given. The correlation
correlation was found between the spattering and fume generate rate between welding current and spattering level was preliminarily ob-
[7]. Liu et al investigated the metal transfer and spatter characteristics tained. Increasing welding current was found to be an accessible way
of self-shielded FCAW; it was found that the explosive spatter and arc for reducing spatters.


Corresponding author.
E-mail address: jiachuanbao@sdu.edu.cn (C. Jia).

https://doi.org/10.1016/j.jmapro.2019.03.013
Received 12 September 2018; Received in revised form 29 January 2019; Accepted 12 March 2019
Available online 20 March 2019
1526-6125/ © 2019 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
C. Jia, et al. Journal of Manufacturing Processes 40 (2019) 122–127

distribution are due to the environmental influences. On one side, for


underwater welding the liquid droplets are found to be with large size
up to 7 mm [11,12]. On the other side, the serious disturbances from
surrounding water seriously affect the droplets’ trajectory and beha-
viours. This makes it possible to form larger spatters during complex
metal transfer processes. For self-shielded FCAW in air, the influences
from air are much gentler. The tiny scattered spatters were formed
because of violent metallurgical and chemical reactions in the con-
tinuously melted flux and metal by arc. Much gas was released from the
mixed slag and metal causing complicated forces on the liquid.

3.2. Spatters loss coefficient measurement


Fig. 1. The experimental system of underwater welding in shallow water.
For further investigating the spatter loss influences on wet welding
2. Experimental setup quality, spatter loss coefficient were measured according to the fol-
lowing equation.
As shown in Fig. 1, Bead-on-plate welding experiments were con- the weight of generated spatters
* 100%
ducted in shallow water (fresh water, 0.4 m deep) in an oblong tank. A the weight of metal provided by the consumed wires
regular welding power source for GMAW was employed. Experiments (1)
were conducted in the direct current electrode positive (DCEP) mode.
Because the welding material was self-shielded flux-cored wire, the
The base material was Q235 mild steel with the dimension of
melted slag and liquid metal were inevitably mixed together. It was
200 × 50 × 8 mm. The water tank moved linearly on a platform driven
assumed that the flux-cored wire was well manufactured with the same
by a step motor, while the torch was kept stationary. A rutile type flux-
quality and fill rate along the fed wire; Pure metals of slag and weld
cored wire specially designed for underwater wet welding was used as
seam were acquired after removing the slag on the surface. Eq. (1) has
welding material. The preset welding parameters are listed in Table 1.
been modified to better calculate the spatter loss coefficients of un-
Please note that an improper parameter set (W0) was intentionally
derwater wet FCAW at different parameters as follows
applied to produce more spatters.
MS (MS M1 )
100%
MS (2)
3. Results and discussion
where M1 and M2 represents the weight of substrate before and after
3.1. Spatters appearance welding respectively; MS represents the weight of pure metal provided
by the consumed wire. Please note that M1 and M2 are easy to be
Typical spatter appearances obtained under water and in air are measured; while MS cannot be measured directly, which therefore is the
compared in Fig. 2 (test case W3 and A1). Red lines (about 5 mm key to obtaining accurate .
away from bead edge) are drawn in Fig. 2 to emphasize the dis- First, MW is defined as the total weight of consumed flux-cored wire
tribution difference. For the weld seam acquired in water as shown in including the flux core and steel tape. The length of consumed wire can
Fig. 2(a) and (b), most of the generated spatters cannot weld with the be calculated using wire feed rate and recorded welding time. The unit
substrate. The spatters far away from the seam could be easily re- weight per meter of the wire is measured as about 18.33 g/m. Thus, the
moved. Several large spatters next to the wed seam were firmly value of MW can be obtained. Second, through measuring sample wire,
welded on the substrate surface. Please note that, almost all of these it is found that the weight of steel tape accounts for 77% of the flux-
small spatters did not weld with the substrate except some small cored wire; the weight of metal powder inside the flux occupies about
spatters with diameter less than 0.5 mm observed in the arc crater 10% of the total flux. Therefore, the weight of pure metal provided by
area (marked by a red circle), as shown in Fig. 2(a). According to the consumed wires roughly equals 79.3% (77%+23%*10%) of the
Fig. 2(b), the generated spatters under water were distributed in a total weight of the flux-cored wire. The following equation is used.
rather narrow range (5 mm to the seam edge) and with large size. By
MS MW * 0.793 (3)
comparison, the spatters generated in air have a much wider dis-
tribution range; amount of spatters are found to be farther than 5 mm Before measuring M2 , the substrate was dried; the slag and spatters
away from the bead edges. on the plate were removed as much as possible. These measures ensure
The experiments were conducted using the same material and that the value of M2 M1 approximately equals the weight of deposited
welding parameters, but only under different surroundings. Therefore, metal provided by the consumed wires. Then, the value of
it is thought that the above differences of spatter loss appearance and MS (M2 M1) equals the weight of metal provided by the wires that
are not deposited on the plate, i.e. the generated spatters.
Table 1 The cleaned beads are shown in Fig. 3. It is obvious that it is difficult
Preset welding parameters. to clean up all the attached FCAW spatters generated in air or under
No. Arc voltage Welding Wire feed Welding speed Surrounding water, because some spatters are either fused together with the sub-
(V) Current*(A) rate (m/min) (mm/min) strate or are surrounded by some irregular shrinking area. These welded
spatters lead to the increase of M2 ; the value of MS (M2 M1) is
W0 33 215 2.9 200 Water
smaller than the actual weight of generated spatters. The measurement
W1 24 190 2.4 200 Water
W2 24 200 2.6 200 Water
results are listed in Table 2. The measured values indicate that welding
W3 24 220 3.0 200 Water parameters have a significant effect on spatter generation during un-
W4 24 240 3.3 200 Water derwater wet FCAW.
W5 24 260 3.7 200 Water With improper welding parameters for experiment W0, the spatter
W6 24 280 4.2 200 Water
loss coefficient is up to 60%. While comparing the experiments from
A1 24 220 3.0 200 Air
W1 to W6, the spatter loss coefficients decrease significantly and

123
C. Jia, et al. Journal of Manufacturing Processes 40 (2019) 122–127

Fig. 2. Spatter loss appearance and distribution in different surroundings.


((a) and (b) show the underwater weld seams before and after cleaning; (c) and (d) show the weld seam in air before and after cleaning).

regularly. As shown in Fig. 4, the measured spatter loss coefficients Table 2


decrease almost linearly along with the increasing welding current. The Spatter loss coefficients at different currents.
minimum spatter loss coefficient of 14.6% (test case W6) is less than No. M1 (g) M2 (g) MS (g)
half value of the maximum one of 32.9% (test case W1). It could be
recognized that the increasing welding current alleviated the total W0 587.4 593.4 15.1 60.3%
spatter loss significantly. W1 611.7 623.1 17.0 32.9%
W2 620.7 634.4 18.4 25.5%
W3 603.0 620.1 21.3 19.7%
W4 573.1 592.4 23.3 17.2%
3.3. Spatter distribution features W5 588.3 610.6 26.9 17.1%
W6 630.4 656.2 30.2 14.6%
The unique accumulation of spatters during underwater wet FCAW
bring about negative effects on bead formation. First, overlap defect
occurred along with the generated large-size spatters. Second, the interactions between liquid spatter and weld pool. On one side, the
spatters near weld bead significantly influenced the slag coverage and spatter loss reduces the total amount of liquid metal in weld pool.
bead formation. Consequently less deposited metal solidified with obvious shrinking in
Typical overlap defect marked with red circle 3 is shown in Fig. 5. the direction perpendicular to welding direction (red circle 2 in Fig. 5).
More spatters congregated at the toe of the bead increase the occur- On the other side, the slag formation changed due to the joint actions by
rence of this defect. Also, due to the lack of slag coverage on the bead, weld pool and adjacent spatter (close to weld toe). A groove between
spatters directly contact with the bead and get welded on it more easily. them was formed forcing the slag gathering in it, as shown by red cir-
The overlap defect deteriorates the consistence of the bead appearance. cles 1 and 2 in Fig. 5. Furthermore, the squeezing effect from spatter
The negative effect on slag and bead formation could be due to the and slag limited the outward expanding of weld pool, which probably

Fig. 3. Cleaned beads appearance welded under different welding currents.

124
C. Jia, et al. Journal of Manufacturing Processes 40 (2019) 122–127

Fig. 4. Spatter loss coefficients variation along with increasing welding current. Fig. 6. Schematics of the spatter influences on slag formation.

deteriorated the weld bead shrinkage. It is believed that when a spatter 3.4. Spatter loss improvement
dropped at a certain place, i.e. neither too close to nor too far away
from the weld bead, the above-mentioned slag formation changes could It has been found that two methods are effective in reducing the
happen. For current experiments, the distance between the spatter and spatter loss during underwater wet FCAW. First one is to select proper
the weld boundary is about 1 mm (Fig. 5). welding parameters for improving welding process stability. Second
Fig. 6 gives a possible explanation of the spatter influences on slag one is to apply higher welding current.
formation. It was thought that the spatter dropped down first on the Fig. 7 shows the electrical signals of a typical arc extinguishing
front-left of the welding area with certain distance. Since the molten period. The welding current dropped sharply to nearly 0 A; while the
pool was always surrounded by slag (as shown by the slag distribution arc voltage increased rapidly to 80 A (open-circuit voltage). Since the
in Fig. 2(a)), when the molten pool moved near the spatter, the slag at arc extinguishing aggravates the spatter loss through several ways, it is
the molten pool boundary contacted the spatter first, as illustrated in effective to reduce the spatter generation by maintaining a stable arc
Fig. 6(a). The spatter, which has been fixed to the substrate, prevented burning. The stability of welding process can be guaranteed by applying
the slag from distributing outwards. Along with more consumed wire, proper welding parameters. As shown in Table 3, the frequency of arc
more slag accumulated near the spatter. The accumulated slag solidi- extinguishing has been calculated according to its electrical signals. The
fied quickly as it was closer to the surrounding water. Thus, the soli- low frequency indicates that the arc burning was rather stable under
dified slag and the spatter jointly limited the outward expansion of the proper welding parameters. Thus the negative effects on spatter gen-
molten pool at this place, as shown in Fig. 6(b). The shrinkage ampli- eration brought about by the arc extinguishing can be avoided by ap-
tude of the weld bead was measured from 1 to 1.5 mm. This range in- plying proper welding parameters.
dicates the accumulated slag can invade into welding area for about From another point of view, the corresponding standard deviations
1.5 mm. It is thought that even if the slag invades further, it can be in Table 3 indicate that the welding process stability did not experience
melted by the arc and be mixed into the weld pool and cannot affect the significant improvement along with the increased welding current.
width of the molten pool. As shown in Fig. 6(c), once the molten pool Except the W0 experiment, fluctuations of welding current and arc
passed by the spatter, the slag distribution was off the limit. The width voltage of other experiments are not obvious. Furthermore, Fig. 8 shows
of the molten pool recovered accordingly. the probability distribution of obtained electrical signals. Please note
that, two expressions are employed for each set of data. Fig. 8(a) and (c)

Fig. 5. Typical defects and negative effects caused by congregated spatters.

125
C. Jia, et al. Journal of Manufacturing Processes 40 (2019) 122–127

deviation statistics in Table 3. However, overall the welding current


signals of experiments (W1-W6) show low concentration ratio in-
dicating obvious fluctuations; the fluctuation ranges are almost the
same with minor differences. From Fig. 8(c), it could be found that the
arc voltage was controlled well fluctuating in a narrow range by the
employed power source. While from Fig. 8(d), although the probability
distribution differences of high arc voltage are amplified, no obvious
rules could be drawn. Therefore, the welding process stability of ex-
periments (W1-W6) is equal.
According to the calculated spatter loss coefficients in Table 2, the
increasing welding current led to significant drop of spatter loss coef-
ficient. Therefore, the increasing stability should not be considered to
be the main reason for reducing spatter loss with elevated welding
Fig. 7. Electrical signals of typical arc extinguishing. current.
There are mainly three possible explanations for this spatter loss
reduction. Firstly, the droplet transfer mode could be changed. It has
Table 3
been observed that the percentage of repulsive globular transfer mode
Arc extinguishing frequency and standard deviation at different parameters.
decreased while the percentage of surface tension transfer mode in-
No. Standard deviation of Standard deviation of Frequency of extinguishing creased by elevating the current. Guo et al. (2016) also observed similar
Current Voltage (HZ) transfer patterns [9]. With less globular transfer and more surface
W0 154.553 18.635 3.063 tension transfer, the spatter loss decreased accordingly. Secondly, re-
W1 80.368 4.102 0.068 sistance forces blocking the droplet transfer could be decreased. The
W2 73.897 3.202 0.034 decomposition of flux and vaporization of surrounding water can be
W3 84.125 3.199 0.000 increased by elevating the current. Consequently, the volume of main
W4 95.542 3.434 0.034
bubble increases and the distance between the bubble wall and the
W5 84.933 3.072 0.000
W6 91.017 3.178 0.000 droplet is enlarged. The enlarged distance makes it difficult for the
bubble wall to approach the droplet, and therefore, the buoyancy acting
upon the bubbles is less likely to resist the droplet transfer. Thirdly,
show the original probability distribution curves; Fig. 8(b) and (d) are higher welding current may produce bigger bubbles, which could
obtained via taking the logarithm of the above two curves to amplify provide large space for droplet transfer and arc burning. Thus, the
the differences in the low and high probability area. The unstable welding process will be less influenced by the surrounding water; less
welding process of W0 experiments could be easily recognized from the droplet transfer is changed into spatters. It was thought that, stable arc
curves. burning with less extinguishing and droplet transfer with less spatters
It could be seen that W2 has highest probability distribution and both depended on correct matching the droplet transfer and bubble
narrowest variation range. This is consistent with the standard evolution, which are closely related to parameters selection.

Fig. 8. Probability distribution of current and voltage at different parameters. (a) and (b), (c) and (d) are based on the same data with different vertical axes.

126
C. Jia, et al. Journal of Manufacturing Processes 40 (2019) 122–127

4. Conclusions References

Generated spatters loss during underwater wet FCAW were in- [1] Rowe MD, Liu S. Recent developments in underwater wet welding. Sci Technol
vestigated through experiments and analysis. Main conclusions are Weld Join 2001;6:387–96.
[2] Maksimov SY, Machulyak VV, Sheremeta AV, Goncharenko EI. Investigation of
summarized as follows: influence of microalloying with titanium and boron of weld metal on its mechanical
properties in underwater welding. Paton Weld J 2014:6–7. 76-79.
(1) Most of the spatters generated during underwater welding have [3] Łabanowski J. Development of under-water welding techniques. Weld Int
2011;25:933–7.
large sizes and tend to congregate near the seam in a narrow range. [4] Cheng F, Zhang S, Di X, Wang D, Cao J. Arc characteristic and metal transfer of
The spatter loss coefficients can be measured based on calculating pulse current horizontal flux-cored arc welding. Trans Tianjin Univ 2017;23:101–9.
the total pure metal from flux-cored wire. [5] Zhu Z, Fan K, Liu H, Ma G. Characteristics of short-circuit behaviour and its influ-
encing factors in self-shielded flux-cored arc welding. Sci Technol Weld Join
(2) The congregated spatters seriously influence the evenness of weld seam 2016;21:91–8.
and cause weld bead shrinkage and overlap defect.Accumulated spatters [6] Li K, Gao H, Li H, Gong S. Droplet rebounded spatter in dry hyperbaric gas metal arc
invade into the formation room of slag and weld bead, which further welding process. Int J Adv Manuf Technol 2014;74:693–8.
[7] De Meneses VA, Gomes JFP, Scotti A. The effect of metal transfer stability (spat-
limits the width of molten pool due to the squeezing effect. The spatter
tering) on fume generation, morphology and composition in short-circuit MAG
congregation leads to increasing occurrence probability of overlap. welding. J Mater Process Technol 2014;214:1388–97.
(3) According to the measured values, the spatter loss coefficients sig- [8] Liu HY, Li ZX, Li H, Shi YW. Study on metal transfer modes and welding spatter
nificantly decreased along with the increasing welding current.By se- characteristics of self-shielded flux cored wire. Sci Technol Weld Join
2013;13:777–80.
lecting proper parameters, the occurrence of extinguishing can be sig- [9] Guo N, Xu C, Guo W, Du Y, Feng J. Characterization of spatter in underwater wet
nificantly reduced. It was deduced that spatter loss can be effectively welding by X-ray transmission method. Mater Des 2015;85:156–61.
improved by elevating the welding current level, in which case droplet [10] Chu W-H, Tung P-C. Development of an automatic arc welding system using a
sliding mode control. Int J Mach Tools Manuf 2005;45:933–9.
transfer mode, droplet transfer resistance and bubble shielding could [11] Zhang Y, Jia C, Zhao B, Hu J, Wu C. Heat input and metal transfer influences on the
contribute. weld geometry and microstructure during underwater wet FCAW. J Mater Process
Technol 2016;238:373–82.
[12] Jia CB, Zhang Y, Zhao B, Hu JK, Wu CS. Visual sensing of the physical process
Acknowledgement during underwater wet FCAW. Weld J 2016;95:202s–9s.

The authors are grateful to the financial support for this project from
the National Natural Science Foundation of China (Grant No. 51675310).

127

You might also like