New developments in advanced welding

Review of process mechanism

Most researchers believe that arc constriction increases the current density and heat intensity at the anode root, enabling a substantial increase in penetration depth to be achieved (Howse and Lucas, 2000). A-TIG shows a visible constriction of the arc when compared with the more diffuse conventional TIG at the same current level (Simonik et al., 1976; Ostrovskii et al., 1977; Savitskii, 1979; Savitskii and Leskov, 1980; Howse and Lucas, 2000). It has been suggested that the arc constriction is produced by the effect of vaporised flux elements, namely oxygen or halogens (such as fluorine), capturing electrons in the outer (cooler) regions of the arc owing to their higher electron affinity as shown in Fig. 3.13 (Simonik et al., 1976; Savitskii 1979; Howse and Lucas, 2000).

Savitskii and Leskov (1980) have proposed a further mechanism involving the interaction between the arc and the surface of the weld pool, because they could not observe arc constriction on a water cooled copper anode with

the flux. Since the surface curvature of the weld pool depends on a balance of arc pressure (cathode jet) and surface tension, the lower surface tension caused by the oxygen in the A-TIG flux should lead to greater surface curvature. This being the case, the cathode jet will be less able to dissipate metal vapour from the pool surface to the outer regions of the arc because the pool surface itself becomes an obstructive wall for the metal vapour, similar to a keyhole. As a result, the arc should be constricted owing to the concentration of metal vapour with low ionisation potential in the centre region of the arc. However, Savitskii and Leskov (1980) did not take account of the convective flow in the weld pool, which is described below.

Ostrovskii et al. (1977) argued that the main mechanism was a recirculatory flow driven by the electromagnetic force (the Lorentz force) resulting from the increase in current density at the anode root. He believed that the strength of the cathode jet should decrease and so not cause greater surface curvature of the weld pool, as stated in Savitskii and Leskov (1980), because the pressure difference between the cathode and anode would become very small as a result of the arc constriction. In fact, the keyhole phenomenon is not observed in the A-TIG process.

Heiple and Roper (1981, 1982) proposed that the change in the magnitude and direction of surface tension gradients at the weld pool surface caused by surface active elements such as oxygen, sulphur, selenium, etc. should change the direction of a recirculatory flow, namely marangoni convection, in the weld pool. Figure 3.14 shows schematically the model of marangoni convection driven by the temperature coefficient of the surface tension (Ohji et al., 1990). Heiple and Roper (1981, 1982) suggested that an outward fluid flow with a wide and shallow weld was caused by a normal negative temperature coefficient of surface tension, whereas an inward fluid flow and resultant narrow and deep weld was caused by a positive temperature coefficient (see Fig. 3.14).

Recently, Tanaka et al. (2003) proposed a numerical model of the weld pool taking account of the close interaction between the arc plasma and the weld pool. The time-dependent development of the weld penetration was predicted at a current of 150 A in TIG welding of stainless steels containing low sulphur (40ppm) and high sulphur (220 ppm). It was shown that calculated convective flow in the weld pool of an argon-shielded TIG process was dominated by the drag force of the cathode jet and the marangoni force. The other two driving forces, namely, the buoyancy force and the electromagnetic force, were significantly less important. Tanaka et al. (2003) also concluded that change in the direction of recirculatory flow in the weld pool led to dramatically different weld penetration geometry.

СО

Temperature (T)

3.3.4 Current understanding of process mechanism

From Sections 3.3.3 and 3.3.4, the most plausible mechanism at present is proposed, as follows.

In the case of the conventional TIG process an outward fluid flow in the weld pool is caused by the drag force of the cathode jet and the marangoni force associated with a normal negative temperature coefficient of surface tension. This causes the heat input from the arc to transfer from the centre to the edge on the weld pool surface. This heat transfer leads to a shallow gradient in surface temperature across the weld pool. It also leads to the metal plasma distribution being expanded widely across the whole weld pool surface, owing to the much lower ionisation potential of the metal compared with that of the shielding gas. As a result, a diffuse anode root is formed. Strong convective flow outward at the surface of the weld pool leads to a shallower weld than heat transfer by conduction alone.

In the A-TIG process, the temperature coefficient of surface tension changes from negative to positive due to the surface active elements, such as oxygen, from the decomposition of the flux. The marangoni force associated with a positive temperature coefficient of surface tension is larger than the drag force of the cathode jet, and causes inward fluid flow. As a result of this inward flow, the heat input from the arc should transfer directly from the surface to the bottom of the weld pool. This heat transfer causes a steep gradient in the surface temperature of the weld pool, which also leads to the
metal plasma distribution being localised at the centre on the weld pool surface. Consequently, a constricted anode root is formed and the anode spot appears to be located at the centre of the weld pool surface. Furthermore, the constricted anode root should lead to higher current density at the anode, which should also promote the inward recirculatory flow driven by the electromagnetic force. Thus the multiplication effect of the electromagnetic force and the marangoni force appears to cause strong inward recirculatory flow in the weld pool. Strong inward convective flow of the weld pool leads to a deeper weld than for heat transfer by conduction alone.

This proposed mechanism suggests that the deep weld penetration can be achieved by the activating flux even if the welding process is changed from TIG to alternative processes such as plasma, laser and electron beam. In fact Howse and Lucas (2000) have reported that the flux equally increased the depth of the weld penetration for both the plasma process and the laser process, although it did not in the case of the electron beam process. The electron beam did not show major increases in penetration as a result of the activating flux although the beam power density was modified to simulate that of a typical TIG arc (Howse and Lucas, 2000). However, Ohji et al. (1991) reported that the deep weld penetrations were achieved independently of sulphur contents (20ppm, 60ppm and 90ppm) of the same type 304 stainless steels as enough defocused electron beam was employed for welding. In the electron beam process, only the marangoni force affects the convective flow in the weld pool because both the drag force of the cathode jet and the electromagnetic force can be neglected (Fujii et al. 2001). However, the marangoni force is strongly dependent not only on the temperature coefficient of surface tension but also on the sulphur (or oxygen) concentration coefficient of surface tension (Winkler et al., 2000). Ohji et al. (1991) suggested that the latter coefficient of surface tension was very important for understanding the phenomena of the weld penetration in the electron beam process because the evaporation rate from the weld pool was much higher than that in TIG. This is due to the vacuum environment used in the electron beam process. The convective flow caused by the gradient of surface tension in liquids was first reported by James Thomson (Scriven and Sternling, 1960). Thomson (1855) provided the first correct explanation of the spreading of an alcohol drop on a water surface, the well-known ‘tears of wine’, and related phenomena. These phenomena are convective flows caused by changes in surface tension driven by the evaporation of the alcohol and the existence of an alcohol concentration coefficient of surface tension in the wine.

3.3.5 Examples of applications

Typical applications of A-TIG are precision welds in relatively thick (3-12 mm) material where advantage can be taken of a single pass. The A-TIG should find particular application in the orbital welding of tubes (Lucas and Howse, 1996). The tube can be welded in a single pass with a simple square butt joint, while three or more passes would be required with conventional TIG. The A-TIG process is also suitable for thinner wall material because the welds can be made at higher speeds and lower heat inputs than welds with conventional TIG (Lucas and Howse, 1996). For example, A-TIG can make a weld joint of 2 mm thickness in AISI 304 stainless steel at 800mm/min welding speed, which is double that of conventional TIG, while the reduced heat input obviously results in less distortion (Okazaki and Okaniwa, 2002). Disadvantages of using A-TIG are the rougher surface appearance of the weld bead and the need to clean it after welding (Lucas, 2000). The as - welded surface is significantly less smooth than that produced with the conventional TIG, because there is significant slag residue on the surface of the weld produced with the A-TIG process. It often requires rigorous wire brushing to remove it (Lucas, 2000).

Some typical applications of the A-TIG process, such as in nuclear reactor components, car wheel rims, steel bottles and pressure vessels have been reported (Lucas and Howse, 1996). AISI type 316 stainless steel tube, 70 mm diameter and 5 mm wall thickness, was reported to be welded in a single pass with a simple square butt joint without filler wire by using the activating flux under the conditions of pulse current 150 A, background current 30 A, arc voltage 9.5 V and welding speed 60mm/min (Lucas, 2000). Full weld penetration was achieved independently of the orbital positions.

In another report (Kamo et al., 2000) AISI type 304 stainless steel tube, 60.5mm diameter and 8.7mm wall thickness, with 4mm root thickness and narrow-gap grove was reported to be welded using the activating flux. The welding conditions were welding current 100-130 A, arc voltage 10-11V and welding speed 80-100mm/min. The A-TIG required only three layers and three passes to complete the joint whereas conventional TIG required seven layers and seven passes. There were no defects such as insufficient fusion or cracking at any position, including the vertical-up and vertical - down positions, using the A-TIG process. The integrity of the weld joint produced by the A-TIG process was confirmed by both mechanical and metallographic tests (Kamo et al., 2000).

The A-TIG was also used for repairing cracks in metal at a nuclear power plant (Takahashi et al., 2002, Tsuboi et al., 2002). Type INC0NEL600 nickel - based alloy was reported to be welded using an activating flux cored wire as a filler wire while welding underwater at double atmospheric pressure. The integrity of the weld joint and a deep weld penetration about 4.5 mm were achieved in this process.

Sire and Marya (2001) have proposed a new technique, called the FBTIG (flux bounded TIG) process, for producing a deep weld penetration in aluminium alloys. In this process, silica (SiO2) flux was used to restrict the arc current to a narrow channel to enhance the weld penetration depth. The silica flux was pasted on the aluminium alloy surface, leaving a flux gap around the joint, whereas a fully flux coverage of the joint area was always maintained in the A-TIG process. The current was restricted to the gap due to the high electric resistance of the silica. It was possible to take full weld penetration of type 5086 aluminium alloy, 6mm thickness, at 175A and 150 mm/min with a flux gap of 4 mm.