In many welding operations, an increase in productivity often requires the use of higher welding speeds. However, at higher welding speeds, the weld bead begins to exhibit serious geometric defects such as undercutting and humping. Undercutting is a groove at the weld toe that is left unfilled by the weld metal during solidification, while humping is the periodic undulation of the weld bead profile. At the present time, the physical mechanisms responsible for these high-speed weld defects are not well understood. Thus, it is not always clear as to what must be done in order to suppress or eliminate the formation of these high-speed weld defects, thereby facilitating further increases in productivity without sacrificing the overall quality of the weld.

One of the main objectives of this study was to comprehensively identify and document the occurrence of the high-speed weld defects with respect to the controlling parameters of the gas metal arc welding (GMAW) process. To accomplish this, a robotic GMAW system was utilized to produce bead-on-plate welds over a wide range of preset welding speeds and welding powers using SAE-AISI 1018 plain carbon steel plate, ER480S-6 and ER480S-3 electrode wires, Argon, Mig Mix Gold™ (MMG) and TIME™ shielding gases. A specialized LaserStrobe video imaging system was used to obtain video images of the events taking place during the formation of a weld bead defect without the intense light from the welding arc. When the welding power was between 5 and 9 kW, the GMAW filler metal transfer mode was spray transfer and the usable welding speed was limited by the onset of the humping phenomenon. When using Argon shielding gas, the limiting welding speed increased slightly from 9 mm/s to 15 mm/s as the welding power was increased from 5 to 9 kW. When using MMG™ and TIME™ shielding gases, the limiting welding speed was 47 mm/s and 59 mm/s, respectively, at 5 kW welding power. This was more than 400% and 550% increase in welding speed and productivity relative to the Argon shielded welds. However, unlike the Argon shielded welds, the limiting welding speeds of the welds made using the reactive shielding gases decreased to about 22 mm/s as the welding power was increased to 9 kW.

When using Argon shielding gas and more than 9 kW welding power, the metal transfer mode was rotational and the welding speed was limited to 15 mm/s by the onset of humping. However, when using the reactive shielding gas, the welding speed was limited to 22 mm/s by the onset of a new, as yet unreported weld defect that was distinctly different from humping. This new discontinuous weld bead defect was broken up into several good bead segments by regularly and irregularly-spaced valleys or depressions where melting of the base metal occurred but no filler metal was deposited. The discontinuous weld bead defect was found to be caused by the inconsistent deposition of molten filler metal during rotational transfer mode. The long molten filler metal string on the end of the electrode wire was erratically fragmented and required time to reform prior to the resumption of the transfer of filler metal. The temporarily disruption of filler metal deposition created a depression which broke up the otherwise good weld bead. The fragmentation of the molten filler metal string during rotational transfer and subsequent formation of the discontinuous weld bead defect are phenomena that have not previously been observed or reported in the open literature.

Based on direct observations made during the formation of humped GMA welds using the LaserStrobe video imaging system, a phenomenological model was developed to explain the occurrence of the humping defect. It was surmised that the momentum of the backward flow of the molten weld metal in a wall jet from the arc gouged front portion of the weld pool toward the tail of the weld pool is responsible for the hump formation at the tail of the weld pool. As well, the momentum of the backward flow of the molten weld metal in the wall jet prevents backfilling of the front portion of the weld pool. As welding continues, the thin wall jet becomes longer and eventually solidifies thereby choking off the continued flow of molten weld metal to the hump. With the continued flow of the molten weld metal towards the tail of the weld pool through the wall jet, a new hump begins to form further along the weld bead. This sequence of events results in the periodic undulation of the weld bead.

This phenomenological model of humping suggests that any welding techniques and processes that weaken the backward flow o f molten weld metal will suppress the hump formation until much higher welding speeds. It was shown that higher welding speeds were possible without humping when welding in the downhill position because of the effect of the forward acting component of the gravity vector on the rearward acting momentum of the molten metal in the weld pool. In addition, gravity acts to push the molten weld metal in the hump forward to promote backfilling of the weld pool. Higher welding speeds were also possible when the workpiece was preheated primarily because this decreased the rate of solidification of the wall jet and the weld metal accumulated at the tail of the weld pool thereby allowing more time for a good weld bead to form. Finally, it was shown that significantly higher welding speeds were possible when using reactive shielding gases because o f the effect of these gases on reducing the surface tension and improving the wetting of the molten weld metal. In addition, the molten filler metal droplets from the electrode were shown to impinge over a larger area of the weld pool. This results in a wider weld pool and reduced rearward momentum of the weld metal, thereby allowing higher welding speeds without humping.

Dimensional analysis was utilized to create dimensionless GMAW process maps that can be used to predict the occurrence of the high-speed weld defects using the preset GMAW process parameters. From the dimensional analysis, a dimensionless parameter that combined the limiting welding speed and the influence of the shielding gas was plotted against a second dimensionless parameter that represented the heat generated during GMAW. The result was a dimensionless GMAW process map that correlated well with data from the present study and previously published experimental data. The dimensionless process map shows a distinct boundary between good and defective weld beads as well as the transition from spray to rotational transfer. Using the dimensionless GMAW process map, the occurrence of a highspeed weld defect can be predicted for the first time using the controllable weld process parameters.