Analysis of Metal Transfer in Gas Metal Arc Welding

This study shows that the transition of metal transfer mode in gas metal arc welding occurs much more gradually than is generally believed

BY Y-S. KIM AND T. W. EAGAR

ABSTRACT. Droplet sizes produced in transfer. These transfer modes show dif­ metal transfer phenomenon. These have GMAW are predicted using both the ferent arc stabilities, weld pool penetra­ had limited success. static force balance theory and the pinch tions, spatter production, porosity pop­ In this study, the droplet size and instability theory as a function of weld­ ulation and level of gas entrapment. droplet transfer frequency are analyzed ing current, and the results are compared Lesnewich (Ref. 1) showed that the mode both theoretically and experimentally. with experimental measurements. The of metal transfer depends on many op­ In the first section of this paper, the equi­ causes for the deviation of predicted erational variables such as welding cur­ librium drop sizes are calculated using droplet size from measured size are dis­ rent, electrode extension, electrode di­ the static force balance analysis and the cussed with suggestions for modification ameter and polarity. Later, A. A. Smith pinch instability analysis. In the second of the theories in order to more accu­ (Ref. 2) reported that an entirely differ­ section of this paper, measurements of rately model metal transfer in GMAW. ent type of metal transfer mode is pro­ droplet size at different welding currents The mechanism of repelled metal trans­ duced when using carbon dioxide gas are compared with the theoretical pre­ fer is also discussed. The transition of shielding as compared with argon shield­ dictions. The limitations of the static metal transfer mode has been considered ing. force balance theory and the pinch in­ as a critical phenomenon which changes With many factors influencing metal stability theory in the prediction of the dramatically over a narrow range of transfer, theoretical models such as the droplet size are discussed. In order to welding current. This transition has been static force balance theory (Refs. 3-5) account for the deviation between these investigated experimentally using high­ and the pinch instability theory (Refs. theories and the experimental data, a speed videography which shows that the 6-8) have been proposed to explain the modification of the static force balance transition is much more gradual than is theory is proposed. The modified theory generally believed. The mechanism of is tested using a pulsed current welding the transition is discussed using a modi­ experiment. fied static force balance theory. KEY WORDS Previous Studies Introduction Modeling Factors Affecting Metal Transfer Modes In gas metal arc welding (GMAW), GMAW Metal Transfer there are various modes of metal trans­ Droplet Size Predict The operational variables affecting fer such as globular, repelled globular, Transfer Frequency the mode of metal transfer are the weld­ Taper Formation ing current, composition of shielding projected spray, streaming, and rotating I Shielding Gas gas, extension of the electrode beyond O Electrode Extension the current contact tube, ambient pres­ CC V. 5. KIM is Assistant Professor, Department Static Force Bal. Theory sure, active element coatings on the elec­ < of Metallurgy and Materials Science, Hong Pinch Instability Theory trode, polarity, and welding material. UJ Ik University, Seoul, Korea. T. W. EAGAR is tn Measurement Among these variables, welding current ui Co-Director, Leaders for Manufacturing Pro­ is the most common variable that the cc gram, Richard P. Simmons Professor of Met­ welder adjusts to obtain the desired allurgy, Department of Materials Science and metal transfer mode. At low welding cur­ Engineering, Massachusetts Institute of Tech­ nology, Cambridge, Mass. rents, globular transfer mode occurs,

WELDING RESEARCH SUPPLEMENT I 269-s while spray transfer mode occurs at rel­ metal transfer modes. where CD is the drag coefficient, Ap is atively higher welding currents. 1) Static Force Balance Theory. The the projected area on the plane perpen­ At the lower welding current range static force balance theory postulates dicular to the fluid flow, pf is the den­ of the spray transfer mode, projected that the drop detaches from the elec­ sity of the fluid, and Vf is the velocity of transfer occurs in which the droplet di­ trode when the static detaching forces the gas. Therefore, the plasma drag force ameter is approximately the same as the on the drop exceed the static retaining acting on the liquid drop can be approx­ diameter of the electrode. As the weld­ force. Four different forces are usually imated by modifying Equation 4 to allow ing current increases, the metal transfer considered: the gravitational force, elec­ for the area occupied by the electrode. mode changes from projected transfer tromagnetic force, and plasma drag The surface tension force, which acts mode to streaming transfer mode, then force are detaching forces, while the sur­ to retain the liquid drop on the electrode to a rotational transfer mode. This sets a face tension force is a retaining force. is given as follows: practical upper limit on the current be­ The gravitational force is due to the mass F^=27tay (5) cause unstable metal transfer begins of the drop and acts as a detaching force with rotational transfer. when welding in the flat position: where a is the radius of the electrode and y is the surface tension of the liquid The transition of metal transfer modes metal. described above are observed in se­ F =-JtR'p H j8 Waszink (Ref. 22) investigated the quence only when the welding material il. relative magnitudes of the detaching is steel and the shielding gas has an where R is the droplet radius, pc| is the forces and showed good agreement with argon-rich composition. With other ma­ density of the drop, and g is the gravita­ experimental results within the range of terials and with other shielding gases, tional constant. globular transfer, however, in the spray not all metal transfer modes are ob­ The electromagnetic force on the transfer mode, the theory deviates sig­ served. When carbon dioxide, helium, drop results from divergence or conver­ nificantly from the experiment. and nitrogen are used as shielding gases, gence of current flow within the elec­ In addition to the above-mentioned the repelled globular transfer mode is trode. When the current lines diverge in usually observed (Refs. 9-11) and nei­ limitations, the static force balance the­ the drop, the Lorentz force, which acts ory has difficulties in explaining several ther streaming transfer nor rotational at right angles to these current lines, cre­ metal transfer phenomena in GMAW. transfer is observed. When mixtures of ates a detaching force. The electromag­ Firstly, the effect of electrode extension argon and carbon dioxide are used, the netic force is given by Lorentz's law: on metal transfer is difficult to explain rate of drop transfer was found to in­ since the electrode extension will not crease linearly with the composition of F„=JxB (2) affect the force balance. Secondly, the the argon gas (Ref. 1 2). analyses of metal transfer using this the­ Considering the effects of shielding where J is current density and B is mag­ netic flux. ory have been performed mostly using gas on the metal transfer modes de­ steel electrodes and argon shielding gas. By assuming that the current density scribed above, there have been several Other systems which produce a re­ on the drop is uniform, the total electro­ attempts to increase the workable range pelling transfer mode cannot be ex­ magnetic force on a drop can be ob­ of welding current by suppressing the plained with this theory. rotational transfer mode using argon- tained by integrating Equation 2 over the 2) Pinch Instability Theory. The pinch based shielding (Ref. 13). When helium current conducting surface of the drop instability theory was developed from and/or carbon dioxide are added to the (Ref. 3). the Rayleigh instability model (Ref. 23) argon gas, the range of welding current of a liquid cylindrical column. Since for projected spray transfer is greatly in­ F_ = fi An spheres can have a lower free energy creased. RsinO than the liquid column, a disturbance of In addition, when a shorter electrode In - COS I the proper wavelength in the liquid col­ extension and a larger electrode diame­ where /, 2 umn tends to cause the liquid column ter are used, the transition current is in­ In- to break up into drops. Rayleigh derived (1-COSf?)' l + COS (3) creased (Ref. 1). Amson (Ref. 14) ob­ the conditions for the liquid column in­ served that as the pressure increases, the stability assuming a simple sinusoidal transition current increases. However, where I is the welding current and \i0 is perturbation for an invicid system. The Perlman, etal. (Ref. 15), found that when the permeability of free space. The ge­ perturbation of the cylinder is solved the pressure reaches 5 atm, spray trans­ ometry used in Equation 3 and a graph with an exponential function of the form fer becomes irregular and fluctuation of off2 as a function of the conduction zone the voltage increases. The use of thin angle is given in Fig. 1. When the con­ a,„(t) = eia) (6) coatings on the electrode consisting of duction zone is small such that the cur­ alkali, alkaline earth, rare earth ele­ rent lines converge, f2 becomes nega­ JB_ ments, and certain oxides have been tive, i.e., the electromagnetic force acts where CO' rf)LM pa3 shown to increase the stability of metal as a repulsive force. However, when the transfer (Refs. 16-18). conduction zone is large enough so that 2/Tar (7) the current lines diverge, f2 becomes X Review of Existing Theories positive and the electromagnetic force of Metal Transfer becomes a detaching force. p = density, X = wavelength of the fluc­ tuation, l (r|) = modified Bessel func­ The plasma drag force on the liquid m tion of the first kind order m, l' (Ti) = the There are two well-quoted theories drop can be estimated by considering m first derivative of l (T|). of metal transfer. These are the static the drag force on a sphere immersed in m force balance theory (Refs. 3-5) and the a fluid. The drag force on a sphere im­ When m = 0 (sausage mode), the pinch instability theory (Refs. 6-8). In mersed in a fluid of uniform velocity maximum frequency is obtained with n, addition to these theories, the plasma field is (Ref. 23) = 0.696, and the most probable drop force theory (Ref. 19) and the critical ve­ size (R0) is calculated to be twice the di­ locity theory (Ref. 20) have been pro­ Pfvj ameter of the liquid column. CnAp posed to explain the transition between (4) The pinch instability theory of metal

270-s I JUNE 1993 transfer analysis postulates that the pinch force on the liquid column of JXiiXXXXKiKli; K)ij )q molten metal due to the self-induced electromagnetic force enhances the break-up of the liquid column into droplets. An approximate analytical so­ lution of the critical wavelength of this instability has been derived (Ref. 24) : „ 2na

1- JhlL 2n2R„ (8)

As seen in Equation 8, the welding 30 60 90 120 I50 current reduces the critical wavelength Fig. 1 — Variation of f as a function of the instability of the liquid jet and thus ANGLE THETA IN DEGREES 2 of q value. decreases the droplet size. In this way, the pinch instability theory claims to ex­ plain the general trend of decreasing drop size with increasing welding cur­ 8 is assumed to be 150 deg, which is the For the surface tension force, it is as­ rent. conduction zone angle when the droplet sumed that the interface between the liq­ Anno (Ref. 25) derived the frequency size is twice the size of the electrode. uid drop and solid electrode is perpen­ of fluctuation for a viscous jet with a sur­ As seen in Fig. 1, the value of f2 does not dicular to the electrode axis. Also, the face charge and showed that viscous ef­ change significantly when 8 is larger diameter of the drop holding neck was fects and surface charges have stabiliz­ than 60 deg. Thus, this assumption will assumed to be the same size as the elec­ ing effects. Allum (Ref. 8) showed that not cause any significant error in the trode diameter. The surface tension data the viscous effects are negligible in liq­ electromagnetic force calculation. were assumed to be: 0.9 N/m for alu­

uid metal, but the stabilizing effects of For the drag force, the value of CD in minum (Ref. 27), 1.3 N/m for titanium surface charge are significant in the low Equation 4 depends on the Reynold's (Ref. 28), and 1.8 N/m for steel (Ref. 27). welding current range. number of the shielding gas. Since the The total detaching force under vari­ The pinch instability theory suffers velocity of the plasma in GMAW is not ous droplet size at a certain welding cur­ the same problems as the static force available, the plasma velocity was as­ rent is the summation of the electromag­ balance theory. These include difficul­ sumed to be 100 m/s, which is the same netic force, the gravitational force, and ties in explaining the effect of electrode as the plasma velocity in GTAW (Ref. the plasma drag force. At the crossover extension, and the repelling mode of 27). The Reynold's number with this ve­ point, where the surface tension hold­ metal transfer. locity is calculated to be approximately ing force and the total detaching forces 3) Other Theories. When using steel 9000, which lies in the Newton's law meet, the equilibrium droplet size is de­ electrodes with argon shielding, the region. Thus, the values of CD for the termined. Figure 2 shows the total de­ transition of metal transfer mode from plasma is 0.44 (Ref. 21). For less-devel­ taching force at various welding currents globular to spray transfer has been re­ oped plasma jets, 10 m/s was used for as a function of droplet size for steel ported to occur over a very narrow cur­ the velocity of the fluid. The value CD electrodes with shielding gas velocities rent range: less than 10 A (Ref. 1). In an for 10 m/s gas flow rate is also calcu­ of 10 m/s, respectively. The crossover attempt to explain the sharp transition lated to be 0.44. Ap is the projected area point at each welding current defines in metal transfer mode as found by of the sphere exposed to the fluid on a the equilibrium droplet size from the Lesnewich (Ref. 1), Needham, etal. (Ref. plane perpendicular to the direction of static force balance theory. The equilib­ the motion and is given by Equation 10 19), using the static force balance the­ rium droplet size of the steel electrode ory, has proposed that the transition oc­ with shielding gas velocity of 1 0 m/s and curs when the welding plasma starts to K(R~ (10) 100 m/s are summarized in Fig. 3. exert a drag force on the drop.

Calculation of Equilibrium Drop Size ^340 •28G> /260 220 /200 > 2 1200 .400 ^360/ /380. From the static force balance theory, /560 440/380/ •320/ 240 .180/ CD the droplet size can be calculated under 520 the assumption that the drop detaches 900 Z/ / from the electrode when the sum of the detaching forces equals the holding 7/7A force : LU 600 o = F... F. + F, cc (holding force) (detaching force) (9) o 300 U- In calculating these forces acting on < the liquid drops, a number of assump­ F- i 0 I I I I I I I I i tions are made. Firstly, in calculating the •o" 0 0.12 0.16 0.20 0.24 0.28 electromagnetic force on the droplet, the electrons are assumed to condense DROP RADIUS (CM) uniformly on the liquid droplet only and Fig. 2 — Variations of the detaching force as the droplet size at the steel electrode tip increases. The assumed argon plasma velocity is 10 m/s.

WELDING RESEARCH SUPPLEMENT I 271-s 0.300 — sJOm/s 2 0.240 c o ^^. o 5 100m/s\ CD 3 0.180 < <*• 0.120 L ^\^__^ CL O ~~ —— Q 0.060 CC 200- O 1,1,1,1,1 160 240 320 400 480 560 200 300 400 500 600 WELDING CURRENT IAI WELDING CURRENT (A)

Fig. 3 — The equilibrium droplet size of the steel electrode calcu­ Fig. 4 — The forces acting on the drop at a steel electrode tip (argon lated from the static balance theory. The shielding gas is assumed to plasma velocity: 100m/s). be argon.

The effect of velocity differences in predicts drop sizes which are much rent electrode positive (DCEP) condi­ the shielding gas is not significant and smaller than the equilibrium drop size tions. An alumina tube was inserted into decreases as the current increases. This predicted by the static force balance the contact tube, leaving only 5 mm (0.2 is because the influence of the electro­ theory. in.) for contact length rather than the magnetic force becomes dominant as normal contact length of 24 mm (0.94 current increases (Fig. 4). At low weld­ Experimental Procedures in.) in commercially available contact ing currents, the welding plasma does tubes. This reduced the variability in not form a strong enough jet to produce Mild steel (AWS ER70S-3), alu­ joule heating due to changes in the lo­ a 100 m/s velocity jet, thus, it is likely minium alloy (AA1100, AA5356), and cation of the current contact. that the droplet size will follow the pre­ titanium alloy (Ti-6AI-4V) were used in Analysis of metal transfer was per­ diction of the 1 0 m/s jet. However, as the experimental portion of this study. formed using high-speed videography the welding current increases, the The shielding gases used were pure with a laser back-lighted shadowgraph ic droplet size will follow the prediction argon, pure helium, 25%Ar-75%He, method (Ref. 30). In this method, a spa­ of the 1 00 m/s jet. In addition, this pre­ 50%Ar-50%He, 75%Ar-25%He, Ar- tial filter is located at the focal point of diction shows that the droplet size de­ 2%02 and carbon dioxide. The weld­ the objective lens, where the parallel creases smoothly as the welding current ing equipment included a constant cur­ laser light becomes focused. Thus, the increases. rent power supply, and a voltage con­ spatial filter transmits most of the laser The droplet size from the pinch in­ trolled electrode feed with a "low iner­ light and excludes most of the intense stability theory is calculated using Equa­ tia" motor. The transistorized welding arc light. The high-speed video camera tion 8. Assuming that the most probable current regulator used in this study was is capable of producing images at a max­ of the constant current type that can sup­ imum 1000 full frame pictures per sec­ wavelength of the instability is Xc/0.696, the droplet size is calculated and plot­ ply DC current with less than 1 % ripple ond (pps). In most of the analyses of ted as a function of welding current in (Ref. 29). This system is capable of puls­ metal transfer, this maximum 1000 pps Fig. 5. The droplet size decreases con­ ing the DC current to a maximum of 5 filming rate was used. The droplet trans­ tinuously as welding current increases; kHz for small superimposed signals. The fer rate was measured for ten seconds however, the pinch instability theory welding was performed under direct cur­ and an average droplet transfer rate for

Globu or >|« Spray ^I0m/s 0.240 s • • ^"""^-^^^ sta ic Force Balance > - 100 m/s^^ ^Th sory 0.2I0- o 0.180 • • 10 ~) • Q ••

Fig. 5 — The droplet size of steel electrodes when shielded with Ar- Fig. 6 — Comparison of predicted and measured droplet size of the 2%02 calculated from the pinch instability theory. steel electrode when shielded with Ar-2%02.

272-s I JUNE 1993 each welding condition was calculated. Table 1 — The Combination of Welding Parameters Used in This Study The droplet size was measured from the still image on the screen once every sec­ Shielding Gas Electrode ond for ten seconds and averaged. The Extension Arc Length Welding variation in measured droplet size and Argon Helium co (mm) (mm) Current frequency was ±5%. 2 The parameters used as variables in Mild steel (a) (a) (a) 16,26,36 14:Ar 180-420 A this study include welding current, weld­ 6:He 8:C0 ing material, shielding gas, electrode ex­ 2 Aluminum (a) 16,26,36 14:Ar 80-220 A tension, and arc length. The combina­ (1100,5356) tions used are shown in Table 1. All the TF6AI-4V (a) (a) 16,26,36 14:Ar, 8 C02 120-260 A welding was performed as bead-on- plate, and the plates were prepared such (a) Welding was performed that there was no mill scale on the sur­ dieted from the pinch instability theory with helium and carbon dioxide shield­ face. The electrode diameter used was is much too small in the globular trans­ ing. Because of errors in direct measure­ 1.6 mm (YH, in.) for all materials. fer mode region and does not tend to­ ments of droplet size due to distortion, The arc length using argon shielding ward the experimentally determined the average droplet sizes that are plot­ gas was constant at 14 mm (0.55 in.), droplet sizes at high currents. ted are not the directly measured data. but with helium and carbon dioxide Figure 7 shows the measured fre­ The average sizes were calculated by di­ shielding gas, the arc lengths were 6 and quency of drop transfer. Again, no sign viding the melting rate of the electrode 8 mm (0.24 and 0.31 in.),respectively, of a sharp transition in the metal trans­ by the frequency of metal transfer, both because of the higher electrical resistiv­ fer rate is present. Using a definition of of which were experimentally mea­ ities of these welding plasmas. spray transfer mode in which the droplet sured. Figure 10 shows the variation of diameter is the same as the diameter of droplet transfer frequency as a function Measurement of Droplet Size the electrode, one obtains a transition of welding current for both of the shield­ and Transfer Frequency at a current of approximately 255 A. The ing gases. transition from projected spray transfer With helium shielding, at low weld­ Effect of Welding Current mode to streaming transfer mode is also ing currents, the droplet sizes are much Figure 6 shows the experimental re­ a gradual phenomenon. The droplet size bigger than predicted by the static force sults of droplet size measured as a func­ decreases gradually, and it is very diffi­ balance theory. As the welding current tion of welding current when shielded cult to define the transition current based increases, a change occurs in the droplet with Ar-2% oxygen gas using a 2.6 cm on either droplet size or frequency. size around 240 A. This jump corre­ (1.02 in.) electrode extension. The sponds to the transition from the repelled droplet size predicted from the static Effect of Shielding Gas globular transfer mode to the projected force balance theory and the pinch in­ spray transfer mode. In the spray trans­ stability theory is also shown in the fig­ Figure 8 shows the repelled transfer fer mode, the droplet size predicted from ure. As seen in Fig. 6, the measured mode just before detachment, when the static force balance theory agrees droplet size decreases gradually as weld­ shielded with carbon dioxide gas. The fairly well with the experimentally mea­ ing current increases, and there is no drops are blown away from the base sured droplet size. With carbon dioxide sharp transition. In the globular transfer metal by the strong cathode jet, and they shielding, a repelled transfer mode was mode, the predicted drop size from the are distorted significantly. As the weld­ observed up to the maximum welding static force balance is reasonably close ing current increases, the droplet size current tested in this study, i.e., 400 A. to the measured values. As the welding and the distortion of the drops become The droplet size continued to decrease current increases, the predicted droplet less pronounced. as the welding current increased, and size starts to deviate from the measured The average droplet size is plotted in there was no dramatic change in the values significantly. The drop size pre- Fig. 9 as a function of welding current droplet size as current was increased.

it o UJ • to 240 • 180 • 120 • 60 •

# ni •— •I • • I •- l I i I i 180 210 240 270 300 WELDING CURRENTIAI

Fig. 7 — Frequency of drop transfer of the steel electrode with Ar- Fig. 8 — Repelled metal transfer of steel electrodes shielded with 2%02 shielding. C02 gas.

WELDING RESEARCH SUPPLEMENT I 273-s 200 250 300 5550 200 250 5500 350 WELDING CURRENT (A) WELDING CURRENT (A)

Fig. 9 — Experimentally measured droplet size for steel electrodes Fig. 10 — Frequency of drop transfer of steel electrodes shielded with shielded with helium and carbon dioxide. helium and carbon dioxide.

With these shielding gases, streaming trode extensions: 2.6 and 3.6 cm (1.02 narrow range of welding current in both transfer mode was not observed. and 1.43 in.). A steel electrode shielded alloys 1100 and 5356. In order to explain the repelled trans­ with Ar-2% 02 gas was used. As the The droplet size in the globular range fer mode observed with these shielding electrode extension increases from 2.6 is within the range of the prediction from gases, it is necessary to introduce a re­ to 3.6 cm, the current of the globular- the static force balance theory. As the pelling force into the static force bal­ projected spray transition deceases from welding current increases, the droplet size ance theory. A possible candidate for 255 to 240 A. As seen in Fig. 11, the ef­ from the static force balance theory devi­ the repelling force in helium and car­ fect of electrode extension on metal ates significantly from the experimental bon dioxide shielding gases is the cath­ transfer is significant. However, this ef­ results as in the case of steel electrodes. ode jet force. The cathode jet force in fect is very difficult to understand in The current of the globular-projected GTAW has been estimated as in Equa­ terms of either the static force balance spray transfer transition was 120 A. tion 11 (Ref. 31). theory or the pinch instability theory. For the Ti-6AI-4V electrode, the mea­ The possible cause for the effect will be sured droplet sizes are shown in Fig. 1 3. discussed in a later section of this paper. cathode jel With this electrode, two different metal Sn (11) transfer modes are shown: repelled Effect of Welding Material transfer due to a strong cathode jet (Fig. In the range of welding current from 14) was observed at low welding cur­ 200 ~ 400 A, the cathode jet force is cal­ Metal transfer with aluminum and rents, and projected transfer was ob­ culated to vary from 1 X 10"2 to 6 X 10'2 with Ti-6AI-4V electrodes in addition to served at high welding currents. As the N, which is comparable to the detaching the steel electrodes was studied. Figure welding current increases, there is a sud­ force. 12 shows the variation of the droplet size den reduction in the droplet size. This with welding current for an aluminum jump corresponds to the transition from Effect of Electrode Extension electrode, along with the droplet size the repelled globular transfer mode to predicted by the static force balance the­ the projected spray transfer mode, as is Figure 11 shows the droplet size vari­ ory. The transition from globular trans­ also found in steel electrodes shielded ation with welding current at two elec­ fer to spray transfer occurs over a very with helium gas.

• • A: Extension 3.3 CM 0.200 • • . 10 m/s • • : Extension 2.6 CM 0.300 2 • 0. I60 5 • _o 0.240 if) 100 m/s"-"— z> s cn 0.120 i - D • n 0.180 < < 0.080 tr Q_ 0.120 • • O Q- • CC **A o • Q 0.040 rr • - o 0.060 • • • n I i 1 , 1 I n 1 ' 1 1 150 200 250 300 120 150 180 210 240 WELDING CURRENT |A| WELDING CURRENTIAI

Fig. 11 — Effects of electrode extensions on the droplet size of steel Fig. 12 — Comparison of predicted and measured droplet size of an electrodes. The shield gas is Ar-2%02. aluminum 1100 electrode with Ar-2%02 shielding.

274-s I JUNE 1993 In the projected spray transfer region, 0.40 the droplet size predicted by the static force balance theory predicts larger droplet sizes compared with the experi­ 3 0.30 mental results. It is believed that the de­ w viation is caused by the value of the sur­ face tension used: the surface tension < 0.20- for the Ti-6Al-4V alloy was unavailable cr and the surface tension for pure titanium CL • • was used in the calculation. o g O.IO As an illustration of the strength of Fig. 13 — Comparison the cathode jet force with Ti-6Al-4V of predicted and mea­ sured droplet size of a electrodes, Fig. 1 5 shows successive pic­ _L tures of a trajectory of the drop detached TI-6AI-4Velectrode 140 180 220 260 with from the electrode and repelled by the ing. cathode jet in the welding plasma. The WELDING CURRENTIA] welding material used was zirconium, which has similar thermophysical prop­ erties as the Ti-6AI-4V electrode, and the shielding gas was argon. Initially, the drop travels toward the weld pool. As it approaches the weld pool, the cathode jet becomes stronger such that the velocity of the drop is reduced and finally the drop is rejected away from the weld pool by the cathode jet, causing spatter on the base plate. This phe­ nomenon is also frequently observed with the Ti-6Al-4V electrode. As the welding current increases, the anode jet dominates the plasma flow, and the droplet transfer mode changes from repelled transfer to projected spray transfer.

The Cause of Droplet Size Deviation between the Static Force Balance Theory and the Experimental Results

The discrepancy of the droplet size from the static force balance theory as compared with the experimental data may be explained by examining the va­ lidity of the assumptions made in the calculation of the static force balance theory. One of the most important as­ sumptions is that the electrode should Fig. 14 — Repelled metal transfer of TI-6A1-4V electrodes caused by the strong cathode jet force. The shielding gas is Ar-2%0 shielding. remain cylindrical and maintain its full 2 diameter at the point where the drop is formed. If the diameter of the neck is changed either by surface melting or by Table 2 — Tapering of Electrode at Various Welding Currents with a Steel Electrode Shielded deformation, the holding force will be with Ar-2% 02 affected as can be seen from Equation 5. Welding current (A) 205 237 253 281 310 The geometry of the drop holding Taper length (mm) 0 2.4 2.7 4.1 5.1 neck was investigated under various conditions, and it was found to be sig­ nificantly changed due to formation of a taper at the electrode tip — Table 2. As seen in Table 2, the tapering of the electrode occurs in the range of welding currents where the discrepancy between the theoretical droplet size and experimental data becomes significant. Figure 1 6 shows a fully developed taper at high welding current (>280 A). The tapering of the electrode occurs because the anode spot reaches this surface of the electrode and generates condensa­ Fig. 15 — Successive pictures of a drop rejected by the strong cathode jet from the base metal. tion heating on the cylindrical surface The electrode material is zirconium and the shielding gas is pure argon.

WELDING RESEARCH SUPPLEMENT I 275-s 0.240

CL • : Experimental Data CC 0.060 Q A : Modified Static force Balance Theory X 180 210 240 270 300 WELDING CURRENT (A)

Fig. lb — A typical shape of the taper formed at the end of steel elec­ Fig. 17 — Comparison of droplet size predicted by the static force trode when shielded with Ar-2%07. balance theory and by the modified static force balance theory with measured values.

of the electrode (Ref. 32). When enough by measuring the size of the drop hold­ tional area of the electrode. The con­ heat is generated on the surface, the sur­ ing neck at different welding currents densation heat is generated when elec­ face will melt and the liquid metal will and substituting the tapered electrode trons enter the electrode from the be swept downward by either the gravi­ radius for the radius of the drop holding plasma. When shielded with argon, tational force and/or the plasma drag neck in Equation 5. It was assumed that some portion of this heat enters onto the force. When this melting and sweeping most of the welding current flows di­ cylindrical surface since electrons con­ action occurs over a significant length rectly into the drop. As seen in the fig­ dense not only on the surface of the liq­ of cylinder, a taper will develop at the ure, the droplet size predicted from the uid drop, but also on the surface of the end of the electrode. modified theory follows the measured solid electrode. The joule heat is the heat The tapering of the electrode tip will droplet size. generated by the electrical resistance of reduce the effective diameter of the drop The effect of the electrode extension the electrode and occurs uniformly in­ holding neck, thus reducing the holding may also be explained by considering side the electrode. force in Equation 5. The reduced hold­ the taper formation. The total heat input When there is no condensation heat ing force will produce a smaller droplet into the electrode occurs via electron input on the cylindrical surface of the size thus creating a streaming droplet condensation heat and joule heat as de­ electrode, the temperature of the solid transfer mode. On the other hand, with fined by Equation 1 2 electrode will increase as the joule heat steel electrodes shielded with helium generation increases. Thus, with longer and with Ti-6AI-4V electrodes shielded 2 electrode extension, the melting of the Q l.=\-kT/e+V„ a +

In order to test the tapering theory, pulsed current welding experiments 0.300 with which tapering can be controlled Fig. 18 — Comparison Base Duty between the dropiet Current Cycle (%) were designed. The experimental pro­ size of the steel elec­ cedures for the pulsed current welding 0.240 o 200 5 trode form the static ^^\ may be found elsewhere (Ref. 33). Fig­ A I 80 5 force balance theory ^% ure 18 shows the minimum droplet sizes cn — Theoretica and minimum droplet O.I 80 — measured at 300-, 400- and 500-A peak size from pulsed cur­ 3 O - A currents, along with the droplet size pre­ rent welding. The < shielding gas is CC O.I 20 — dicted from the static force balance the­ CL ory. Steel electrodes shielded with Ar - Ar-2%02 Shielding. O CC 2% oxygen were used at base currents Q 0.060 of 1 80 and 200 A. The load duty cycle was 5%. The pulsing parameters were I I , I , I , I selected such that there is no taper for­ I60 240 320 400 480 560 mation on the electrode. The droplet size PEAK CURRENTIAl was measured at the pulsing frequency,

276-s I JUNE 1993 which produces the minimum droplet sizes at a given peak current. As seen in the figure, the droplet size predicted from the static force balance theory agrees within ±10% with the experi­ mental data. This experiment suggests that the cause of the deviation of droplet size in DC welding is tapering of the electrode tip. Fig. 19 — Successive pictures of pendent drop motion prior to detachment from the steel elec­ Comparison between the Static trode (total elapsed time: 60 ms). The shielding gas is Ar-2%02 shielding. Force Balance Theory and the Pinch Instability Theory tographic analyses show that the peak References The pinch instability theory and the velocity of the pendent drop reaches 20 1. Lesnewich, A. 1958. Control of melting static force balance theory have been cm/s and the dynamic force due to the rate and metal transfer in gas-shielded metal used in modeling metal transfer with pendent drop movement is calculated arc welding: part - control of metal transfer. somewhat disappointing results. The to be approximately 2 X 1 0"2 N. This is Welding Journal 37(9): 418-s to 425-s. drop size predicted from the pinch in­ enough to account for the deviation of 2. Smith, A. A. 1970. CO, welding of stability theory as in Equation 8 is un­ the drop size in the globular mode. Also, steel, The Welding Institute, Cambridge, U.K. able to predict the trends of the mea­ the surface tension value used in this 3. Amson, J. C. 1 962. An analysis of the sured drop size as seen in Fig. 6. One of study to calculate the holding force may gas-shielded consumable metal arc welding the fundamental requirements for the system. British Welding Journal 41(4): be larger than the actual value of the sys­ 232-249. pinch instability phenomena to occur is tem and, hence, may be incorrect. Com­ 4. Greene, W. J. 1960. An analysis of that the liquid metal should be in the bining these two effects, the predicted transfer in gas-shielded welding arcs. Trans. form of cylinder, which is at a higher droplet size could be made to agree AIEE Pan 2, 7: 194-203. state of free energy than a correspond­ more closely with the experimentally 5. Waszink,). H., and Graat, L. H. J. 1 983. ing liquid metal sphere. However, ob­ measured data. Experimental investigation of the forces act­ servation with a high-speed video cam­ ing on a drop of weld metal. Welding Jour­ era shows that as soon as the solid metal nal 62(4):109-s to 11 6-s. Conclusions melts, it forms a spherical liquid drop, 6. Lancaster, J. F. IIW Document No. 21 2- which is already in the lower free en­ 429-78. 1) Metal transfer with steel electrodes ergy state as compared with a cylindri­ 7. Allum, C.). 1985. Metal transfer in arc shielded with Ar-2% oxygen shows a cal liquid column. Thus, there is no driv­ welding as a varicose instability: part 1 — vari­ gradual transition, from globular to pro­ ing force, and it is not logical to apply cose instability in a current-carrying liquid jected spray, followed by streaming cylinder with surface charge, journal of the pinch instability theory to a problem transfer mode. Physics D: Applied Physics 1 8: 1431 -1446. in which a cylindrical liquid column 8. Allum, C. J. 1985. Metal transfer in arc never exists. Also, the repelled globular 2) The static force balance theory can predict the droplet size in the globular welding as a varicose instability: part 2—de­ transfer mode and the effect of the elec­ velopment of model for arc welding. Journal transfer range, but it deviates signifi­ trode extension is very difficult to ex­ of Physics D: Applied Physics 18: plain by the pinch instability theory. cantly in the spray transfer range. The 1447-1468. From these observations, it is concluded cause of the deviation is the geometry 9. Defize, L. F. 1962. Metal transfer in that the pinch instability theory is an in­ change of the electrode due to a taper gas-shielded arcs. Physics of the Welding Arc. appropriate way to explain metal trans­ formation at the electrode tip. The Institute of Welding, London, U.K. fer phenomena in either globular trans­ 3) In order to analyze repelled glob­ 112-114. fer or projected spray transfer. ular transfer by the static force balance 10. Hezlett, T. B., and Gordon, G. M. theory, it is necessary to determine an­ 1 957. Studies of welding arcs using various However, in streaming metal trans­ other force that will act as a repelling atmospheres and power supplies. Welding fer mode, a liquid column instability force. A possible candidate for the re­ Journal 36(8): 382-s to 386-s. phenomenon is observed as seen in Fig. pelling force is the cathode jet force on 1 1. Jahn, R. E. and Gourd, L. M. 1 957. Gas-shielded welding of steel. British Com­ 18. In this case, a cylindrical liquid jet the drop. monwealth Welding Conference. The Insti­ is formed at the end of the electrode, 4) When the taper is not formed, the tute of Welding, London, U.K., pp.148-1 55. and it disintegrates into several drops static force balance theory can predict 12. Elisstratov, A. P., Chernyshev, G. G., away from the electrode tip. As long as the droplet size very accurately in the and Spitsyn, V. V. 1 975. Transfer of electrode the diameter of the liquid jet remains the spray transfer range, as has been shown metal during welding in shielding mixtures. same, the droplet size will remain the in pulsed current welding. Avt. Svarka 12:61-62. same. This may explain the plateau of 5) The pinch instability theory fails 13. Church, J. G. U.S. Patent #4463243, drop size in the high current range of to explain the effect of the electrode ex­ July 31, 1984. 14. Amson, ). C, and Salter, G. R. 1963. streaming transfer as seen in Fig. 6. In tension or of changes in the shielding Analysis of the gas-shielded consumable this case, where a cylindrical liquid col­ gas on the metal transfer mode. The umn exists the pinch instability theory metal-arc welding system-effect of ambient static force balance theory as modified gas pressure. British Welding Journal 9: might be applicable. by the taper formation can adequately 472-483. The modified static force balance the­ explain metal transfer. 15. Perlman, M., Pense, A. W., and Stout, ory predicts a larger drop size than the R. D. 1969. Ambient pressure effects on gas experimentally measured value in gen­ Acknowledgment metal-arc welding of mild steel. Welding eral. One of the possible causes of this Journal (48)6: 231-s to 238-s. deviation is the drop movement on the This research was funded by a grant 16. Cameron, J. M ., and Baeslack, A. J. electrode as seen in Fig. 19. The figure from the United States Department of 1956. A straight-polarity, inert-gas process of welding mild steel. Welding Journal 35(5): shows successive pictures of pendant Energy under contract number DE- 445^149. drop motion over a 60 ms period. Pho- FG02-85ER-13331.

WELDING RESEARCH SUPPLEMENT I 277-s 1 7. Agusa, K., Nishiama, N., and Tsuboi, demic Press, New York. 28. Hand Book of Chemistry and Physics, j. 1981. MIG welding with pure argon shield- 22. Waszink, J. H., and Graat, L. H. J. 63rd ed. 1982. F-28, CRC Press, Boca Raton, ing-arc stabilization by rare earth additions 1979. Der einflues der gasstroemung auf die Fla. to electrode wires. Metal Construction 9: tropfenabloesung beim plasma-mig scheis- 29. Eickhoff, S. T. 1988. Gas metal arc 570-574. sen. Grosse Schweisstechnische Tagung, DVS welding in pure argon. M.S. thesis, M.I.T., 18. Lucas, W., and Amin, M. 1975. Effect Berichte57: 198-203. Cambridge, Mass. of wire composition in spray transfer mild 23. Lord Rayleigh, 1897. Proceedings of 30. Allemand, C. D., Schoeder, R„ Ries, steel MIG welding. Metal Construction 2: the Mathematical Society, pp. 10, 4. D. E., and Eagar, T. W. 1985. A method of 77-83. 24. Lancaster, J. F. IIW Document No. filming metal transfer in welding arc. Weld­ 19. Needham, J.C., Cooksey, C. J., and 212-429-78. ing Journal 64(1): 45-47. Milner, D. R. 1960. Metal transfer in inert- 25. Anno, |. N. 1977. The Mechanics of 31. Lin, M. L. 1985. Transport processes gas shielded-arc welding. British Welding Liquid Jets. Lexington Books, Lexington, affecting the shape of arc welds. Ph.D thesis, Journal 7(2): 101-114. Mass. M.I.T., Cambridge, Mass. 20. Waszink, J. H., and Van Den Heuvel, 26. Chapman, B. 1980. Glow Discharge 32. Kim, Y. S. 1989. Metal transfer in gas G. ). P. M. 1 982. Heat generation and heat Processes. John Wiley and Sons, New York, metal arc welding. Ph. D thesis, M.I.T., Cam­ flow in the filler metal in GMA welding. Weld­ 53. bridge, Mass. ing Journal 6'\ : 269-s to 282-s. 27. Mondain-Monval. The physical prop­ 33. Kim, Y. S., and Eagar, T. W. Metal 21. Szekley, J. 1979. Fluid Flow Phenom­ erties of fluids at elevated temperature. IIW transfer in pulsed current gas metal arc weld­ ena in Metals Processing, pp. 256-257. Aca­ Document No. 212-264-73. ing. To be published in Welding Journal.

The Welding Journal is printed with environmentally ty • *» V<% ^ friendly soy-based ink — another way we're helping to reduce the emission of pollution-causing chemicals into the atmosphere.

WELDING JOURNAL

278-s I JUNE 1993