Heat Transfer and Penetration Mechanisms with GMA and Plasma-GMA Welding

Transcription

1 Heat Transfer and Penetration Mechanisms with GMA and Plasma-GMA Welding The heat content of transferring metal drops appears to determine the total cross-sectional area of weld penetration, while the impact of the drops on the liquid metal weld pool determines the depth of penetration BY W. G. ESSERS AND R. WALTER ABSTRACT. Heat input to the workpiece during GMA and plasma-gma welding can be measured with a simple calorimetric apparatus. A distinction can be made between the heat supplied by the arc via convection, radiation and conduction, the heat generated in the cathodic region and the heat contained in the transferring metal drops. For weld penetration in the workpiece, the heat content of the transferring metal drops appears to determine in large measure the total cross-sectional area of the penetration, while the impact of the drops on the liquid weld pool determines the depth of penetration. An electromagnetic method is described, by means of which every drop produced when welding with filler metal wire can be made to move in a predefined direction. As a result, the drops strike the weld pool at the points desired, i.e., the points at which penetration is required. Introduction In gas-metal-arc (GMA) welding, the heat which is transferred to the workpiece (cathode) is determined by a number of processes: 1. The phenomena which appear in the cathode region, such as thermionic emission and the interaction of positive ions with the cathode surface. 2. The energy transferred from the arc column by convection, radiation and conduction. 3. The heat developed in the filler metal. This heat is transferred to the workpiece via the molten drops. The individual part played by each of the various heat transfer mechanisms has been partly reported in a recent publication. 1 The present paper extends the earlier work with a number of new observations. In addition, a picture which has been obtained of the penetration mechanism with GMA welding is described. Finally, it is shown how this knowledge can be used to produce weld penetration of a required shape. Heat Transfer During Welding In order to measure the amount of heat transferred to the workpiece by the arc and the transferring drops, a simple water-filled calorimeter was used (Fig. 1). A metal strip which was used for a bead-on-plate test was placed in the calorimeter in such a way that it was almost completely immersed in the water. Only the upper surface was just above water level. The traverse speed was chosen such that no gas bubbles appeared in the water. A rotating blade ensured that the temperature distribution in the water was virtually constant. Water temperature variations were continuously recorded. The maximum temperature of the water during all tests was below 35 C (95 F). Although Paper presented at the 61st AWS Annual Meeting held in Los Angeles, California, during April 13-18, W. G. ESSERS is Senior Research Scientist and R. WALTER is Research Assistant, Philips Research Laboratories, Eindhoven, The Netherlands. there was some heat loss from the surface of the metal strip, the resulting error appeared to be relatively small, i.e., smaller than 5% of the heat input to the workpiece. The plasma-gma process 2 can be used either with of without filler metal. Therefore, it is possible with this process to differentiate between heat due to the arc and heat due to the transferring weld metal. It was, for this reason, used in addition to the usual GMA process. Figure 1 illustrates the two processes GMA and plasma-gma. Figure 1A shows GMA welding with cold shielding gas; Fig. 1B represents plasma-gma welding, using an experimental welding torch. In the latter case the filler metal is surrounded by thermally-ionized gas, a plasma created by a second arc between a nonconsumable electrode and the workpiece. The plasma has a temperature of more than 10,000 K :i and is, therefore, an excellent conductor in which filler metal melt-off takes place under control. As both views of Fig. 1 show, the workpiece consisting of 10 mm (0.39 in.) thick mild steel plate is placed in a water-filled calorimeter. The dotted line in Fig. 1 B illustrates a possibility by which welding with a non-transferred plasma can be performed. In this case switch S2 is closed and switch SI is opened. These arrangements allow the heat input to the workpiece to be measured in varying conditions. The filler metal consisted of mild steel with a diameter of 1.2 mm (0.05 in.), the shielding gas was argon + 7% CO, and the plasma orifice was 10 mm WELDING RESEARCH SUPPLEMENT I 37-s

2 ^nsa 6 D Q. C o 0) 10 8 i r a) o plasma without wire addition and JTO current through the workpiece, _b)» plasma without wire addition c)a plasma-gma welding d) x GMA-welding I total current in (A) Fig. 2 Measured heat input to the workpiece as a function oi the total current supplied to the system Fig. 1 Schematic representations oi GMA welding (A) and plasma-gma welding (B) above a calorimeter: 1 shielding gas; 2 filler metal; 3 workpiece; 4 water-filled calorimeter; 5 plasma-arc (0.39 in.) in diameter. The following tests were carried out: 1. Using the arrangement of Fig. IB, the heat input to the workpiece was measured with an arc of which the current did not flow through the workpiece but was collected by the lower nozzle. Switch S1 was opened and S2 was closed.. No filler metal wire was added. In other respects this non-transferred arc had the same shape as a transferred arc but only touched the workpiece. In this case heat transfer takes place by convection, radiation and conduction. The workpiece was not subjected to cathodic heat. The efficiency of this heat transport mechanism was 23%, defined by the expression: Efficiency Heat input to workpiece Total power input to process -X 100% The value of the efficiency appears to be more or less constant over the whole range of currents used (±3%). 2. When the current through the arc also passed through the workpiece, so that this formed the cathode, significantly more heat was absorbed by the workpiece. (Once again, no filler was used.) The efficiency in this case was found to be 54% and this remained constant over the whole range of currents (±3%). 3. In the next tests a filler metal wire was fed into the plasma, the wire being connected to a second power source as is usual with plasma-gma welding. At higher currents even more heat per ampere supplied to the process was transferred to the workpiece. The efficiency was found to be 65% over the whole range of currents (±3%). 4. Finally, normal GMA welding was used (Fig. IA). This gave the highest heat input to the workpiece per ampere supplied to the process. This also determines the efficiency, which was found to be 71% over the whole range of currents used (±3%). Figure 2 shows plots of the measured heat input to the workpiece against the total current supplied to the system for test nos. 1 to 4 as described above. There is a notable difference between the heat absorbed by the workpiece in GMA welding (curve d) and in plasma-gma welding (curve c). In the case of GMA welding more heat is taken up by the workpiece per ampere supplied. This can be explained by the fact that in both cases there is only one cathode namely, the workpiece. With GMA welding there is also only one anode the filler metal. With plasma-gma, however, there are two anodes the filler metal and the non-consumable plasma anode. In the latter case, part of the anode heat is removed by the cooling water of the non-consumable electrode. If one compares curves a, b and d in Fig. 2, one can draw a conclusion as regards GMA welding. Although the ratios between the amounts of power transferred to the workpiece are not exactly equal over the whole range of currents, the amount of heat transferred to the workpiece by radiation, convection and conduction amounts to about 34% (±3%) of the total heat input in the case of GMA welding. The passage of current in the cathode area delivers about 41% (±3%). Finally, the metal drops account for about 25% (±5%) of the total heat transfer to the workpiece. The part played by the drops is further discussed later. The data quoted above apply naturally only to the conditions described. Penetration of the Weld in the Workpiece Heat from the Arc From the tests described above, it appears that of all the heat transferred to the workpiece during welding, the heat from the arc accounts for the largest part, about 75%. However, other tests have shown that this power does not contribute significantly to weld penetration in the workpiece. Experiments using a torch as shown in Fig. 1B and no added filler metal, with only the plasma arc transferred to the workpiece, have shown that radiation, convection and conduction from the arc together with current passage at the cathode can have only a very limited influence on the depth penetration. A 10 mm (0.39 in.) diameter transferred plasma arc was used with an argon gas flow of 8 liters/min (17 cfh)* plasma gas and 18 liters argon + 2 *Liters/min = cfh. 38-s I FEBRUARY 1981

3 Table 1-Influence of Heat Content of Molten Weld Metal Droplets on Weld Penetration 1 * 1 lw"" Depth of penetration, mm M, g/cm lc '> H M, J/cm,di H,. J/g"' H J/cm,n current in filler wire (A) Fig. 3 The ratio H l/h yl as a function ol the current through the 1.2 mm diameter filler metal (mild steel) 18 '150 A plasma current. 8 liter argon/min plasma gas. 18 liler argon + 2 liter CO,/min shielding gas, diameter plasma orifice 10 mm, wire extension 33 mm, 3.78 g filler metal per cm weld length, 1.2 mm diameter mild steel filler metal composition: % C; % Mn; % Si; 0.01% Al max. ""[ is the current through the filler metal. lrl Total amount of molten metal per cm weld length. ""Heat necessary to melt M from room temperature at 1536 C (2795 F). le 'Heat content of the droplets. <rl Heat content of the filler metal per cm weld length. liters C0 2/min shielding gas. Three runs were made over a 10 mm (0.39 in.) thick mild steel plate with 180, 240 and 300 A plasma current. The traverse speed was 0.24 m/min (9 ipm). The penetration depths were 0.1, 0.2 and 0.3 mm (0.0039, , and in.), respectively. These tests illustrate the small effect on the penetration depth of the combination of radiative, convective and conductive transfer and current passage. Obviously these small penetrations were obtained only with the fairly low arc-current densities mentioned. Higher current densities (smaller diameter of the plasma orifice) would produce increased penetration. However, current densities of this order of magnitude will continue to be discussed throughout this paper. Other experiments have also shown that with GMA and plasma-gma welding the arc heat influences the depth ot penetration only to a limited degree.' The heat supplied from this source has a great influence on the width of the weld and on the contact angle between the weld bead and the surface of the workpiece. The greater the amount of this heat, the wider the weld bead and the better the wettingin of the weld metal to the work- piece.- couple. The temperatures showed a variation of about 70 C (126 F) for welding currents of 125 to 235 A through the 1.2 mm (0.047 in.) diameter mild steel filler metal, using the plasma-gma process. The values obtained were 2100 and 2170 C (3812 and 393~8 F), respectively. In order to determine the influence of the heat content of the drops on weld penetration in the workpiece a number of tests were carried out using the plasma-gma process. The plasma current was kept constant at 150 A. The plasma gas flow was 8 liters argon/min with 18 liters argon + 2 liters C0 2/min shielding gas. The diameter of the plasma orifice was 10 mm (0.39 in.). The tests were of the bead-on-plate type using a 1.2 mm (0.047 in.) diameter mild steel filler metal and a 10 mm (0.39 in.) thick mild steel plate. The traverse speed was chosen such that the amount of weld metal per cm weld length was constant. The depth of penetration and the cross-sectional area of the weld were measured in each of the tests. From these values the total amount of molten metal per cm weld length (M) was calculated. The heat content of the drops from the filler metal was derived from the following equation: 6 H dr = 0.81 T dr + 92 (1) where H dr = heat content of the drops )/g: T dr = temperature of the drops, C, as obtained by Jelmorini et al. The tests were carried out for a wide range of currents through the filler metal (l ). The results are given in Table 1 When H, is the heat content of the filler metal per cm weld length and Hj, is the heat necessary to heat and melt the mass M from room temperature to 1536 C (2797 F), then the ratio H,/H can be determined. This ratio is plotted against the current through the filler metal in Fig. 3. It can be seen that H,/H M is close to 1 over the whole range of currents used, varying in fact from 0.89 at 322 A to 1.11 at 100 A. With currents lower than 250 A, all the energy required to melt the weld area metal is contained in the overheated drops. Above 250 A the heat content of the drop is slightly too small. Heat of the Filler Metal It is known from the literature that the metal drops transferring from the filler metal to the workpiece are strongly overheated. It can be reasonably assumed that this extra heat contributes to the melting of the workpiece. The values of drop temperature found by Ando et a/.' and Jelmorini et a/ 3 are in good agreement with each other. Ando measured the heat content of the drops by means of a calorimeter and from this calculated the temperature. Jelmorini measured the temperature of the falling drops directly by catching them on a thermo tn ll (i *o >a CD E current in filler wire (A) Fig. 4--The drop mass (A), the drop frequency (B) and the drop velocity (C) as a function of the current through the filler metal (mild steel). The polarity and diameter of the filler metal are indicated, (v in illustration has same meaning as f) Kinetic Energy In addition to heat, another form of energy can be transferred from the drops to the workpiece: their kinetic energy. For each centimeter of weld length, this is given by: E f mv" 2s (2) In which f is the drop frequency, m the average mass of the drops in kg, v the average velocity of the drops on hitting the weld pool in m/sec, and s is the traverse speed in cm/sec. To obtain an idea of the value of this form WELDING RESEARCH SUPPLEMENT I 39-s

4 of energy, the number, the velocity and the mass of the transferring metal drops must be determined. In the present case they were measured during plasma-gma welding on mild steel. For this purpose the metal transfer was filmed at a rate of 7000 frames per second, and the film was examined on a motion analyzer. From the results the following data could be calculated: the drop mass (m), the number of drops reaching the weld pool per second (f) and the velocity (v) of the drops immediately before impact. The main variables were then the wire diameter and the wire polarity, with a constant traverse speed of 0.41 m/ min. The plasma current was 150 A and the wire extension 40 mm (1.57 in.). For welding with positive polarity on the wire, 8 liters argon/min plasma gas was used, and for negative polarity welding 0.1 liters CO,/min was added. In Fig. 4A the drop mass is plotted as a function of the current in the filler metal (l w ). It is obvious that, with 1.6 mm (0.063 in.) diameter filler metal, larger drops detach than with the same current and 1.2 mm (0.047 in.) diameter filler metal. The difference between positive and negative polarities of the filler metal is marked only below 140 A in the 1.2 mm (0.047 in.) diameter filler metal, when significantly larger drops are detached with negative polarity welding. At currents below 110 A drop detachment at the negative pool becomes irregular, with very large drops. Figure 4B shows the relationship between the drop frequency and the magnitude of the current, l w. In this respect there is no difference between positive and negative polarity when 1.2 mm (0.047 in.) diameter filler metal is used. The diameter of the filler metal does have a great influence, however, at least with welding currents above 130 A. For the same value of l, a 1.6 mm (0.063 in.) diameter delivers fewer drops per second. The velocity of the drops at the moment of impact with the weld pool as a function of l is plotted in Fig. 4C This shows clearly that there is a difference between the three examples: drops from a 1.2 mm (0.047 in.) diameter filler metal are faster with positive polarity than with negative polarity, but drops from the 1.6 mm (0.063 in.) diameter filler metal are slower, at least at high values of l.. The last result is to be expected, since the current density, which determines the magnitude of the electromagnetic pinch force, is greater in the 1.2 mm (0.047 in.) than in the 1.6 mm (0.063 in.) diameter filler metal (for the same value of current) while the restraining product of average momentum and drop frequency. v xp (10" kgm/sec I Fig. 5 Depth of penetration of weld bead as function of v x P: x 1.2 mm mild steel filler metal, positive polarity; mm filler metal, negative polarity;o 1.6 mm filler metal, positive polarity oi the metal, (v in illustration has same meaning as i) force through surface tension is smaller, being proportional to the diameter. The lower velocity of the drops detaching from the negative wire can be explained from the fact that, at least for part of the time, the point of contact of the cathode was somewhat above the detaching drop. As a result the current did not always pass through the liquid drop, so that the pinch force must have been smaller than with positive polarity. For the whole range of currents used, the various diameters of filler metal and different polarities the kinetic energy transferred to the workpiece by these drops was less than 1J/cm weld length. Thus in comparison with the heat content of the drops (Table 1), their kinetic energy can be neglected. In the following sections it will be shown that the high speed of the drops is important in another way. Momentum of Drops High-speed cinematography shows (hat the impact of each drop causes a marked indentation in the weld pool, particularly at high currents. At low currents, i.e., 100 A through a 1.2 mm (0.047 in.) diameter mild steel filler metal, the pit fills immediately after impact and there is an appreciable lag before the following drop creates a Fig. 6 Schematic representation of arc deflection with a transverse magnetic field: 1-welding torch; 2 filler metal; 3 arc; 4 workpiece; 5 electromagnet new pit. At higher currents, i.e., 170 A and more through the 1.2 mm (0.047 in.) diameter filler metal, the pit no longer fills before the subsequent impact. The drops fall constantly into the same pit so that the heat contained in the overheated drops is transferred very efficiently to the bottom of the weld pool. This causes the well-known fingershaped penetration. For this reason we have calculated the momentum of the drops when they reach the weld pool. In Fig. 5 the depth of penetration of the weld bead is plotted against the product of the average momentum (P) of the drops and the drop frequency (/). It appears that there is a relationship between the value of the "total impact per second" and the depth of penetration, irrespective of the history of the drops, whether they have been formed from 1.2 or 1.6 mm (0.047 or in.) diameter filler metal, or whether the wire has positive or negative polarity. When we consider GMA welding processes in which the transferring metal follows more-scattered paths we find that the penetration exhibits other profiles. With CO, welding, for example, the transfer of metal is much less focused on one point; in this case the drops strike the weld pool over a wide area. As a result, there is no fingershaped penetration. The same effect can be seen with plasma-gma and GMA welding when very high current densities and short electrode (filler metal) extensions are used. No true rotational transfer occurs, but a small conical arc is created. High-speed cinematography has shown that the very thin tip of the wire tends to rotate. In consequence the drops are less directed to one point only, and finger-shaped penetration is absent. This applies even more to welds made with rotational transfer. In that case the drops are deposited along the circumference of a wide circle having a diameter of 10 mm or more. The total impact per second at any point where the drops strike the weld pool is so small that penetration is very limited. Magnetic Effects on the Direction of Movement of the Detaching Drops From the considerations given above it appears that with GMA and plasma-gma welding, the penetration of the weld in the workpiece takes place mainly at the point where the drops fall into the weld pool. As a rule that is also the point at which the arc strikes the workpiece. However, as will be seen from the following, this is not 40-sl FEBRUARY 1981

5 0 o v, r, Fig 7 A motion analysis of two successive drops detaching irom a 1.2 mm diameter filler metal (mild steel), wire current = 200 A, plasma current = 100 A; the alternating field was pulsed with a frequency of 31 Hz. B drops deposited on a rapidly moved plate (1.5 m/min), when oscillating the wire tip. C cross-section of a bead on plate when the traverse speed is 0.23 m/min; otherwise all the parameters were the same as with A and B 0 0.5*10 ' 10' magnetic induction ( T ) Fig. 8-Penetration depth (A) and weld width at P-1 mm (B) as a function of the magnetic induction of the pulsed alternating transverse field. The pulse frequency was 85 Hz. The current through the 1.2 mm diameter mild steel filler metal was 230 A and the plasma current was 100 A. 1 T = 10* Gs Fig. 9-Cross-sections of bead-on-plate tests: A without a transverse magnetic field; B with a pulsed 85 Hz altetnating transverse magnetic field, of 0.4 x 10-' T; C-with a sinusoidal alternating transverse magnetic field (20 Hz). Wire current-230 A, plasma current-100 A, 1.2 mm diameter mild steel filler metal always necessarily true. The point of impact of the drops on the workpiece can be influenced in a number of ways. One which is normal with arc welding is to move the torch in the required direction. If this is done at a given frequency and speed the result is types of wetting-in which cannot otherwise be obtained. Another method is to deflect the arc by means of a transverse magnetic field. This method has been described several times in the literature, by Dilthey/ Dick 8 and Akulov et al", among others. An arrangement for magnetic arc deflection is shown schematically in Fig. 6. The field of the two coils are at right angles to the arc and parallel to the direction of welding. If an alternative magnetic field is used, the arc will oscillate at right angles to the welding direction at the same frequency as that of the field. The deflection of the arc is proportional to the strength of the magnetic field and the current through the arc With GMA welding, and also with plasma-gma welding, the use of such a magnetic field results not only in oscillation of the arc but also causes Fig. 10 Root run oi a V-groove joint without support at the back of the joint; 185 A through 1.2 mm diameter mild steel filler metal and 150 A plasma current. The pulsed alternating transverse magnetic field had a frequency of 46 Hz; the wire speed was pulsed with a frequency oi 92 Hz the detaching metal drop to move in the same direction as the arc. The liquid drop, so long as it is still attached to the wire, carries current and is, therefore, subjected to a similar force as that on the arc. In the last phase of detachment in particular, when a liquid neck has formed, the direction of this metal particle will also be determined by the magnetic field. When the drop detaches, it will move further in the direction given to it by the magnetic field at the moment of detachment. If the frequency of the magnetic field is set at half that of the drop frequency, a drop will detach each time the end of the wire reaches its maximum deflection. This is particularly the case with a pulsed magnetic field. The oscillation of the wire end then promotes a regular detachment of the drops, probably as a result of the extra acceleration imposed on the drop with respect to the wire end at each maximum deflection of the liquid section High-speed cinematography was used to investigate this method of drop detachment and the movement towards the weld pool. Figure 7A gives the motion analysis of two successive drops. The plasma-gma process was used with a 1.2 mm (0.047 in.) diameter mild steel filler metal. The distance of the drops when falling into the weld pool, was 9 mm (0.35 in.). As the film shows, the liquid tip of the wire moves to the right and to the left under the influence of the pulsed alternating magnetic field. It has already been remarked that the drops detach when the wire filler metal end is at its maximum deflection. This process can be enhanced by giving the wire motion an accelerating pulse at that moment. A current pulse is less suitable for this purpose since the force on the wire filler metal tip: F = / x B (3) where F = the Lorenz force on the wire tip induced by the transverse WELDING RESEARCH SUPPLEMENT I 41-s

6 magnetic field; / = the current density in the wire tip; B the induction of the transverse magnetic field. F also increases as a result of the increased current density, so that the deflection of the filler metal end is changed. An accelerating pulse on the wire filler metal speed, however, has a definite effect on drop detachment: the drop is shaken off the end of the wire. By synchronizing the accelerating pulse with the change over of the magnetic field, each drop is shaken off the wire at just the right moment. Figure 7A indicates that the drops follow a path which is in line with the alignment of the wire tip at the moment of detachment. The film speed was 3000 frames per second. The position of the drop in the figure was obtained from every 5th frame of the film. If a workpiece in the form of a metal plate is rapidly moved under such an oscillating metal shower, a result as pictured in Fig. 7B is obtained. The material has no chance to melt completely together and the drops lie apart, or in some cases fused together at the points on which they were deposited. This is well demonstrated in Fig. 7B. On the metal plate in Fig. 7B there are only three drops fused together. The speed of the plate under the torch was 1.5 m/min (59 ipm). The bead on plate shown in Fig. 7C was made under conditions different from those of Fig. 7B, in that the traverse speed was lower (i.e., 0.23 m/min or 9 ipm). Otherwise, all the parameters were the same, including the dimensions of the plate on which the weld was made. Figure 7C shows once again that penetration occurs at the point where the drops strike the plate. The deflection of the wire tip decreases and increases proportionally with the strength of the magnetic field. The effect of variation of the strength of the magnetic field on the shape of the weld bead, with constant traverse speed, is shown in Fig. 8. The bead-on-plate specimens were all made with 230 A through a 1.2 (0.047 in.) mm diameter mild steel filler metal and a plasma current of 100 A. Only the strength of the magnetic field was varied. The alternating transverse magnetic field used in these tests was pulsed with a frequency of 85 Hz. In Fig. 8A the penetration depths in the bead-on-plate specimens are plotted against the magnetic induction of the field. Since with increasing amplitude of deflection of the oscillating wire tip the drops land further and further from the center of the weld, so that their heat is dissipated over a greater area, the penetration steadily decreases. In Fig. 8B the width of penetration, measured at 1 mm (0.039 in.) above the deepest point, is plotted against the magnetic induction. This shows that the width can increase by a factor of 3, with 40% less depth of penetration, as shown in Fig. 8A. The cross-sections of a few of these bead-on-plate weld tests are shown in Fig. 9. Figure 9A is the cross-section of a bead-on-plate weld without the effect of a transverse magnetic field. The penetration has the well-known finger shape. The width of penetration was 2.5 mm (0.10 in.). Figure 9B shows the cross-section obtained with a pulsed alternating magnetic field with a strength of 0.4 X 10- T = 40 Gauss. The width of the penetration is doubled to 5 mm (0.20 in.). If a sinusoidal alternating magnetic field is used (with a frequency so related to the drop frequency that the drops are spread regularly across the width of the weld), a completely different penetration profile is obtained. The conditions for the weld illustrated in Fig. 9C were identical to those of Fig. 9A, except that a sinusoidal transverse magnetic field was used to spread the drops. The field alternated with a frequency of 20 Hz. In contrast with the finger-shaped penetration of the bead-on-plate weld as in Fig. 9A, a very regularly-distributed penetration was obtained across the whole width of the weld. This effect can be very useful in some practical applications. Finally, Fig. 10 shows a possible application of a pulsed alternating magnetic field. With single-sided welding of the root run of a V-groove joint without support at the back of the joint, the hot metal falls out of the joint when a spray-arc is used for welding. In the test described below our objective was as follows: If we could succeed in directing all the drops to the plate edges alongside the gap, the arc would be able to pass over the gap regularly but without any unfortunate consequences. As has been demonstrated above, it is the impact which determines the depth of penetration. Figure 10 shows the cross-section of such a root run. The current through the 1.2 mm (0.047 in.) diameter mild steel filler metal wire was 186 A and the plasma current was 150 A. Since the drop frequency with these currents was 92 Hz the frequency of the transverse magnetic field was set at 46 Hz. In addition, at the end of each maximum deflection of the wire tip the wire speed was given a very brief accelerating pulse, also at a frequency of 92 Hz. The result was a very regular weld penetration. Conclusion Under the conditions described in this paper, 34% of the heat transferred to the workpiece during GMA welding is supplied from the arc as a result of convection, radiation and conduction. The heat generated in the cathode area contributes 41% while 25% is provided by the overheated metal drops. The heat of the arc and the passage of current from the arc through the workpiece has only a slight influence on the penetration depth when the plasma- GMA process is used in the conditions described. In plasma-gma welding, and also in GMA welding, it is the heat in the transferring drops which determines the mass of workpiece metal which is melted. The overheated metal drops deliver their excess heat to the molten pool. This effect is more intense with high currents, since the drops are then driven deep into the weld pool as a result of the rapid succession of impacts. The penetration is a function of the magnitude of f X P, the total momentum per second. Penetration of the weld in the workpiece can be regulated during GMA and plasma-gma welding by controlling the direction of movement of the transferring metal drops. An alternating transverse magnetic field is an effective measure for this purpose. With the help of such a field the drops can be spread over the whole width of the weld or focused at predetermined locations. As a result, the shape of the penetration can be varied from the normal finger shape to one which, although less deep, is far wider. References 1. Essers, W.G., and Walter, R., "Some aspects of the penetration mechanisms in metal-inert-gas welding," International Conference on Arc Physics and Weld Pool Behaviour, London, May Essers, W.G., "New Process Combines Plasma with CMA Welding," Welding journal, 55 (5), May 1976, pp Ton, H., "Physical properties of the plasma-mig welding arc," /. Phys.D. (Appl. Physics), 8 (8), 1975, pp Ando, K., and Nishiguchi, K., "Mechanism of formation of pencilpoint-like wire tip in MIC welding," IIW Doc lelmorini, C, Tichelaar, G.W. and van den Heuvel, G.I.P.M., "Droplet temperature measurements in arc welding," IIW Doc Erokhin, A.A., "Kinetics of the metallurgical processes in arc welding" (in Russian), Moscow, Dilthey, U., "Beitrag zur Lichtbogensteuerung durch magnetfelder bei mechanisierten Lichtbogenschweiszverfahren." "Dissertation," 1972, Technische Hochschule, Aachen. 8. Dick, N.T., "The application of magnetic fields to TIG welding arcs," Welding Research International, 2 (1), Akulov, A.I., and Kopaev, B.V., "Magnetic control of the arc during argon MIG welding," Avt. Svarka, 1972, pp sl FEBRUARY 1981