Never Underestimate the Importance of Weld Bead Shape

Having come to the realization that we were basically confined to using this “tweener” root opening of ¼ in, how then should we deal with it? At this width, the joint is just wide enough that a side-to-side weave pass would be necessary to ensure that the arc energy would be able to melt the feather edge and the backing bar together without discontinuity. But as I mentioned in last month’s column (“Modified ‘5-55’ Vertical-Up Welding Technique Improves Bead Quality and Weld Shape” (Welding Tips, March 2012), the welder’s (or robot’s) weave technique can make all the difference in a successful weld.

Traditional automated weave techniques employ an oscillation from side to side, with the welding torch remaining perpendicular (90 deg) to the workpiece. Yet the inherent problem with this type of weave technique is that the arc force may not be powerful enough to punch through the sidewall and into the backing bar. This is particularly true if the weave width is ever-so-slightly too large.

Alternatively, some integrators have offered automated travel carriages have the capability of pivoting the torch about a fixed point. But the drawback with this type of weave technique is that the resulting contact-tip-to-work distance (CTWD) tends to be relatively short on the sidewalls and relatively long in the middle of the joint. This long CTWD with its associated lower welding amperage has been shown in some circumstances to actually cause cold-lapping in the middle of the joint. So, a pivoting torch in-and-of itself is not a truly viable option.

Going back to the traditional side-to-side weave technique, let’s assume for a moment that the robot programmer discovers the correct oscillation speed, travel speed and dwell time to ensure that complete fusion through the joint was achieved. Furthermore, all of this was accomplished while keeping the welding heat input below 35 kJ/in. How does this set us up for the next weld pass that needs to be deposited?

So now we’ve brought the focus of this analysis to the theme implied by the title – bead shape. With a ¼ in root opening and a simple side-to-side weave technique (maintaining the 90 deg angle to the workpiece), it proved to be a daunting task to avoid depositing a weld bead that was not convex in bead shape. In the context of this article, I will define a convex bead shape as one where a sharp transition of the bead to the sidewall is evident. I will also refer to this as a “valley”, which can be gleaned from the illustration in Figure 3.

Even when an extended dwell time (or pause) at the sides of the joint was programmed into the robot logic, a highly convex weld bead still resulted. These valleys were reminiscent of what pipeline welders often refer to as “wagon tracks.” Furthermore, because we were using an ER70S-6 GMAW welding wire, SuperArc® L-59, an unavoidable minimum amount of silicon islands was being produced during the welding process. Despite switching to the ER70S-3 GMAW welding wire, SuperArc® L-58 (with its lower level of silicon and manganese), there was little impact in reducing the number or size of these silicon islands.

When we combined even the slightest amount of convexity in the root pass weld bead with silicon islands trapped in these valleys between the weld and the sidewall, we had a recipe for disaster. These silicon islands cannot simply be “burned out” of the joint with the next “fill” pass due to the relatively low arc force of the GMAW process as mentioned above.

At this point we believed that if one of the two components of the “recipe” was removed, we would be able to achieve a weld that was free of trapped slag. But, as with everything else in our business, the proof is in the weld. So next we proceeded to test our theory that our success rate would be much closer to 100 percent if we were able to eliminate just one of these factors. This “test” was simply the application of a wire wheel to the joint to remove the silicon islands resulting during welding of the root pass, and subsequently duplicating our welding procedure as before.

But we quickly discovered that even when there weren’t any silicon islands embedded in the wagon tracks, these valleys still posed a problem. As mentioned earlier, there was a heat input restriction imposed, which essentially means that we are compelled to use a lower wire feed speed and hence a lower welding current. Because of this, during the fill pass the arc force was not powerful enough to penetrate completely through to the bottom of the valley – regardless of the presence or absence of silicon islands. The combination of these elements would manifest itself in what was interpreted on a radiograph as a line of lack of fusion at the sidewall.

So at this point, we had come to the conclusion that what was needed was a “smooth” transition to the sidewall. We realized that the very best way to accomplish this was to program the robot and/or an automated welding carriage to weld the joint in the same manner that a human being would weld the joint. That is, the welding technique should employ an oscillation and pivot. This would allow us to preferentially distribute the weld metal at the sides of the joint, as shown in Figure 4, as opposed to depositing the weld metal primarily in the middle of the joint.

Further welding tests confirmed that a slightly convex weld bead with a smooth transition to the sidewall made all the difference in preventing lack of fusion at the sidewall. Transfer of the success in our lab to the customer jobsite was ultimately achieved when a FANUC® robotic welding arm was combined with a travel carriage. Since the travel carriage removed the “responsibility” of vertical welding progression away from the robot arm, the chances that the robot arm would “drift” out of the joint were eliminated. Combining the repeatability of robotics with the oscillation-pivot weave technique that yielded a smooth transition to the sidewall, consistency over the entire 60+ ft joint was finally realized.