Robotically welding grooved butt joints

How robotics can assist with large components too complex to fixture

Over the years article after article has been written about how robotics can help improve the quality and efficiencies of the welding process. Most of these articles tackle applications in which light-gauge components are welded in rigidly designed fixtures that provide excellent fit-up and minimal variations to weld joints from part to part.

But what if your components are too large to fixture and you have to live with the large tolerances that result from cutting, forming, and manual fitting processes? What if you are welding butt joints on 3-in.-thick plates and the plate edges were flame-cut? Can robotics still improve quality and efficiency for these applications? The short answer is yes.

One of the most challenging aspects of these applications is making butt welds on groove joints because they generally have large variations requiring both positional and volumetric changes to create good-quality welds.

Conventional Robot Programming Approach

With the conventional approach to robotic weld setup, the robot is programmed by manipulating the mechanical arm (jogging) and recording (teaching) a series of positions (points) in space and along the weld seam (see Figure 1). Depending on the robot and welding equipment used, the weld parameters can be set up in files or schedules in the robot data structure. They can also be directly coded in the robot program or set directly in the welder as a “job” (this last is the least flexible option).

After the program is completed, the robot simply replays this structure with a high degree of repeatability. If there are no variations in part fit-up and part locations, the conventional approach works perfectly.

The fact is, when welding large components, it is rarely possible to apply the conventional approach unless the welded assemblies are very simple and can be fixtured easily. As anyone who has ever welded groove joints on plate with any type of automation can attest, variations in fit-up can wreak havoc on overall weld quality. Joint misalignment and variation in edge preparation and groove openings can lead to many quality issues and ultimately joint failure if not handled properly. Fit-up variations can also lead to weld cost increases, including increased weld times and increased consumption of weld filler metal and gas.

The challenge in automating these welds lies mostly with the variations in fit-up from part to part.

If we look at common groove design such as a V groove with the following dimensions, the weld cross-sectional area (excluding the weld cap) is shown in Figure 2.

If the root opening on this joint is increased by just 1/16 in., the weld volume increases by 11.2 per cent, which is equivalent to approximately two full ¼-in. fillet welds. This information is in no way new; however, understanding this problem is key to developing a successful robotic welding operation.

Figure 1

Seam Tracking and Touch Sensing Limitations

Traditional robotic processing of large weldments on which these joints are usually found generally uses through-the-arc seam tracking and touch sensing functions (see Figures 3 and 4). These functions are used to find the joint and then track the joint while welding. They are also available in most robots via software and some hardware.

While this approach is very common and is still considered an advanced processing setup, there is a major limitation with its implementation for welding groove joints. As noted previously, groove joints on large components most often have large variations in fit-up, causing the weld volume to vary. Here are a few added challenges the robot needs to address:• Increased joint volume will lead to weld joint underfill.• Decreased joint volume will lead to weld joint overfill. • Depending on the severity of the changes in fit-up, weld pass and layer sequences may need to change to avoid cold lap between passes. Cold lap can lead to lack of fusion and inclusions.• Changes in the joint fit-up may lead to nozzle drag.• Side-wall fusion issues may occur.

Problems with simply using the through-the-arc seam tracking and touch sensing functions alone occur on these types of weld joints as well. For instance, welding parameters do not adapt to the changes in the weld joint volume.

If adaptation is used, there is no way to verify if the joint should even be welded or not. For example, if the weld groove is too small or too large to be welded successfully, the robot would still continue the process.

Depending on the joint design, the weld pass and layer sequences may have to be altered when the joint varies beyond a certain point. In these cases, even using these advanced heavy welding application features, the robot simply welds using the preset parameters regardless of the condition of the weld joint.

Improved Methodology for Multipass Groove Joint Welding

With a clear understanding of some of the challenges, you can customize an approach to automatically compensate for these changes.

The first step is to provide an accurate joint mapping method with enough resolution for your application. With the joint mapping process, before welding the robot searches and maps out the joint, determining accurate positioning of the weld joint and volume information so that the welding parameters can be adjusted accordingly. Specialized algorithms are also created that use the joint measurements to determine if the weld pass and layer sequence have to change with the changing groove dimensions. Algorithms are also created that prevent cold lap between layers. Also, all collected joint mapping data can be used for quality control purposes if required.

Joint analysis is the next step. As mentioned previously, one of the problems with welding these joints with traditional robotic methods is that when the joint is searched properly and the location can be found repeatedly, the search simply provides an offset so the robot welds in the right location. If the joint fit-up is in a condition that should not be welded, the robot will weld it regardless. Joint analysis provides a means of checking and verifying against a set of acceptable tolerances. With joint analysis, the robot analyzes the joint and then makes a decision whether to weld or stop, depending on the configuration. Alarms are initiated when a joint fails analysis, and the operator then can determine a proper course of action for the failed joint, such as welding by hand or adjusting joint prep. When using this methodology, you can achieve a very high degree of quality as it will prevent any groove joint that has undesirable fit-up from being welded. It is also a very clear indicator of how good your manufacturing processes are that are used in the fitting process.

Figure 2: Cross-sectional area (excluding cap): 0.56 in2

A Library-based Approach

In addition to joint mapping and joint analysis, using a library-based approach to manage welding data can make the processing of these welds faster. The library data, stored in the robot, contains the specific parameters that enable the robot to autonomously create the weld path. For example, a set of library data would be unique to a joint style or a machined boss to be welded to a plate.

These library parameters are used, among other things, to provide data that feeds a series of algorithms used to search the joint and provide triggers for joint pass/failure status to create the required weld parameters.

The advantage of this approach is that overall programming time improves dramatically because the final “point programming” only requires a few reference positions to be taught to give some general directional information to the robot.

The library system allows the robot programmer to very quickly add new joints simply by configuring the robot program with the new library. This triggers new searching algorithms for the new style of weld joint and also pulls the corresponding analysis data from the library to compare against the joint mapping data. Final programming requires only a few taught points for directional purposes.

The biggest advantage of the library-based approach is that it can significantly reduce the time it takes to get new parts running on the robot. Quite often it is possible to produce “first off” quality production parts, which is difficult to achieve with a welding robot in these types of applications.

The total end result of this methodology can provide high-quality welds on parts with large variations and it does this with a high degree of safety. Combining this methodology with the library-based approach can also help increase the throughput of production parts.

Kevin McWhirter is president of Autonomous Welding Inc., 705-288-4887, www.autonomouswelding.com.