Voltage change in the GMAW process due to the influence of a droplet travelling in the arc.

Voltage change in the GMAW process due to the influence of a droplet travelling in the arc Peiyuan Zhu*and S.SimpsonExperimental measurements have been made to investigate the meaning of the fluctuation or noise of electrical signals for the gas metal arc welding process through globular to spray transfer mode,with particular attention being paid to the so-called‘drop spray’transition mode.The results reveal that the welding arc voltage is significantly affected by the molten droplet travelling in the arc.A sharp drop in arc voltage appears just after the detachment of the droplet,reaching its minimum quickly and then recovering,according to the location of the droplet in the arc. Although the full explanation of the feature requires further study,we believe that an important influence is the geometrical effect.The existence of the droplet in the arc path significantly affects the geometrical shape of the arc and arc attachment at the anode,and hence the overall voltage of the arc.Keywords:GMA welding,Droplet transfer,Arc voltageIntroductionGas metal arc welding(GMAW)is a commonly used metal joining industrial technology.Although it is a very well developed industrial process,the complexity of the phenomena that occur in the presence of solid and liquid metal,shielding gas and arc plasma means there are still gaps in our knowledge preventing us from fully under-standing and controlling the process.Over many years of study in thefield,a large number of publications have appeared reporting features and findings of the process,based both on experimental observation and theoretical studies.1–9Among all the features of the GMAW process,metal transfer has attracted most attention around the world.10–14 The formation of the liquid drop and the size of the droplets for various conditions have been both observed experimentally3,5,12,13and simulated in great detail using various theoretical computer models10,11,14,15The interactions between the arc and the formation of a droplet on the wire are understood reasonably well.However,the interaction between the travelling droplet and the surrounding arc has not drawn much research paring the size of a droplet in a typical welding process with the size of the welding arc, it seems that the free travelling droplet will cause serious distortion of the arc,and the degree of the distortion will depend strongly on the size and location of the droplet. Other factors,such as the sheath surrounding the droplet that establishes a potential difference between the arc plasma and the metal droplet,16and both electrical and thermal properties of the liquid metal droplet,will also affect the arc plasma.This paper reports an experimental study to observe the overall impact of a travelling droplet on welding arc characteristics.The possible influence of the arc voltage and currentfluctuations on the formation of the free droplet was observed.This was achieved by correlating precisely the welding arc electrical signals (arc current and voltage)with the shape,size and location of the free droplet images recorded simulta-neously in real time.The authors found that the detachment of the droplet will cause a sharp voltage drop of2–3V.The location of the droplet also seems to have a strong correlation with the arc voltage change during its recovery process.This is particularly clear when GMAW is done in the so-called‘drop spray’17transfer mode,and provides a possible explanation for the regularfluctuation of the arc voltage,which was previously unexplained.It is suggested that a free travelling droplet will affect the arc characteristic significantly because of the geometrical distortion of the arc plasma,and the changes in the arc attachment to the anode,and its influence is responsible for some of the regularfluctuations observed in the arc voltage dynamics.Experimental arrangementThe GMAW process with various welding conditions was performed using a typical industrial welding power source,operating in the constant voltage mode.Copper coated steel electrodes of1.0mm diameter(ER70S-6) andflat steel plates were used as the consumable and workpiece respectively,with contact tip to plate distance of20mm.Argon based shielding gas(y3%O2,y5% CO2,and y92%Ar)was used with aflow rate of 20L min–1throughout this study.School of Electrical&Information Engineering,University of Sydney,NSW2006,Australia*Corresponding author,email Peiyuan@.auß2005IoM Communications LtdPublished by Maney for the Institute of Materials,Minerals and MiningReceived23February2004;accepted24June2004DOI10.1179/174329305X39275Science and Technology of Welding and Joining2005VOL10NO2244Figure1shows a schematic of the experimental arrangement for this study.The arrangement consists of a typical He–Ne laser shadowgraph set-up,a high-speed digital camera,a synchroniser with LCD time display,and a multichannel data acquisition system. The high speed digital camera was capable of recording8000frames s–1,but was operated at a speed of2000frames s–1with camera shutter speed of 561025s.This was fast enough for investigation of the droplet transfer(transfer rates in the order of 102s21)in a typical welding system.The camera was controlled by a personal computer,and the recorded optical image frames were stored in the computer directly.A synchroniser was built and used to synchronise the digital camera with the electrical signal collection system.The synchroniser drives an LCD time display that marks each frame with a time relative to the start of the collection of electrical signals,with a resolution of 1024s.Figure2shows a typical recorded optical image, recorded at a time of0.5245s from the start of electrical signal collection(the digital display has units of1023s). The data acquisition system for measuring welding voltage and current consisted of a multichannel data card,installed in and controlled by a PC computer. The time resolution of the system was set to be1024s,five times higher than the rate of optical image collection.A current probe was used to measure the welding current while the voltage was measured in between the welding plate and the anode nozzle,as shown in Fig.1.With this arrangement,the measured voltage is the sum of three terms,the voltage drop due to the nozzle–anode contact resistance,the voltage drop across the anode wire,and the arc voltage.The arrangement allowed us to observe the dynamics of the consumable wire/droplet(a series of timed optical images)and the welding current and voltage curves at the same time.Results and discussionThe welding current and voltage as a function of time for three different wire feed rates,corresponding to spray(droplet diameter generally smaller than wire diameter),drop spray(droplet diameter slightly larger than wire diameter)and globular(droplet diameter much larger than wire diameter)modes respectively,are compared with each other in Fig.3.The‘drop spray’mode wasfirst observed by Ma18in 1982,and Norris in2003suggested that it shouldbe2A typical sample of a timed laser shadowgraph image, showing a detacheddroplet 1Schematic of the experimental arrangementincluded in a new metal transfer classification list.17Although it appears only at a very narrow band of operation conditions,the clearly regular fluctuations and steady cyclic characteristics make it ideal for understanding the influence of the droplet on the arc plasma.In this research,most of the attention was onthe ‘drop spray’transfer mode.The measurements were made using 1.0mm steelwire,with wire feed rates of 0.1588m s –1(375in min –1)for curve (a),0.1291m s –1(305in min –1)for curve (b)and 0.0931m s –1(220in min –1)for curve (c)in Fig.3.It can be seen from Fig.3that these electrical signals are fluctuating and noisy in general,but in very different manners for different transfer modes.Although it is difficult to interpret the meaning of these signals in full detail,it is clear that these electrical signals reflect the relevant transfer mode’s features and are rich in information.In Fig.3a ,higher frequency in voltage with lower signal level as compared to the other two cases can be seen.This corresponds to a high frequency but smaller sized droplet transfer (spray)mode with a more or less constant arc length.The current curve in Fig.3a shows the lower frequency components,indicating that the inductance of the power supply has limited and smoothed the change in the current,reducing the high frequency component.Figure 3b shows very regular fluctuations in both voltage and current curves,with the frequency being the same as the droplet transfer frequency.This is a key characteristic of the drop spray mode,which has previously been wrongly regarded as being due to power supply modulation.17The fact that these regular fluctuations have the same cyclic period as the droplet transfer rate implies thatthey are related to the droplet transfer processes.The regularity of the fluctuations suggests that the size and the process of each individual droplet transferred in thismode are relatively steady,which was confirmed by therecordings of the droplet images.Figure 3c shows a rather variable high frequency (.50Hz)signal on a low frequency background (3–7Hz).It was found that the droplet size varied a lot and that the transfer mode is actually changing between spray,drop spray and globular modes,with rather significant oscillations in arc length,periodic with the low frequency modules.This is not discussed in detail in this study,since this is a much more complicated situation,as the stick-out instability 7is likely to be causing the low frequency oscillation.The authors believe that the regular voltage fluctua-tions observed in Fig.3b are not due to the change in the stick-out length of the anode wire,for two reasons.First,the stick-out length,and hence the arc length,is relatively steady in this transfer mode,and second,the arc length is at its longest just after droplet detachment and at its shortest just before detachment,whichwould3Typical welding voltage (left)and current (right)as functions of time for a spray,b drop spray and c globular transfermodesinfluence the voltage change in the opposite direction to that observed.Figure 4shows the change in the stick-out length of the anode wire for various stages of a single droplet transfer process in this mode.It can be seen fromFig.4that the stick-out length of the anode wire does not change much in this mode,yielding a relativelysteady arc length.It can also be seen from Fig.4that the stick-out length of the anode increases when droplet detachment is approached and is shortest just after detachment.Characteristic features of drop spray transfer mode As discussed above,the drop spray mode has very regular fluctuations in both voltage and current,with the frequency being the same as the droplet transfer frequency (Fig.3b ).By comparing the electrical signal fluctuations with the laser shadowgraphs for the corresponding times,we find that each cycle of the fluctuation corresponds to one droplet transfer process.The regularity of these fluctua-tions implies that the droplets are all transferred in a very similar manner.The images captured with the high speed camera confirm that the droplet sizes are similar (ranging from 1.26mm to 1.42mm in diameter)at this mode.Therefore,this very stable transfer mode is particularly suitable for investigating the influence of the droplet transfer process on the electrical character-istics of the welding arc.The influence of droplet formation,detachment and travel on the welding arc can be investigated by assessing the droplet images together with the arc voltage signal at the corresponding time.Figure 5shows details of the electrical signals for oneperiod of fluctuation together with nine images,with a point plotted on the voltage curve when an image frame is available at the corresponding time.The nine images were selected after frame by frame investigation of all the images recorded to give a general picture of thedroplet transfer process.The droplet diameter observedfor this case is about 1.3mm.Necking effect and droplet formationIn Fig.5,we can see the liquid necking just before twoindividual droplet detachments,at t 50.5240s andt 50.5370s respectively.Images at t 50.5365s and t 50.5370s show two stagesof the necking process,with the first showing early signof necking,and the latter just before detachment.Fromthese two pictures,we conclude,at lease for this transfermode,that the necking process lasts for about 561023s.The voltage change due to the neckingwould,therefore,naturally occur only in this timeinterval just before detachment.It is hard to measure experimentally the change ofresistance of the liquid on the tip of the wire during thenecking (forming a droplet)process.However,theinfluence of the necking on the voltage change (resis-tance changing during necking)can be estimated bycalculating the change of the resistance of the liquid atnecking using a similar size simulated droplet formationprocess.11We used simulated droplet shapes during the forma-tion 11to estimate the resistance of the droplet at various formation stages by assuming a constant resistivity of 16061026(V cm)for the liquid metal.The resistance of the liquid R can be estimated from R ~ðrS d lwhere r is the resistivity of the liquid metal,l is the liquid length,and S is the cross-sectional area of the liquid.Figure 6shows the simulated droplet shape at various stages of the necking,using the dynamic dropletformation model 11and the estimated resistance of theliquid on the anode tip as a function of the ratio ofliquid volume on the tip to the liquid volume at themoment of detachment.Note that Fig.6provides only an indicative estimation of the resistance for the necking process.It can be seen from Fig.6that the resistance of the liquid does not vary much until the fraction of liquid reaches 0.995.The resistance is of the order of 0.01V ,before it rapidly rises to 0.032V just before detachment.For a 200A current arc,this resistance change implies that a voltage change of the order of half a volt,due to the necking effect,occurs immediately prior to detachment.In Fig.5,we saw a voltage increase of about 0.78V from t 50.5355s,when no obvious neck was observed,to t 50.5372s,when the droplet was detaching.This is in accordance with the neck resistance change estimated from the calculation.However,compared with the random voltage variation level of about 0.5V and a clear voltage drop of 2.5–3V observed after detach-ment,we conclude that the voltage drop after detach-ment cannot be explained as simply a recovery from the necking effect.Characteristic points at droplet detachment time From Fig.5,it is clear that the droplet is detachedbetween 0.5240s (the droplet is still attached)and0.5245s (the droplet has clearly separated with thewire).4Stick-out length for various stage of droplet transfer in drop spray modeSince the detachment of the droplet from the anode wire is a sudden event,it is reasonable to expect a sudden change in the arc at the same time.A turning point of the voltage curve at t 50.5242s was observed,which gives a quite accurate estimation of the detach-ment time.There is no sudden change in welding current (also plotted in Fig.5)at the time of detachment.This can be explained by the fact that the inductance of the power supply not only causes a phase delay but also smooths any sudden changes.The current change is relatively slow,although also periodic,with an averagevalue of about 200A.It is difficult to see a relationshipbetween the droplet transfer process and the currentvariation.The arc voltage has a clear relationship with thedroplet transfer process,through its dependence onparameters such as the arc length and geometry,and thearc–electrode attachment.This characteristic of thevoltage curve provides quite direct information abouthow the droplet transfer affects the welding arc in thedrop spray transfer mode.Through careful study of more laser shadowgraphimages and the voltage curves for each of thesefluctuation periods in this transfer mode,it was foundthat the sudden turning points,which are followed byrather sharp voltage drops,on the voltage curve can betaken as characteristic points corresponding to thedetachment of each droplet.There is often a rise involtage,of the order of about 0.5–1.0V,justbefore5Arc voltage,current and details of droplet transfer for the drop spray modedetachment,which is probably due to the liquid necking effect.The necking voltage change varies,depending on the individual droplet formation and detaching process,and sometimes is not obvious.The sharp decrease in voltage of 2–3V after detach-ment is a characteristic feature of the process.Figure 7shows a section of the voltage curve with the detachment characteristic points marked.Influence of droplet travelling in the arcIt is interesting to see the locations of the droplet at certain points of the voltage curve.In Fig.8,we see that the droplet detaches at point a at t 50.5242s,1.73mm away from the wire tip at point b (t 50.5250s),and about halfway through point c (t 50.5285s),and merges with the pool at t 50.5325s (point d ).At t 50.5365s (point e ),a new neck appears,indicating that detach-ment of the next droplet is about to occur.We observed a 2.6V drop after the detachment of the droplet over about 0.861023to 1.361023s,reaching a minimum at t 50.5250s.After a levelling out,the voltage rises gradually to the nextdetachment.8Droplet at various locations of the arcpathThe droplet in the arc has a significant impact on the arc pressure distribution,due to the geometrical influence,since the droplet diameter we are discussing is about1.3–1.4mm,which is not small compared with the dimension of the arc path,especially near the anode wire.This geometrical influence certainly affects the shape of the arc,and it is understandable that arc voltage would be affected as well.In Fig.9,arc temperature and pressure profiles,for a 200A free burning Ar arc11mm in length,calculated using a two-dimensional free burning model,19are plotted together with a droplet(1.4mm in diameter) on the same scale.The calculation was carried out using similar boundary conditions as those described in Zhu et al.6Thefigure only serves the purpose of geometrical scale comparison of a droplet with a clear path arc(with no droplet in it).The size of the droplet is certainly significant,and the arc will undoubtedly be distorted, and the degree of the distortion will clearly depend on the location of the droplet.From Fig.9,it is concluded that the distortion is most serious when the droplet is near the anode wire tip.Features of spray transfer modesWhile there is a clear correlation between the arc voltage and droplet transfer process in the drop spray transfer mode,the same is not obvious for the spray transfer mode.Figure10shows recorded optical images together with the arc voltage curve(this is a detail of a section of Fig.3a.).For this case,it is not so easy to see the influence of the droplet transfer process at different stages,since the droplet is transferred in lesss time than that needed to travel though the arc path,and there are always several droplets simultaneously at different locations in the arc.Although our image recording rate(2000frames s–1)is only about four times higher than the droplet transfer rate (y500drops s–1),it is clear from our results that the influence of an individual droplet cannot be separated from that of other droplets in the arc path,and the identification of the transfer process stage is more difficult.We conclude that,due to the presence of several droplets at different locations of the arc path simulta-neously,the spray transfer mode has lower amplitude but much less regular voltagefluctuation.In this transfer mode,the arc length and also the arc current do not vary much.Summary and conclusionsExperimental measurements have been made that correlate precisely the electrical dynamics of the GMAW process with the droplet images recorded simultaneously in real time.The results indicate that the electrical signals,in particular thefluctuations in the arc voltage,are rich in information.The droplet formation and transfer process (detachment,travel from the wire tip to welding pool) affect the arc plasma and are responsible,to some degree,for thefluctuations of the voltage curve.One can identify the detachments of the droplets from character-istic maxima of the voltagefluctuations,especially for the transfer modes with droplet size similar or slightly greater than the wire diameter.The results suggest that the welding arc voltage can be significantly affected by a molten droplet travelling in the arc.A rapid drop in arc voltage appears just after the detachment of the droplet.Although a full explanation of the feature requires further study,it is believed that one of the important influences is the geometrical effect of the droplet in the arc.The existence of the droplet in the arc path significantly affects the geometrical shape of the arc,and hence the overall voltage of the arc.This is yet to be confirmed by further study,especially theoretical computer simulations.Reference1.A.Lesnewich:Weld.J.,1958,37,418s–425s.2.J.H.Waszink and L.H.J.Graat:Weld.Res.Suppl.,1983,62,108–116.3.J.C.Needham,C.J.Cooksey and ner:Br.Weld.J.,1960,7,101–114. 10Droplets and voltage curve for stream spray modencaster:‘The physics of welding’,2nd edn;1986,Oxford,Pergamon.5.Y.S.Kim and T.W.Eager:Weld.J.,1993,71,269s–278s.6.P.Zhu,M.Rados and S.W.Simpson:Plasma Sources:Sci.Technol.,1995,4,495–500.7.E.Halmøy:‘Simulation and instabilities in GMAW and SAW’,IIW Document212-895-96;1996,International Institute of Welding.8.P.Zhu,M.Rados and S.W.Simpson:Weld.J.,1997,76,269-S–274-S.9.R.T.Deam,S.W.Simpson and J.Haidar:J.Phys.D.Appl.Phys.,2000,33,1393–1402.10.J.Haidar:J.Appl.Phys.,1998,84,3530–3540.11.S.W.Simpson and P.Zhu:J.Phys.D.Appl.Phys.,1995,28,1594–1600.12.Q.Lin,X.Li and S.W.Simpson:J.Phys.D.Appl.Phys.,2001,34,347–353.13.V.A.Nemchinsky:J.Phys.D.Appl.Phys.,1994,27,1433–1442.14.J.J.Lowke:Austral.Weld.J.Res.Suppl.,1997,42,32–35.15.F.Wang,W.Hou,S.Hu,E.Kannatey-asibu and W.Schultz:J.Phys.D.Appl.Phys.,2003,36,1134–1152.16.M.Mitchner and C.H.Kruger:‘Partially ionized gases’,130;1973,New York,JohnWiley&Sons.17.J.Norrish:‘A review of metal transfer classification in arc welding’,IIW Document XII-1769-03;2003,International Institute of Welding.18.J.Ma:‘Metal transfer in MIG welding’,PhD Thesis,CranfieldInstitute of Technology,1982.19.J.J.Lowke,P.Kovitya and H.P.Schmidt:J.Phys.D.Appl.Phys.,1992,25,1600–1606.。