Arc Instability Phenomena in GMA Welding

Transcription

1 Arc Instability Phenomena in GMA Welding The effect of arc length, power source characteristics, oxidizing and shielding gas on arc stability were investigated potential BY P. J. MODENESI AND J. H. NIXON ABSTRACT. This paper analyzes arc and metal transfer instability phenomena in GMA welding that depend on arc length, the oxidizing potential of the shielding gas and the power supply V/I characteristic. The instability process was associated with abrupt changes in arc current and voltage levels and could seriously affect operation by causing repulsive globular transfer in short circuiting conditions. As instability effects were observed for oxygen levels up to approximately 2%, it can be significant in applications that demand a low oxidizing gas media, as in the welding of alloy steels and nonferrous alloys. The experimental program consisted of two sets of trials using either constant voltage or constant current operation. When constant voltage was used, strong fluctuations in arc length and current level were observed during the transition from unstable (repulsive globular metal transfer) to stable (spray transfer) operation. Significant variations in arc length could not be detected for constant current operation but a clear change in metal transfer and arc appearance could occur if the arc was shortened or elongated. Based on results of the experimental program, a descriptive model based on differences in cathode spot operation was proposed for the instability process. Introduction In the GMAW process, a welding wire is continuously fed through a contact tube where electrical connection is made to the power supply. An electric arc is sustained between the wire and the P. J. MODENESI is with Federal University of Minas Gerais, Belo Horizonte, Brazil. J. H. NIXON is with Cranfield Institute of Technology, Bedford, U.K. workpiece and melts both. To have the process operating stably, at least two basic requirements are to be satisfied: ) the mean wire melting speed has to be equal to the speed at which the wire is being fed; and 2) the molten metal from the wire has to be transferred to the weld pool causing minimal process disturbances. The former requirement can be attained over a large interval of welding current, while the latter is limited to current values or waveforms that promote spray transfer. However, even under these conditions, odd arc instabilities and disordered metal transfer have been observed and ascribed to different factors (Refs. -4). The present study describes an experimental program carried out to study process instability phenomena in GMAW that were linked to shielding gases of low oxidizing potential, short arc length and power supply characteristic. When welding with a constant voltage power supply, this instability was characterized by strong fluctuations in arc length and current. Significant variations KEY WORDS Arc Instability GMAW Arc Length Shielding Gas Power Supply Globular Transfer Alloy Steels Nonferrous Alloys Spray Transfer Descriptive Model in arc length were not detected with a constant current supply but a clear change in metal transfer and arc appearance could be produced by changing arc length. These phenomena were first observed in GMA narrow groove welding trials with a helium-rich (Ar-85%He-.5%CO ppm 0 2 ) shielding gas mixture and, later, in bead-on-plate welding with another mixture that contained 38% He and 2% C0 2 (Ref. 5). Both shielding mixtures were characterized by a low C0 2 /0 2 content and were originally developed for welding stainless steels (Ref. 6). With these mixtures, a stable spray operation was achieved only if an arc longer than about 5 mm (0.2 in.) was used. Experimental Procedure and Results The experimental apparatus consisted of a welding jig with a GMAW transistorized power supply and a data logger system to record welding current and voltage levels. Mild steel plates (thickness = 6.5 mm; 0.25 in.), ER70S-3.2 mm (0.045 in.) welding wire and different shielding gas mixtures (Table ) were used throughout the work. In some trials, a gas mixer was used to add oxygen to the gas mixtures. The experimental program consisted of two sets of trials using either constant voltage (CV) or constant current (CC) operation. Constant voltage trials were used to characterize the time characteristics of the instability phenomenon because, with this type of power supply, a clear transition in process behavior could be observed during welding. However, the current changes in this operation mode provided little base for a more fundamental analysis of the problem. Therefore, CC trials, WELDING RESEARCH SUPPLEMENT I 29-s

2 Table Composition of Base Gas Mixtures Mixture Argonshield 5 Argonshield 20 Argonox Argonox 2 Argonox 5 Helishield Helishield 2 Helishield 0 Nominal Composition Ar-5% C0 2-up to 2% 0 2 Ar-20% C0 2-tip to 2% 0 2 Ar-% 0 2 Ar-2% 0 2 Ar-5% 0 2 Ar-85% He-.5% CO Vpm 0 2 Ar-75% He Ar-38% He-2% C0 2 in which voltage level differences associated with the process stability could be measured, were also performed. In these trials, the stubbing of the wire when it touched the weld pool was reduced by a short-circuit detector that made the power supply switch its output to constant voltage when arc voltage dropped below 5 V. Trials with Constant Voltage Bead-on-plate welds were deposited so that the arc was initiated with a low voltage setting (approx. 27 V) to force a short and unstable arc. After one or two seconds, the voltage was increased to its test value and welding continued until a transition into stable operation took place. This transition was marked by a sudden increase in welding current. The time for transition (t s = t tg, Fig. ) was measured from current and voltage traces. Pure argon, He 0 (Table ) and mixtures of each gas with oxygen were used for shielding. Results are presented in Fig. 2 for a wire feed speed of 50 mm/s (355 in./min). Despite the Table 2 Experimental Conditions Used in the Trials with Constant Current Parameter Shielding Gas Wire Feed Speed (mm/s) (a) Welding Current (A) Contact Tube-to-Work Dist. (mm) (b > (a) mm/s = 2.36 in./min. (b) mm = in. (c) Trials performed inside a hyperbaric chamber filled with argon. large spread of these results, a clear tendency toward longer transition times for both lower oxygen content and test arc voltage can be observed. The transition from unstable to stable operational mode was quite clear with abrupt changes in current level and arc length. Current differences of 40 to 90 A were measured, while high-speed photographs indicated a variation in arc length from 0 to mm (0.039 in.) during the initial unstable operation to 7 to 8 mm ( in.) in stable operation. At the same time, metal transfer changed from repulsive globular to spray. For all shielding gases but pure argon, the process did not return to unstable operation after reaching the stable mode. For pure argon shielding, the process could return to unstable operation with the arc length shortening abruptly. Trials with Constant Current An extensive experimental program was undertaken with CC operation. Five shielding gas compositions were investigated: argon, Ar-25%He, Ar-75%He, helium and Ar-2%0 2. As the main objective of this step was to investigate arc voltage differences associated with both the stable and unstable processes, most Welding Condition Ar, Ar-25%He, Ar-75%He, He and Ar-2% and and 300 4, 7, 20, 23, 26, 29< c» and 32< c > of the experimental trials were performed with pure inert gas shielding, which was expected to favor the unstable process. The Ar-2%0 2 mixture was used to assess the effect of oxygen on voltage levels. Over 60 bead-on-plate welding trials with different contact tube-to-work distances (CTWD), wire feed speeds (w) and current levels were performed (Table 2). Arc length was not measured during these trials. However, as wire melting rate is only weakly affected by shielding gas composition and arc length (Ref. 7), it was assumed that, for a given welding current and melting rate level, the electrode extension was approximately the same for all shielding mixtures and that a change in CTDW corresponded to a similar change in arc length. A few trials were performed inside a hyperbaric chamber, which was evacuated and filled with argon at atmospheric pressure, to check the influence on results of the shielding profile from a commercial GMAW torch. Results indicated that welding stability worsened for shorter arcs (or shorter CTWD). Typical voltage traces from welding trials with different CTWD values are shown in Fig. 3. A significant re- I Oxygen Content» 0% Fig. Typical result of a trial using constant voltage. A Plot of current vs. time; B plot of voltage vs. time. I um current level at unstable mode, l it current level at stable mode, and f s time necessary for a change in process operation D ST io f V M^dMMA ^tifl [«L Qt] ts -* *- 2 * aa st Fig. 2 - Effect of oxygen content and mean test voltage on the time necessary for process stabilization. Wire feed speed: 50 mm/s (355 in./mi n). Shielding gas: A Argon + oxygen; B He-0 + oxygen. f 9" ri2- E F 8-4' 30 * * Voltage (V) r i I x Voltage (V) % 2% Oxygen Content x 0% * % 3% 4% 220-s ISEPTEMBER 994

3 WFI nimr RF^FAKTH?l IPPI FMCMT I TT e suit was the presence of two or three different voltage levels when the process was unstable (Fig. 4): ) short circuit voltage, V 0, 2) intermediate arc voltage level, V s,, and 3) high arc voltage level, V um. In this condition, two very distinct metal transfer modes and arc configurations were identified by photography in a same trial Fig. 5. The spray arc (Fig. 5A) was similar to that observed in CV operation after the process was stabilized (t > t s ), while the repulsive globular transfer (Fig. 5B) was also observed in CV operation, but before the process was stabilized (t < Figure 6 shows histograms of arc voltage from welding trials with different CTWD values. For the shortest CTDW values, histograms clearly present two or even three voltage peaks that correspond to V uns, V st and V. Both V uns and V st tended to increase with CTWD and helium content in the shielding gas Fig. 7. The difference between V uns and V st, however, decreased with helium content and also with welding current Fig. 8. On the other hand, results indicated that oxygen caused an increase in (V uns -V st ) by reducing V s,. Discussion The favorable influence of oxygen on arc stability, metal transfer and bead shape in GMAW welding of steel is well established. In terms of arc stability, oxide films are considered essential to the formation and stabilization of cathode spots that are generally located on the workpiece close to the molten pool (Ref. 8). When pure argon shielding is used, the arc will quickly consume the oxide layer beside the weld pool and will move outward erratically on the surface of the plate resulting in a process that is more difficult to control (Ref. 9). In an oxidizing medium, the oxide layer located close to the weld pool can be continuously regenerated resulting in the arc root being fixed at this position. Experimental results elsewhere indicate that the addition of up to 5% oxygen to argon reduces the spray-globular transition current of steel welding wires. This has been related to different factors such as an increase in arc temperature (Ref. 0), magnetic effects associated with the paramagnetic nature of oxygen (Ref. ), and a reduction in the surface tension of the molten electrode tip (Refs. 2, 3). The presence of an oxidizing component in the shielding mixture smooths the weld bead profile, reducing its wetting angle and reinforcement height, improves bead penetration and reduces the tendency to undercut by decreasing the surface tension of the molten pool (Refs. 40- I J I n ^ " 30- Ol j» o- (D O) 03 -i > O > J Time (ms) ' CTWD = 7mm Time (ms) Vuns Vst Vo 40- f 3 0- > Time CTWD -WvWWVWlWr^^ H Time (ms) Fig. 3 - Voltage traces for different CTWD distances obtained in constant current trials (pure argon, = 260 A and w = 33 mm/s). Fig. 4 Schematic representation of the traces in Fig. 3 (V 0 - short circuit voltage, V st -intermediate voltage level, V U m ~ nigh voltage level). Fig. 5 Metal transfer modes observed in a trial that presented a voltage trace similar to that represented in Fig. 4. A Stable spray transfer; B repulsive transfer. 3, 4) or by stabilizing the position of the arc root (Ref. 5). The present work has shown further instability phenomena in GMAW that are strongly influenced by arc length, oxidizing potential of the shielding gas and power supply characteristic. In CV operation, a transition from an unstable (dominated by repulsive globular metal transfer) to a stable (spray transfer) operation mode was observed. During this transition both the arc length and the welding current (Fig. ) increased considerably. The time required for this transition varied with arc voltage and gas composition, and, for gases with a low oxygen content, the unstable period lasted for several seconds.

4 Fig. 6 - Histograms of the voltage trace from CC trials with different CTWD values (Ar-25%He, I = 260 A and w = 33 mm/s). 0.B J o.oa ,2, Jr "! 4 mml 20 mml This behavior may be associated with instabilities that are commonly observed at the start of GMA welding in steel and often explained by experienced welders by the low initial temperature of the workpiece. As the process usually stabilizes after a few seconds for the shielding gases commonly used with low-carbon steels, the problem has not yet been perceived as significant. However, if a short arc is used and the shielding gas has a low oxidizing potential (2% or less of oxygen or carbon dioxide), this instability could last more than 0 s. In CC operation, the instability was observed only when a short arc was used during which high arc voltage periods (Fig. 3) and repulsive globular transfer (Fig. 5B) were observed. As the periods of high voltage were not present when welding with a long arc, the occurrence of repulsive globular transfer was associated with V uns. This behavior could also lead to an increase in mean arc voltage when arc length (or CTDW) was reduced Fig. 7.! : -<" v / ~ h / U I r >"'; y \ : in c i < a; o.oo , S I 0.5- * mm 23 mm A...IV I. J \ ' Almost no reference to these phenomena was found in the literature, although many references do exist to unexpected instabilities and process behavior changes. Lucas and Amin (Ref. 3) assigned alterations in metal transferto the deoxidation level of GMAW steel wires of the same grade. Exploding drop transfer was associated with high oxygen levels in the wire, while stream spray transfer predominated for low oxygen levels. In a paper studying arc stabilization by rare earth additions, Agusa, ef al. (Ref. 6) mention unstable operation with periodic spray transfer and wire stubbing in the welding of mild steel with pure argon shielding. Scale on the plate surface could prevent wire stubbing. Kiyohara, ef al. (Refs. 7, 8), linked changes in drop size and wire melting rate in the spray GMAW of aluminum with arc length. Rodwell (Ref. 4) studied arc disturbances and variable weld bead shape in the spray GMAW of stainless and mild steels. Sudden and violent. - changes in arc length and current level were associated with contamination on the wire surface and contact tube wear. Middleton (Ref. 9) observed an unexpected increase in voltage at short arc lengths during the development of an arc-voltage-control system for the GMA welding of Inconel with a constant current power supply. Foote (Ref. 20) observed arc instabilities in bead-on-plate GMAW trials with a helium based gas mixture along with certain wire compositions. A paper by Hazlett and Gordon (Ref. 2) presented welding current traces with abrupt changes in current level but made no comment regarding this behavior. The influence of power supply output characteristic on the phenomena described in this paper can be explained by analyzing the interrelationship between power supply and arc characteristics and considering that the arc can operate in two different forms characterized by different voltage levels (V st and V uns ) for the same arc length. Figure 9 shows schematically the transition from unstable (point ) to stable (point 4) operation for a power supply of arbitrary slope (dv/dl) ps. When an arc, operating initially at point (unstable operation), switches to the stable form, the voltage that it requires from the power supply will drop by (V uns -V st ), and it will tend to operate at point 3. However, as this results in an increase of welding current from l uns to l max, the wire melting rate will exceed its feeding rate, and the arc length will tend to increase until equilibrium is restored at some point (4). Therefore, the changes in both arc length and current will depend on the slope of the power supply, and these changes will be more intense when the slope is small. Considering only the voltage drop in the arc, the variation in current due to the change in the form of arc operation can be given by: 'sr 'uns V M-v*W dv /dl t il)? 40- CD cn 30- _ *- ' 20- Vm Vun e *"^ Vst =*"** < Vst - Vuns ' Vmean <> Vst Vuns > Se- 0) CJ c Sem 5 CO a- ""»-.. - Current -*-260 A *-300A CTWD (mm) CTWD (mm) ( Helium Content (%) 00 Fig. 7 Mean voltage, V st and V uns levels for pure argon and Ar-2%0 2 trials with 260 A and wire feed rate = 33 mm/s (34 in./min). Fig. 8 Effect of current level and gas composition on the difference between V um and 222-s I SEPTEMBER 994

5 WFI nimr. PFQFAPPH Cl IPPI CUCMT I ooo,. Where V uns () and V st (4) are the voltage levels for unstable and stable process operations at points and 4, respectively. When the power supply has a CC output characteristic, the transition will be restricted to a change from point to point 2 and, therefore, only minor alterations in arc length should be expected. A tentative model for the mechanism for this arc instability is proposed below to explain some of the observed phenomena. This model is based on results from the present work and from others linked with welding and vacuum arcs. Figure 0 presents a schematic representation of the GMAW arc region. Point C indicates the usual location of cathode spots and r is their displacement to the electrode-arc axis (OA). In GMA welding of steel and most nonrefractory metals, electron emission in the cathode is associated with oxide layers on the metal surface that are removed during the emission (Ref. 22). When a shielding gas relatively rich in oxygen is used, the cathode spots will be located just at the edge of the weld pool and r will be at its minimum value (r 0 ). However, if the oxidizing potential of the shielding gas is not sufficiently high, the cathodic region will have to expand to reach fresh oxide areas and r will be greater than this minimum value. An estimate of r can be obtained by considering that the oxide area that should be consumed by the arc per unit time to sustain the welding current (A,) results both from fresh oxide areas that are brought to the arc region by welding gun translation (A v ) and oxide layers created close to the arc region by any oxygen presented in the shielding (A 0 ): A, = A V + A 0 (2) In a first approximation, A 0 can be considered proportional to the partial pressure of oxygen (p 0 2) in the shielding: and: k r, X r n 2r x v (3) (4) where v is the welding travel speed. Therefore, by rearranging Equations 2, 3 and 4: _A,-k 0 xp 0i 2v (5) Boughton and Amin Mian (Ref. 9) estimate that a um-thick oxide layer would be consumed by the arc at a rate of 0.6 mm 2 /As (0.056 in. 2 /Amin). So, for a 250 A arc, A,=50 mm 2 /s CD CD CO 4-> o > \ \v r-2 2 Uns / -s I Arc \ 4 J^SSJ '2 [Characteristics is^ 3 Current (4 in. 2 /min). If the shielding gas is pure argon (p O2 =0) and the welding speed is 7 mm/s (6.5 in./min), the calculated displacement r of the cathodic region would be equal to approximately 0 mm (0.4 in.). In the presence of a small amount of oxygen in the shielding gas, r would be quickly reduced to its minimum value. The cathodic displacement will result in an increase in the actual arc length of the process Fig. 0: '.(' y r2 +L 2 (6) Which forms an angle 9 with the electrode-arc axis: 9 = tan-\r/l a ) (7) Both / a (r) and 6 will increase if r increases, for instance, by decreasing the oxygen potential in the shielding. G can also be increased by decreasing the arc length l a. When the arc is short, 6 can become so large that it would be very difficult for the arc to preserve its contact with the cathodic region and the resulting configuration would be very unstable. In this condition, the arc may either extinguish or change to a different operating condition. It is proposed that the latter can take place by the development of a new cathode spot that does not depend on the presence of oxide layers to emit electrons and is located on the weld pool. Different mechanisms of electron emission have been proposed in the literature for cold cathode operation, although mainly in relation to vacuum arc processes. Vapor dominated cathodes are reported to develop in vacuum arcs on oxide-free surfaces (Ref. 22) and their presence has also been suggested in welding arcs (Ref. 9). Changes in cathode operation mechanisms in vacuum arcs have also been associated with abrupt variation in voltage (Ref. 23). Power Supply Characteristic Fig. 9- Schematic diagram presenting the relationship between power supply characteristic and the transition from unstable () to stable (4) operation. I: arc length, and ll< 2. This simple model can be used to explain some of the results of the present work. For a given welding condition, r is approximately constant if the process is operating in the stable mode (V = V st ) and is given by Equation 5. Then, if the arc is shortened, G will increase until it reaches a critical value (G c ) above which the maintenance of a contact between the arc and the cathodic region is so difficult that a new cathode spot is formed at the center of the weld pool. This new cathode spot emits electrons by a mechanism that is independent of surface contaminations and, by being less energetically favorable, should need a higher voltage (V = V uns ). As the cathodic region is now concentrated in only one spot, a plasma jet that is directed from the weld pool to the electrode is formed and tends to blow the molten tip of the wire away from the pool (repulsive globular metal transfer). During operation with this single spot, oxide layers on the surface of the workpiece are not consumed and brought closer to the arc region both by the translation of the torch and by surface reoxiv C A L O / = AO i a r = CO l a (r)= AC Fig. 0 Schematic representation of the GMAW arc region.

6 dation. Consequently, the angle G decreases and facilitates again the formation of cathode spots on the plate surface. Then, the arc voltage is reduced to V st and spray operation is restored for a period of time during which the oxide layer is consumed and G increases creating again conditions for the formation of a new single spot in the pool. Therefore, the welding system will alternate periods of multi- and single-spot operation with different voltage levels (V st and V uns) as observed in the CC operation trials. If the shielding gas contains some oxygen, r tends to be smaller than with a completely inert gas shielding (Equation 5). Consequently, a smaller value of / a(r) can be expected in the presence of oxygen and partially explain the lower Vj, values found in welding trials with Ar-2%0 2 when compared with pure argon. Lower values of rand G would also reduce the tendency to the formation of the unstable spot. In CV operation, the arc elongates and the welding current increases during the transition from unstable to stable mode. These changes in arc length and current tend to favor the stable mode and, therefore, make it more difficult for a transition back to the former condition to occur. This agrees with the experimental results that showed that, in CV operation, the transition tended to occur only once. The model presented above is an oversimplification of the complex processes taking place in the arc during the changes in stability observed in this work. Even so, it provides a reasonable explanation for some aspects of the phenomena such as arc length and oxygen dependencies and the influence of the power supply. The model does not consider the influence of time and, therefore, cannot deal with the apparent greater likelihood of unstable operation in the first moments of welding. This time dependency may be linked with workpiece temperature distribution, but no explanation was found for it yet. Conclusions ) When a shielding gas of low oxygen content is used in GMAW of steel, the arc may operate in two distinct modes. These modes are characterized by a difference in arc voltage of about 0V and by distinct metal transfer mechanisms (spray and globular repulsive metal transfer). 2) The voltage difference between these arc modes decreases with an increase in welding current and in the content in helium of the shielding gas. 3) A longer arc and a higher oxygen content in the shielding favor the most stable (spray transfer) arc mode. 4) Power supply output characteristic strongly affects the form in which these arc modes occur. With a constant voltage power supply, repulsive metal transfer tends to occur after the arc is struck and, after a period of time, it can change into a more stable spray transfer. The transition from one mode of metal transfer to the other is marked by abrupt changes in arc length and current level. With a constant current power supply, the two modes of arc operation seem to occur together. 5) The occurrence of the different modes of arc operation may be linked to arc rooting problems and to the operation of distinct mechanism of electron emission from the cathode. References. Hutt, G. A., and Lucas, W Arc disturbances in consumable electrode welding a review of literature. Welding Institute Research Report, 73/ Woods, R. A Metal transfer in aluminum alloys. Welding Journal 59 (2): 59- s to 66-s. 3. Lucas, W., and Amin, M Effect of wire composition in spray transfer mild steel MIG welding. Metal Construction, pp Rodwell, M. H A preliminary investigation into arc disturbances and poor weld appearance in the spray transfer MIG welding of steel. Welding Institute Research Report, 785.0/84/ Modenesi, P. J Statistical Modelling of the Narrow Gap Gas Metal Arc Welding Process, Ph.D. Thesis, Cranfield Institute of Technology. 6. Hilton, D. E., and McKeown, D Improvements in mild steel weld properties by changing the shielding gas theory or practice. Metal Construction, pp Nunes, J. L. (982) Metal Transfer Investigations with a Synergic Power Supply, M.Sc. thesis, Cranfield Institute of Technology. 8. Essers, W. G., van Gospel, M. R Arc control with pulsed GMA welding. Welding Journal, pp Boughton, P., and Amin Mian, M Aspects of the arc root behavior in welding. Second Int. Conf. on Gas Discharges, IEE, pp Brosilow, R Gases for shielded metal arc welding. Weld. Design & Fab., pp Rimskii, S. T, et al, 979. Transfer of electrode metal during welding using shielding gases with oxygen added. Automatic Welding, 979, (0), pp Kennedy, C. R Gas mixtures in welding. Aust. Welding J., pp Salter, G. R., and Dye, S. A. 97. Selecting gas mixtures for MIG welding. Metal Const, and British Weld. J., June 97, pp Norrish, )., and Hilton, D.E Shielding gases for arc welding. Weld. & Metal Fab., May/June 988, pp Cresswell, R. A. (972) Gases and mixtures in MIG and TIG welding. Weld. & Metal Fab., pp Agusa, R., et al. 98. MIG welding with pure argon shielding-arc stabilization by rare earth additions to electrode wires. Metal Construction, pp Kiyohara, M., et al On the stabilization of GMAW welding of aluminum. Welding Journal 56(3): Kiyohara, M., et al Melting characteristics of a wire electrode in the MIG welding of aluminum. Arc Physics and Weld Pool Behavior, The Welding Institute, pp Middleton, P Private communication. 20. Foote, W. ] Welding of C-Mn Steels using the Pulsed Current MIG Welding Process. Ph.D. thesis, Cranfield Institute of Technology. 2. Hazlet, T. B., and Gordon, G. M Studies of welding arcs using various atmospheres and power supplies. Welding Journal 36 (8): 382-s to 386-s. 22. Guile, A. E Electric arcs: their electrode processes and engineering applications. IEE Proceedings, 3K7A), pp Fu, Y. H The influence of cathode microstructure on DC vacuum arcs. /. Phys. D:Appl. Phys., 22, pp e I ^FPTFMRFR Q94