Introduction Global competition in manufactur-ing demands welding processes forhigh efficiency and high quality. Con-ventional welding processes often donot satisfy such needs in modern man-ufacturing. Manufacturing industriescontinuously look for novel arc weld-ing processes to increase or improvewelding productivity/quality (Refs.1–4) as arc welding is still and will con-tinue to be the most adopted processfor welding.

Gas metal arc welding (GMAW) andgas tungsten arc welding (GTAW) aretwo major arc welding processes wide-ly used in manufacturing industries.Continuous efforts have been made toimprove their productivity withoutcompromising the quality. Many newmodifications have been invented, in-cluding tandem GMAW, plasma-GMAW, multiwire GMAW, and GTAW-GMAW (Refs. 5–9). Despite theirdemonstrated advantages, all of themare based on the conventional arcingmechanism. Each electrode only estab-

lishes an arc with the workpiece al-though there may be multiple elec-trodes and parallel arcs. The weld poolis not only heated by several arcs, butalso at a common terminal where arcsare coupled. All these modifiedprocesses share a fundamental princi-ple in the arcing mechanism; the work-pieces to be heated by all the arcs inthe system and the heat input are cou-pled with the mass deposition pro-duced. Novel modifications are stillneeded to increase the depositionwithout increasing the heat input in acertain range. In order to separately control themass and heat inputs, the divided-arcwelding processes have been proposedat the University of Kentucky (Refs.10–12), and it has gained applicationsin various industries including auto-motive and shipbuilding (Ref. 13)among others (Refs. 14–17). Researchfrom Wu et al. (Ref. 14) demonstratedthat double-electrode gas metal arcwelding (DE-GMAW) can enhance thewelding speed compared with conven-tional GMAW. Shi et al. (Ref. 15) ana-lyzed the DE-GMAW as a high-speedwelding method for lap joints on zinc-coated steel sheets for which the zinccoating is less melted by the DE-GMAW than by GMAW with the samedeposition. Lu et al. (Refs. 18, 19) pro-posed and developed DE-SAW as avariant/modification of the DE-GMAW to increase welding speed. This is because the DE-GMAW di-vides the arc and establishes arcs withboth the workpiece and another added

WELDING RESEARCH

FEBRUARY 2016 / WELDING JOURNAL 47-s

SUPPLEMENT TO THE WELDING JOURNAL, FEBRUARY 2016Sponsored by the American Welding Society and the Welding Research Council

Effect of Torch Height on Arc Stabilityin Divided­Arc Processes

The distance between the contact tube and the workpiece has a significantimpact on the arc behavior in double­electrode gas metal arc welding

BY S. J. CHEN, L. ZHANG, G. Q. MEN, Y. X. SONG, S. SU, AND L. W. WANG

ABSTRACT In conventional arc welding processes, an electrode only establishes an arc with theworkpiece. As such, the current through the electrode and workpiece is the same. As hasbeen demonstrated, this mechanism fundamentally restricts the ability to decouple theheat input and deposition. In order to separately control the mass and heat inputs, double­electrode gas metal arc welding (DE­GMAW), which modifies the GMAW by also establish­ing a second arc from the electrode to an added second electrode, and arcing­wire gastungsten arc welding (GTAW), which modifies the GTAW by establishing a second arc fromthe electrode to the welding wire, have been proposed. Since both processes break theconventional arc, this paper refers to them as the divided­arc welding process. While thedivided­arc welding process promises a new mechanism to provide an ability to decouplethe mass and heat inputs as will be demonstrated in this paper, maintaining a stable arc di­vision is the key to ensuring the promised decoupling ability from this novel arc weldingprocess. Studies are needed to determine how the ability to maintain the arc division isaffected by typical variations in manufacturing conditions. Unfortunately, no such studieshave been reported. In this paper, the effect from the variation in the distance between thetorch and workpiece (simply the torch height) is experimentally studied. Since the variationin the torch height is probably the most commonly encountered variation in welding condi­tions, such a study is significant in helping to transition the divided­arc welding processesfrom laboratory to manufacturing.

KEYWORDS • GMAW • GTAW • Arc Behavior • Metal Transfer • Divided Arc

S. J. CHEN (sjchen@bjut.edu.cn), L. ZHANG, G. Q. MEN, Y. X. SONG, S. SU, and L. W. WANG are with the Engineering Research Center of Advanced Manufac­turing Technology for Automotive Components, Ministry of Education, Beijing University of Technology, Beijing, China. L. ZHANG is with Hebei Key Laboratoryof Material Near­Net Forming Technology, Hebei University of Science and Technology, Shijiazhuang, Hebei, China.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 47

electrode (Refs. 10, 12). As such, thearc is no longer established betweenthe electrode and workpiece. Part ofthe current that would otherwise flowinto the workpiece now flows to theadded electrode as a result of the arcdividing. The resultant process, i.e.,DE-GMAW, is thus a modification ofGMAW through arc dividing. Another modification through arc di-viding has been made for the GTAWprocess, resulting in arcing-wireGTAW (Ref. 12). The arc dividing al-lows a second arc to be established be-tween the tungsten electrode and thefiller metal, which melts the wire at ahigh speed while the current on theworkpiece is still controlled at the de-sired level. That is, while the heat in-put is controlled by the current be-tween the tungsten electrode andworkpiece, the deposition speed is in-creased/controlled by the arc dividedfrom the original gas tungsten arc.

Since the currents through the work-piece and the filler metal are freely ad-justable, the heat input and mass dep-osition are separately controlled. As can be seen, the arc dividing re-lieves the inherent coupling betweenthe heat input and deposition in con-ventional arc processes, resulting inthe advantages of high efficiency, highspeed, and low heat input with desir-able coupled controls. However, theseadvantages depend on the success inthe arc dividing. The arc dividing hasto be successfully maintained againstpossible variations in manufacturingconditions that may occur in industri-al settings. Unfortunately, despite ex-tensive various efforts on the divided-arc processes (Refs. 20–23), effectsfrom possible variations in manufac-turing conditions on the successful op-erations of the divided-arc processeshave not been addressed. Withoutstudying the effects from manufactur-

ing variations, the applicability of thedivided-arc processes would not be en-sured to actually deliver their distin-guished advantages for decoupledmass and heat input control. This pa-per studies the effect of the most en-countered variation in welding condi-tions — the variation in the distancefrom the torch to the work, which willbe referred to as to the torch height. In particular, past studies have fo-cused on modeling, monitoring, andcontrol of the divided-arc processes.No mention has been made of the ef-fect on the arc stability of coupled arcs(Refs. 10, 12, 20, 22) from any weldingcondition variations, including that ofthe torch height. Gas metal arc and gastungsten arc welding are the results ofthe effects from many parameters thatare coupled and interacted. These pa-rameters also affect the success of thearc dividing, thus the stability of thecoupled arcs, the controls on the heatinput and mass deposition, and there-by the quality of produced welds. Inmany studies, especially in DE-GMAW,the torch height is abnormally highcompared with conventional GMAW.For manufacturing applications, thetorch height may change significantlydue to the uneven surface of the work-piece. In this investigation, experi-ments were conducted and analyzed tostudy how the torch height affects thearc stability in two major divided-arcwelding processes, DE-GMAW and arc-ing-wire GTAW, to help transitionthese processes to manufacturing applications.

Divided­Arc WeldingProcesses Double electrode GMAW (Refs. 10,20, 21) and arcing-wire GTAW (Ref.12) are presently the two major divid-ed-arc welding processes. This paperwill focus on these two processes,whose principles are discussed below.

DE­GMAW

Double electrode gas metal arcwelding (DE-GMAW) was proposed atthe University of Kentucky in 2004(Ref. 10) as a modification to theGMAW process. It separately controlsthe heat input and deposition byadding a second electrode (noncon-

WELDING RESEARCH

WELDING JOURNAL / FEBRUARY 2016, VOL. 9548-s

Fig. 1 — DE­GMAW system.

Fig. 2 — Arcing­wire GTAW system.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 48

sumable electrode) to bypass part ofthe current, which otherwise wouldflow to the workpiece. As shown inFig. 1, the main power supply, themain welding gun, and the main arcare the same as in the conventionalGMAW welding process. The bypasselectrode, a tungsten electrode, isadded to provide the condition to es-tablish a bypass arc with the wire,which is the electrode in GMAW. Thewire serves as a common positive ter-minal for the two power supplies andthere are two currents flowing into thewire: GMAW current (through theworkpiece) and GTAW current (bypasscurrent). These two currents are“summed” to melt the wire and thendivided at the end of the wire, which isthe positive terminal of the two arcs.The bypass gun and workpiece areconnected by the negative terminal of

the bypass power supply, and mainpower supply, respectively. Gas metalarc welding current and GTA currenttogether form the current to melt thewire, but only the GMAW current con-tributes directly to heat the workpiece.With the DE-GMAW process, GMAcurrent and GTA current can be ad-justed individually such that the depo-sition and heat input can be adjustedindividually during welding.

Arcing­Wire GTAW

Gas tungsten arc welding is typical-ly used for precision joining. However,its productivity in depositing fillermetal is much lower than that ofGMAW because the wire is not melteddirectly by an arc as in GMAW. In or-der to increase the deposition speed inGTAW, the wire melting in cold-wire

GTAW and hot-wire GTAW was stud-ied by Chen et al. at the University ofKentucky, and an innovatively modi-fied GTAW, namely the arcing-wireGTAW process, was proposed in 2012(Ref. 12). As shown in Fig. 2, the arc-ing-wire GTAW process is formed byestablishing a second arc between thetungsten electrode and the wire suchthat the wire becomes an arc terminalthat can melt the wire at high speedssimilarly to GMAW. As a result, thereare two coupled arcs: the arc betweenthe tungsten electrode and the work-piece, which can be considered theoriginal gas tungsten arc, and the sec-ond arc between the tungsten andwire, which can be considered as an arcdivided from the original GTA. Withthe second arc, the wire melting is nolonger dominated by the weld pool andresistance heat. As a result, the fillermetal can not only be melted at highspeeds by the second arc, but also at adesirable, adjustable speed by adjust-ing the current through the wire. Thisdesirable adjustability does not de-pend on the amperage of the currentthrough the workpiece, which directlycontrols the heat input and penetra-tion on the workpiece. In particular, the tungsten elec-trode (GTA torch) serves as a commonnegative terminal to form the two di-vided arcs. The GTA current is the cur-rent to control the direct heating ofthe workpiece and the wire current(simply the GMA current hereafter be-cause the wire is an arc terminal as inthe GMAW process) is the current tocontrol the direct melting of the wire.The heat and mass inputs are thusseparately controlled and the high-speed deposition in GTA becomes sim-ilar to GMAW.

Experimental Procedure A divided-arc process has at leasttwo arcs coupled to each other, al-though there may be more than twodivided arcs (Refs. 23, 24). This studytries to find how the torch height af-fects the arc behaviors in the divided-arc processes and the stability of thearc dividing. To this end, the experi-mental system will be able to recordnecessary signals/information to ob-serve and analyze the arc behaviors inthe two major divided-arc weldingprocesses being studied.

WELDING RESEARCH

FEBRUARY 2016/ WELDING JOURNAL 49-s

Fig. 3 — Experimental system for arcing­wire GTAW.

Fig. 4 — High­speed images of DE­GMAW in Experiment 1.

Table 1 — Experimental System Components

Equipment and Accessories Model, Material, or Size

GTA power supply Time WSM­400 GTA torch WP­18 GMA power supply Kemppi FastMig 350 GMA gun Kemppi MMT 30W Diameter of tungsten 5 mm High­speed camera IDT motion Y4 Shielding gas Pure Argon

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 49

In the experimental system, GTA isestablished using a DC-CC (direct cur-rent-constant current) power supplyand GMA is established using a syner-gically controlled DC power supply inwhich the welding current and voltageare adjusted in relation to the wirefeed without separate settings. The experimental system shown inFig. 3 is the setting for the arcing-wireGTAW. There are four sensors to de-tect the voltages and currents of thetwo divided-arcs and a high-speedcamera to record the arc images. Thehigh-speed camera (without an opticalfilter) records the arc behaviors at3000 frames per second (f/s), whilethe current and voltage sensors sam-ple at a higher frequency. However,the timing for the images from thehigh-speed camera and electrical arc

signals were still synchronized in orderto observe and analyze the behaviorsof the divided arcs with informationfrom multiple sources. Table 1 showsthe details of the experimental system. With the divided-arc weldingprocesses, there are many parametersaffecting the behavior of the dividedarcs. However, many of them may befixed at their experimentally deter-mined values. For example, the posi-tion of the bypass torch (or the fillermetal for the arcing-wire GTAW) in re-lation to the main torch may be opti-mized and retained by a fixture. Assuch, the focus should be on parame-ters that may easily be subjected tovariations. A main parameter that de-termines the stability of the coupledarc and is easily subjected to variationsis the torch height — the distance

from the tip of the contact tube to theworkpiece in DE-GMAW or the dis-tance from the tip of the tungsten tothe workpiece in the arcing-wireGTAW. Hence, experiments were con-ducted to examine how the torchheight affected the stability of the arcdividing, as shown in Table 2. The ex-planation and detailed parameters/conditions are given below. In Table 2, GTA current (IGTA) refersto the current between the tungstenelectrode and the workpiece in the arc-ing-wire GTAW or the bypass currentbetween the main wire and the bypass(tungsten) electrode in DE-GMAW. TheGMA current (IGMA) refers to the cur-rent between the wire and workpiece inthe DE-GMAW or the current betweenthe tungsten and filler metal in the arc-ing-wire GTAW. Because of the use ofthe synergic DC power supply, the GMAcurrent is not a constant parameter butcontrolled by the wire feed speed. Ofcourse, the change in other conditionswill affect the actual GMA current, as inconventional GMAW. In the experi-ments to be performed, the GTA cur-rent is fixed at 100 A and the wire feedspeed varies between two levels. Thewire for both the DE-GMAW and arc-ing-wire GTAW is low-carbon steel witha 1.2 mm diameter. The shield gas ispure argon. The flow rate was 12 L/min.All the experiments were conducted asbead-on-plate. For the arcing-wire GTAW, the wirewas fixed at 3 mm below the tungstentip at the tungsten axis. The wire wasfed at 40 deg with the tungsten behindthe torch without optimization. Forthe DE-GMAW, the actual extension ofwire was variable as the wire feedspeed was a variable parameter (Table2), but the distance between tungstenand contact cube was fixed at 7 mm,and the tungsten was fixed at 30 degwith the wire.

Experimental Results

DE­GMAW

There were three experiments inDE-GMAW. The GTA (bypass arc) cur-rent (IGTA) was fixed at 100 A and theother two parameters (wire feed speedand torch height) were both changed.In particular, the wire feed speed wasdecreased from 2.3 to 1.2 m/min in

WELDING RESEARCH

WELDING JOURNAL / FEBRUARY 2016, VOL. 9550-s

Fig. 5 — Voltage waveforms of DE­GMAW in Experiment 1.

Fig. 6 — High­speed images of DE­GMAW in Experiment 2.

Table 2 — Parameters of Divided­Arc Welding Experiments

IGTA Wire Feed Exposure

Welding Experiments Torch GTA Speed (m/min) Time Process Height (mm) Current (A) (s)

#1 20 100 2.3 200 DE­GMAW #2 15 100 2.3 200 #3 15 100 1.2 160 #4 10 100 2.3 200 Arcing­wire #5 6 100 2.3 200 GTAW #6 6 100 1.3 160

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 50

Experiment 3; the torch height (thedistance from the tip of contact tubeto workpiece) decreased from 20 to 15mm in Experiments 2 and 3. In Experiment 1, the wire feedspeed was 2.3 m/min, which was inthe common range for industrial man-ufacturing. The metal transfer for con-ventional GMAW under this wire feedspeed and wire extension is the shortcircuiting transfer. As can be seen inFig. 4, after the GTA (bypass arc) wasadded, such that the process changesfrom GMAW into DE-GMAW, the met-al transfer changed to a free-flight

globular transfer. In this experiment,the torch height was approximately 20mm, slightly longer than typical inconventional GMAW. As can be seen,the anodes of the two arcs are both onthe wire. The heat melting wire wasthus determined by the sum currentsof GMA and GTA. As can be seen fromFig. 4, the process and the arc dividingwere stable. While the GTA (bypass arc) can beestablished and maintained as expect-ed, its arc column was bent, becomingcurved as can be seen in all images inFig. 4. The coupling between the GMA

and GTA is apparent. Although theGMA was established between thewire and workpiece as conventionalGMAW, the GMA deviated to the tung-sten. There was a bright zone that in-dicated the overlapping of the two arccolumns. The GMA and GTA were di-vided at the end of this overlappingzone. The role of the wire as the com-mon anode changed the wire meltingspeed and the mode of the metaltransfer. In the series of images in Fig.4, the droplet gradually grew. When itsdiameter became large enough, it wastransferred to the weld pool under theinfluence of the gravity and the uniqueelectromagnetic forces associated withdivided current flows despite the sur-face tension. In the entire process ofmetal transfer, the arcs were stableand the metal transfer was smooth.There were no observations that thetwo coupled arcs could affect the sta-bility in this experiment. For Experiment 1, the voltagewaveforms were given in Fig. 5. Thedesirable stability of the DE-GMAWarc can be seen in its relatively smoothvoltage waveforms. The amplitude ofGMA voltage was approximately 30 V(CH2), which was higher than typicalin conventional GMAW in this condi-tion. The simultaneous effects fromthe two currents melted the wire at ahigher speed such that the arc lengthbecame longer. As such, a higher wiretip position was maintained. The GTAvoltage was approximately 20 V (CH4),as the position of the tungsten relativeto the GMA torch was not changed.The waveform for this voltage was alsosmooth without fluctuations, suggest-ing that the arc length of the GTA isstable. The position of the wire tip,from which the GTA is established tothe stationary tungsten, is thus stable.The stable position of the common an-odes as well as the stationary cathodes(workpiece and tungsten electrode)can ensure the stability of the two, di-vided arcs. In Experiment 2, all the parameterswere the same as Experiment 1, exceptthe torch height was decreased from20 to 15 mm. Comparing Fig. 4 andFig. 6, there were some significant dif-ferences in the behaviors of two cou-pled arcs, especially in the mode of themetal transfer. The transfer modechanged from the globular transferback to the short circuiting transfer.

WELDING RESEARCH

FEBRUARY 2016 / WELDING JOURNAL 51-s

Fig. 8 — High­speed images of DE­GMAW in Experiment 3.

Fig. 9 — Voltage waveforms of DE­GMAW in Experiment 3.

Fig. 7 — Voltage waveforms of DE­GMAW in Experiment 2.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 51

As can be seen in Fig. 6, the processbecame unstable just by changing thetorch height. The droplets grew up-ward, fell, and touched the weld pooland exploded at the end of the metaltransfer process. The whole processwas similar to the CO2 weldingprocess. Careful analysis and study onExperiment 2 showed that the changeswere so obvious, not only on the modeof metal transfer, but also on the be-haviors of arcs. The GMA establishedbetween the wire and workpiece devi-ated to the left by GTA (bypass arc).The GTA was not established directlybetween the wire and the tungsten inthe last two images in Fig. 6. The an-ode of the GTA looked like it was es-tablished on the workpiece. The GTA(bypass arc) heated the workpiece di-rectly and decreased the heat thatmelted the wire. The wire melting rate

was reduced and GMA current adaptedthat automatically. The actual exten-sion of the wire was longer than thatin Experiment 1. The behavior of GTAand GMA were both changed, leadingto an unstable process for DE-GMAW. The significant changes in compari-son with Experiment 1 would also bereflected in the voltage waveforms.Figure 7 shows the voltage waveformsof Experiment 2. The most obviouschange was that the waveforms wereno longer smooth, but had many largefluctuations. The GMA voltage wave-form (CH2) can match with the wave-form in the short circuiting transfer(Ref. 25). During the short circuittime, the GTA deviated further to theleft in the fifth image in Fig. 6. Theelectromagnetic forces should be re-sponsible because the two currentsflowed in the opposite directions. In

this case, the electromagnetic forcescan push the two arcs away further.When a short circuit occurred, theGMA current increased sharply andreached its greatest at the end of theshort circuit. In the meantime, theGTA always existed. The magnitude ofthe electromagnetic force acted on theGTA depends on the GMA current.The GTA thus deviated farther at theend of the short circuit time. This de-viation caused the voltage of the GTAto further increase. The GTA voltagewaveform (CH4) thus had about 50 Vat the end of short circuit time. At thistime, the droplet transferred to theworkpiece and the GMA reignited,while the GTA was still established be-tween the tungsten and workpiece.The voltage of the GTA was actuallythe sum of GTA and GMA, which wasabnormally high. In Experiment 3, in order to makethe GMA maintain a certain arc lengthand the two arcs to be in a stable statedespite the reduced torch height, thewire feed speed was decreased from2.3 to 1.2 m/min, while other parame-ters were kept the same as in Experi-ment 2. The exposure time was re-duced to 160 ms to further reduce theinterference of the arc in order to ob-serve the behaviors of the dropletsmore clearly. In this experiment,changing the wire feed speed broughtsome obvious changes. As shown bythe high-speed images in Fig. 8, theGMA current decreased and the heatmelting the wire also reduced. Thelength of the GMA was shorter thanthat in Experiment 2. The metal trans-fer mode was still the short circuitingrather than the globular transfer aswas intended by reducing the wire feedspeed. After the droplet transferred tothe workpiece, the GMA and GTAcould be resumed, respectively. Obser-vation of the series of images in Fig. 8showed that the GTA successfully es-tablished the tungsten and wire on re-suming. However, as the wire melted,forming a small droplet at the wire tip,the GTA (bypass arc) changed immedi-ately. It was no longer establishedfrom the tungsten to the wire directly.The arc column was bent and deviatedfrom the original direction to theworkpiece instead. The change wasvery obvious, especially when thedroplet was being detached and trans-ferred from the wire to the weld pool

WELDING RESEARCH

WELDING JOURNAL / FEBRUARY 2016, VOL. 9552-s

Fig. 12 — High­speed images of arcing­wire GTAW in Experiment 5.

Fig. 11 — Voltage waveforms of arcing­wire GTAW in Experiment 4.

Fig. 10 — High­speed images of arcing­wire GTAW in Experiment 4.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 52

as in the last image in Fig. 8. The GMAextinguished during the short circuit-ing time, during which the deviationof GTA was serious. In Figs. 4, 6, and8, there is a common phenomenon:the GTA was not a straight arc, but acurved arc that bent toward the work-piece. The stability of the GTA mustdetermine the stability of DE-GMAW.How to obtain/maintain a desirablebypass arc is the key to ensuring a suc-cessful arc dividing to maintain a sta-ble process.

Figure 9 is the voltage waveformsin Experiment 3. The significant dif-ferences were that the waveforms werenever smooth compared with Fig. 5.The GMA voltage waveform was a typ-ical short circuiting transfer voltagewaveform. The voltage was almost 0 Vduring the short circuiting time andnear 20 V in the open-circuiting time.The GTA voltage waveform was alsounstable. GMA experienced reestab-lishment and the cycle started overagain periodically. As such, the at-

tempt to resume the arc stability by re-ducing the wire feed speed without in-creasing the torch height failed.

Arcing­Wire GTAW

There were three experiments con-ducted using the arcing-wire GTAW. Thedistance between the tungsten and wirewas fixed at 3 mm. The main arc was aGTA and the second arc was a GMA.The GTA current (IGTA) was fixed at 100A. The major difference in the parame-ters was the torch height, which de-creased from 10 mm in Experiment 4 to6 mm in Experiments 5 and 6. The wirefeed speed decreased from 2.3 to 1.3m/min in Experiment 6. For Experiment 4, the high-speedimages are shown in Fig. 10. The GTA(main arc) was established betweenthe tungsten electrode and the work-piece. The GMA (the second arc) wasestablished between the tungsten andthe wire. In Fig. 10, the droplet grewuniformly when the wire was fed intothe GTA. It was similar to the cold wireGTAW, except there was a second arc.The current flowing into the tungstenelectrode was the sum of the GTA cur-rent and GMA current. The tip of thetungsten was the common cathodewhere the arc divided into two arcs. The directions of the two currentswere approximately the same. In theo-ry, the GTA and GMA (the second arc)should attract each other. However, itwas observed that the GTA deviated tothe left rather than toward the GMAon the right. To explain this phenome-non, we begin with the positive ionsemitted from the wire that must flowto the tungsten at high speedsthrough the GMA. These ions weighedmuch more than the electrons and col-lided with the particles in the GTA. Assuch, the ions in the GMA are impact-ed. In the meantime, any area in theentire workpiece could be the anodefor the GTA and the additional energyconsumption is less sensitive to thechange of the anode position on theflat workpiece surface. The GTA thustended to deviate to the left as can beseen in the macroscopic phenomenonobserved in Fig. 10. From the high-speed images in Fig.10, there was no evidence that GMAcould be established between the wireand workpiece, even when the droplettransferred to the workpiece, and even

WELDING RESEARCH

FEBRUARY 2016 / WELDING JOURNAL 53-s

Fig. 13 — Voltage waveforms of arcing­wire GTAW in Experiment 5.

Fig. 15 — Voltage waveforms of arcing­wire GTAW in Experiment 6.

Fig. 14 — High­speed images of arcing­wire GTAW in Experiment 6.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 53

when there was no arc between thedroplet and workpiece, and even whenthe distance between the droplet andworkpiece was only 1 or 2 mm. Asaforementioned, the GTA and GMAwould tend to attract each other. Forthe GMA, the effect from the GTA isminimal because only a small portionof the GTA is coupled with the GMAdue to the wide spread of the GTA.The impact from the GTA ions on theGMA ions is thus minimal. As a result,the GMA was stable as desired. Thedroplet transferred to the workpiece inthe mode of globular transfer. Howev-er, the arc behaviors were not affectedbecause the droplets did not affect thedistances between any arc terminalsand travel across any arc columns. Figure 11 details the voltage wave-forms for the GTA and GMA. Despitethat there were fluctuations in thewaveforms, they were much smootherthan those in Experiment 3. When thedroplet transferred to the workpiece,there was an approximate increase of 3V in the waveforms. However, the arcstability and arc dividing were not af-fected. Further, the GMA voltage wassmooth and was about 20 V. This fur-ther illustrated that the GMA was stable, as shown in the high-speed

images. While the torch height was an im-portant parameter for the stability ofthe DE-GMAW process, Experiment 4showed a stable arcing-wire GTAWprocess with a torch height of approxi-mately 10 mm. To verify whether thetorch height had a significant impacton the stability of arcing-wire GTAW,it was decreased from 10 to 6 mm inExperiment 5. Figure 12 shows high-speed imagesfrom Experiment 5. As can be seen,the whole process was stable. The GTAand GMA were established as desired.The wire was fed and melted, formingthe droplet. When the droplet becamelarge enough, it touched the work-piece. Then the droplet transferredinto the workpiece immediately. Themode of the metal transfer was similarto conventional short circuiting trans-fer, but there were some fundamentaldifferences. There was no phenome-non of short circuiting in the arcing-wire GTAW. Although the droplettouched the workpiece, the wire andworkpiece are not anode-cathode suchthat there was no current to flowthrough the bridging droplet. As such,while touching the droplet to theworkpiece in GMAW results in short

circuiting, which may generate an ex-plosion at the end of the short circuitprocess, for the arcing-wire GTAW,there was no explosion in the wholeprocess. The electromagnetic force could alsohave contributed to the observed stabil-ity of the arc dividing and the dividedarcs despite the reduced torch height.As aforementioned, because the GMAcurrent is approximately in parallel withthe GTA current, the electromagneticforces will tend to attract the two cur-rents to each other. Since the impactfrom the GTA ions on the GMA ions isminimal, this attraction will not be sig-nificantly neutralized. As a result, thereis a trend that the GMA is attracted tothe GTA such that the ionized particlesproduced by the GTA will improve theconductivity needed to maintain theGMA. As such, the stability of the GMAis enhanced. The stability of the arc di-viding and the arc-wire GTAW process isthus enhanced by the favorable wayhow the electromagnetic forces are pro-duced as in the arcing-wire GTAWprocess. The strong arc stability over-came the possible effect from the torchheight variation. The voltage waveforms from Experi-ment 5 are shown in Fig. 13. The GTAvoltage (CH2) was stabilized at 15 V,which was less than that in Experiment4, as the torch height was decreasedsuch that the arc length for the GTA wasreduced. The GMA voltage (CH4) wasalso stabilized at 12 V, also lower thanthat in Experiment 4. Careful compari-son with Fig. 10 showed that the re-duced arc length made the GTA moreconcentrated. As a result, the conductiv-ity of the GTA provided a conductivechannel needed to establish and main-tain the GMA increase. The GMA volt-age was thus reduced. In Experiment 3 for DE-GMAW, de-creasing the wire feed speed made themode of metal transfer less stable. Tocheck the possible effect from the re-duced wire feed speed on the stability inthe arcing-wire GTAW, the wire feedspeed was reduced from 2.3 to1.3m/min in Experiment 6. Experimen-tal results showed no effect on the sta-bility as can be seen from Fig. 14. In Fig.14, the exposure time was reduced to160 ms with the hope to reduce the in-terference of the arc and observe the arcbehavior more clearly. In this case, thewire further extended toward the GTA.

WELDING RESEARCH

WELDING JOURNAL / FEBRUARY 2016, VOL. 9554-s

Fig. 16 — Arc behaviors of DE­GMAW.

Fig. 17 — Arc behaviors of arcing­wire GTAW.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 54

The droplet was still transferred intothe workpiece after touching, but thedroplet was reduced. It appeared thatthe reduced distance from the wire tothe workpiece, due to the increased ex-tension of the wire, which was fed withan angle, reduced the gap for thedroplet to grow before touching theworkpiece. As a result, the droplettransferred into the workpiece at re-duced sizes as can be seen from Fig. 14in comparison with the droplets in Fig.10 and Fig. 12. The waveforms for thevoltages as shown in Fig. 15 also indi-cated the stability of the arcs.

Analysis and Discussion The major conclusion from thisstudy on the sensitivity of the arc di-viding stability to the torch height isthat the DE-GMAW is sensitive whilethe arcing-wire GTAW is not. DE-GMAW and arcing-wire GTAW areboth based on the arc dividing. How-ever, while the arc is divided at a con-sumable (wire) in DE-GTAW, the arc-ing dividing occurs at a nonconsum-able (tungsten electrode) in the arcing-wire GTAW. This difference causes thedifference in their sensitivity to thevariation in the torch height. In DE-GMAW, the position of the arcdividing root (wire tip) changes in rela-tion to the GMAW torch due to the con-sumable nature of the wire. The bypasselectrode/GTA torch is attached/fixed tothe GMAW torch. The relative position-ing between the root and destinationthus changes. When the torch height re-duces, the root would tend to move to-ward the tip because the tip of the wireis supposed to keep a constant distancefrom the workpiece in the constant volt-age (CV) mode. (A synergic control de-termines the voltage setting based onthe wire feed speed but the current isstill adjusted to maintain this voltage.The arc voltage, thus the arc length, isstill controlled similarly as in CV mode.)The main arc that provides the neededconductive channel to maintain the de-sired arc dividing at the wire tip wouldmove toward the wire tip. The destina-tion (bypass tungsten electrode) wouldbetter immerse the needed conductivechannel. In this case, reducing the torchin a certain range may benefit the stabi-lization of the arc dividing. However, the metal transfer associ-ated with the consumable wire compli-

cates the physical process. The dropletneeds to be detached from the wire tipby a sufficient electromagnetic force.When the wire feed speed is relativelysmall, the current is relatively small.The electromagnetic force that con-tributes to detaching the droplet is rel-atively small and would be insufficientto effectively detach the droplet by it-self. As a result, the droplet grows. Ifthe distance between the contact tipand the workpiece is relatively small,the droplet may touch the workpiece,forming the short circuiting. This canbe seen by comparing Fig. 16A with B.In this case, the main arc extinguishesand the condition for the arc dividingis no longer maintained. As a result,the workpiece becomes the anode toestablish the arc with the tungsten.The arc no longer divides at the wiretip. The process deviates from the de-sired DE-GMAW. Of course, the occurrence of theshort circuiting is the only cause thataffects the arc dividing. If the arclength of the main GMA is too short,the GMA would contract (image 5 inFig. 6) such that it would not be ableto immerse the bypass tungsten elec-trode. The DE-GMAW would still notbe maintained. As such, maintaining asufficient torch height and a sufficientarc length is needed to maintain thearc dividing in the DE-GMAW. For the arcing-wire GTAW, the caseis simpler. The “root” where the arc isdivided is stationary in relation to thewire where the “destination” resides.The length of the added arc is main-tained constant by the CV mode used.The relative position between the“root” and “destination” thus tends tobe maintained stationary — Fig. 17.The droplet formed on the wire wouldno longer extinguish the main arc orbreak the condition to maintain theneeded conductive channel for theadded arc, despite possibly touchingthe workpiece. Changing the torchheight appears to have limited affecton the successful maintainance of thearc dividing as long as the torch is notextremely low such that the wirepoints to the workpiece before it canimmerse into the GTA. The stability ofthe arcing-wire GTAW can thus be eas-ily maintained and is much less sensi-tive to the variation in the torchheight, as observed from Experiments4 to 6.

Conclusion This experiment studied and ana-lyzed the effect of the torch height onthe stability in two innovative divided-arc processes: DE-GMAW and arcing-wire GTAW. The authors found the fol-lowing: 1) The torch height has a signifi-cant impact on the arc behavior in DE-GMAW. Maintaining a stable (bypass)GTA means a successful arc dividing andstable welding process. When the torchheight is sufficient, the droplet maytransfer in globular transfer mode. Thebypass arc can be established stably be-tween the bypass tungsten electrodeand the main wire. Decreasing the torchheight may cause the short circuitingtransfer for the main GMA such thatthe condition to divide the arc at thewire tip is no longer maintained. Theprocess would switch between the suc-cessful arc dividing (desired DE-GMAW)and unsuccessful arc dividing (deviatingfrom the DE-GMAW), which would beunstable. 2) The torch height has less impacton the stability of the arcing-wireGTAW. The stationary position of thearc dividing root in relation to thewire, and the CV mode for the addedGMA that tend to maintain a con-stant distance between the arc divid-ing “root” and “destination,” arelargely responsible for this desirablestability. The independence of themain GTA from possible short circuit-ing between the wire and workpiecealso contributes to this desirable sta-bility and insensitivity to the varia-tion in the arc length.

This work is supported by the Nation-al Science Foundation of China(51375021) and the National Scienceand Technology Major Project(20142x04001171).

1. Pan, J. L. 2000. Prospects for weldingresearch in 21st century. China MechanicalEngineering 11(1): 21–25, 35. 2. Cui, X. M., Li, L. H., and Zhang, Y. H.2004. The research of high efficient weld-

WELDING RESEARCH

FEBRUARY 2016 / WELDING JOURNAL 55-s

Acknowledgments

References

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 55

ing processes. New Technology & NewProcess 7: 13. 3. CH, K. S. R. 1996. High depositionMAG welding: Used for welding bridgesand structures. Welding in the World 38:227–343. 4. Mendez, P. F., and Eagar, T. W. 2002.New trends in welding in the aeronauticindustry. Proceedings of 2nd Conference ofNew Manufacturing Trends, Bilboa, Spain. 5. Li, X. L., Huang, S. S., and Wu, K. Y.2008. Research of the high-speed doublewire pulse MIG welding. Power Electronics42(3): 39–40, 65. 6. Klaus-Jürgen, M., and Thomas, K.2002. Electrical effects and influencingquantities in the case of the hybrid plasma-MIG welding process. Welding and Cutting(2): 87–90. 7. Tusek, J. 1994. Submerged arc weld-ing with double and triple wire electrode.The International Journal of Joining of Mate-rials 6(3): 105–110. 8. Leng, X. S., Zhang, G. J., and Wu, L.2006. Analysis of twin-electrode TIG cou-pled arc pressure. Transactions of the ChinaWelding Institution 27(9): 13–17. 9. Song, Y. L. 2013. Research and devel-opment of high performance welding arc.Electric Welding Machine 343(3): 1–5. 10. Zhang, Y. M., Jiang, M., and Lu, W.2004. Double electrodes improve GMAWheat input control. Welding Journal 83(11):39–41.

11. Lu, Y., Zhang, Y. M., and Kvidahl, L.2013. Heat input reduction in fillet weld-ing using bypass and gap. Welding Journal92(12): 390–400. 12. Chen, J. S., Lu, Y., and Li, X. R., etal. 2012. Gas tungsten arc welding using anarcing wire. Welding Journal 91(10):261–269. 13. Li, K. H., Zhang, Y. M, and Su, P., etal. 2008. High-strength steel welding withconsumable double-electrode gas metal arcwelding. Welding Journal 87(3): 57–64. 14. Wu, C. S., Zhang, M. X., and Li, K.H., et al. 2007. Study on the process mech-anism of high-speed arc welding DE-GMAW. Acta Metallurgica Sinica 43(6):663–667. 15. Cheng, X., Shi, Y., and Fan, D., et al.2012. High-speed welding of zinc-coatedsteel sheets with double electrode GMAW.Electric Welding Machine 42(1): 10–13. 16. Shi, Y., Wang, Z., and Dong, B. T., etal. 2012. Test of thermal cycling curve ofpulsed melting-soldering welding DE-GMAW and CMT for bonding steel withaluminum. Journal of Lanzhou University ofTechnology 38(4):10–13. 17. Li, K. H., Chen, J. S., and Zhang, Y.M. 2007. Double-electrode GMAW processand control. Welding Journal 86(8): 231–237. 18. Lu, Y., Chen, J. S., and Zhang, Y. M.,et al. 2014. Predictive control based dou-ble-electrode submerged arc welding for fil-

let joints. Journal of Manufacturing Process-es 16(4): 415–426. 19. Lu, Y., Chen, J. S., and Zhang, Y. M.2015. Dynamic model of consumable dou-ble-electrode submerged arc weldingprocess. Journal of Manufacturing Scienceand Engineering-Transactions, ASME,137(2): 021001-1–021001-6. 20. Li, K. H., and Zhang. Y. M. 2008.Consumable double-electrode GMAW part2: Monitoring, modeling, and control.Welding Journal 87(2): 44–50. 21. Li, K. H., and Zhang. Y. M. 2008.Consumable double-electrode GMAW part1: The process. Welding Journal 87(1):11–17. 22. Lu, Y., Chen, S. J., Yu, S., et al. 2014.Double-electrode arc welding process: Prin-ciple, variants, control and development.Journal of Manufacturing Processes 16(1):93–108. 23. Zhang, Y. M., and Chen, J. S. 2013.Systems and methods to modify gas metalarc welding and its variants, U.S. Patent,No. 8,563,896. 24. Zhang, Y. M., and Chen, J. S. 2012.Systems and methods to modify gas metalarc welding and its variants, U.S. Patent,No. 8,278,587. 25. Yin. S. Y. 2012. Fundamental and Ap-plication of Shield Gas Arc Welding Process.China Machine Press. Beijing.

WELDING RESEARCH

WELDING JOURNAL / FEBRUARY 2016, VOL. 9556-s

Authors: Submit Research Papers Online

Peer review of research papers is now managed through an online system using Editorial Manager soft-ware. Papers can be submitted into the system directly from the Welding Journal page on the AWS website(aws.org) by clicking on “submit papers.” You can also access the new site directly at editorialmanager.com/wj/.Follow the instructions to register or log in. This online system streamlines the review process, and makesit easier to submit papers and track their progress. By publishing in the Welding Journal, more than 70,000members will receive the results of your research.

Additionally, your full paper is posted on the American Welding Society Web site for FREE access aroundthe globe. There are no page charges, and articles are published in full color. By far, the most people, at theleast cost, will recognize your research when you publish in the world-respected Welding Journal.

Chen 2-16_Layout 1 1/14/16 2:23 PM Page 56