October 15-17, 2024 at Orange County Convention Center in Orlando, Florida
FABTECH Shows ▶
Register
An info icon
A close icon

Join Us in Orlando

Attend Exhibit

Expo Hours

Tuesday Oct 15 9 AM — 5 PM
Wednesday Oct 16 9 AM — 5 PM
Thursday Oct 17 9 AM — 4 PM

Venue

Orange County Convention Center

9800 International Drive
Orlando, Florida 32819

About the Expo

North America's largest metal forming, fabricating, welding and finishing event heads to Orange County Convention Center in October 2024. FABTECH provides a convenient "one stop shop" venue where you can meet with world-class suppliers, see the latest industry products and developments, and find the tools to improve productivity, increase profits and discover new solutions to all of your metal forming, fabricating, welding and finishing needs.

A search icon

North America's Largest Metal Forming, Fabricating, Welding and Finishing Event

From the Blog

New Solutions for Low-Volume, High-Mix Automation

By Jennifer Dallos on on
By Louis Dicaire and Ian McLaren Louis Dicaire is vice president of Marketing, AGT Robotics, and Ian McLaren is global product manager Robotics Automation, ESAB (esabna.com), United Kingdom. Reprinted with permission: The AWS Welding Journal Recent advances in software, vision systems, and 3D laser scanning now combine to enable what was previously impossible: profitable automated solutions for welding, cutting, gouging, and grinding in low-volume, high-mix applications. Using self-learning robotic technology, unique parts that took 25 or more hours to fabricate now take 45 min. By reducing cost per part and enhancing long-term cash flow, companies that once fabricated parts overseas are reshoring their work and simultaneously growing domestic employment by winning more contracts. Good candidates for self-learning robotic applications include those that have a notion of family in their components. Structural steel I-beams provide a prime example. A typical beam will have a W section (main beam), a start plate and an end plate, a few stiffeners, and some connector plates. With just a few common parts, their dimensions can be adjusted to build an infinite number of I-beams that satisfy any need. Previously, structural steel building and bridge components would be entirely welded by an operator using a combination of gas metal arc welding (GMAW) for tacking and flux cored arc welding (FCAW) for the primary welds. Now, self-learning robotic technology can easily weld such parts.
Dicaire lead picture

The next generation of low-volume, high-mix automation welding stations can also include automatic positioning of components, such as this clip.

Laser Scanning For low-volume, high-mix robotic welding to become reality, fabricators first needed a way to reduce the time and cost associated with programming a robot trajectory. Traditionally, trajectories are programmed offline or with a teach pendant, and programming takes hours per part (perhaps 10 to 30 for an I-beam). If part dimensions change (such as changing the length of the beam from 25 to 30 ft), the part requires a new program. Primarily because of programming time, robotics were used only in applications where costs could be disbursed over a high volume of parts. Now, instead of manually programming a robot trajectory, companies can produce a trajectory using a combination of automated technologies and without programming knowledge. Ideally, a fabricator’s customer supplies a CAD file with embedded part dimensions and welding data. From the CAD file, next-generation software can extract the information necessary to generate a trajectory. Unfortunately, most situations aren’t ideal. In that case, creating a robot trajectory starts with 3D laser scanning. Continuing with the I-beam as an example, a fabricator would start by placing the beam, plates, flanges, and connectors on a work surface, such as a conveyor tray or turntable that feeds a robot work cell. The scanner, which can be held by the robot or placed on a gantry, then projects a beam of laser light onto the component and moves along the length of (and/or around) the components. Sensor cameras record the changing shape and distance of the laser lines so that the scanning system can then generate “point clouds.” Based on tens or hundreds of thousands of measurements per second, 3D scanners can measure parts with tolerances as tight as ± 0.02 in. in shop floor conditions. While different systems use different technologies, they all achieve basically the same end: a point cloud from which a software package can then generate a highly accurate CAD model of a part’s exact size, shape, and surface geometry. Even when a fabricator’s customer provides CAD files with embedded part dimensions and welding data, a low-volume, high-mix robotic welding system will use 3D laser scanning to detect and compensate for part imperfections or irregularities (such as a distorted beam). In addition, they can detect situations where a human welder accidentally tack welded a stiffener a fraction of an inch in the wrong direction. By automatically accommodating for unpredictable deviations, 3D scanners — in conjunction with robot trajectory software — can nearly eliminate one of the larger causes of weld rejects in robotic applications. Taking the technology one step further, the 3D scanner can scan the entire work area to prevent accidental collisions of the robot arm. In summary, the robot is aware of everything in its work environment. Expert System Knowing what and where to weld differs greatly from knowing how to write good welding procedures. When customers do not provide procedures, software programs are now available that can extract information from an expert system and apply it to a 3D-scanned joint. An expert system stores information from previous jobs (or contains a database programmed by a welding engineer) to contain welding procedures for specific joints. The system continues to learn as new jobs are added. As good as an expert system is, note that a welding engineer experienced with robotics can almost always improve speed, quality, and consistency by adjusting torch angles, wire stickout, contact-tip-to-work distance, and other welding parameters. Pipe Elbow Case Study A Canadian fabricator of pipe elbows for petrochemical applications provided a good demonstration of self-learning robotic technology. In this case, the part had four primary parameters: material thickness, radius, diameter, and chamfer (bevel). With a common “family” of dimensions, this part was a good candidate for low-volume, high-mix automation. The fabricator received two forged elbow halves (imagine a piece of macaroni split in half lengthwise) from its supplier. To fit the parts together, the fabricator had to spend 25 to 100 h grinding, trial fitting, and more grinding to match the two halves of this high-alloy part. The fabricator originally sought a robotic grinder to reduce labor effort. What it really needed, however, was a self-learning robotic plasma cutting system, a solution that would reduce weld preparation time from dozens of hours to 45 min. In this solution, each half of the elbow is placed open side up on a turntable and clamped in place simply so it doesn’t move (e.g., no precise fixtures). The robot picks up a 3D scanner, scans both halves, and then the software generates a perfect cutting path, checking the cut paths with a virtual fit. The robot sets down the scanner and picks up the plasma torch and cuts using a path generated specifically for those two halves. Finally, the robot picks up a grinding tool for de-burring and surface preparation. The entire process takes just 45 min. Not only does the fabricator reduce labor costs, the fabricator can free labor for other tasks, address labor shortage issues, and even reduce delivery time, being more responsive to its customers. Productivity and Payback In 20 years of conducting arc-on time studies, the leader of ESAB’s Value Added Engineering (VAE) team consistently confirms that a good welding operator using the GMAW process averages 17 to 20% arc-on time. In a fully optimized robotic application, the arc-on time averages 55 to 65%. Part of the increase comes from the fact that robots don’t need to take physical breaks, aside from scheduled maintenance. However, most of the increase comes by designing a robotic work cell with multiple work zones. When possible, cells also use multiple robot arms, as was the case for a manufacturer of industrial shelving, or pallet racks. The racks have the same basic profile as an I-beam, specifically a beam channel with an end-plate requiring a fillet weld on one side and flare bevel weld on the other. Beams could be 4 or 8 ft long. Between the two robots and a “Ferris wheel” positioner/fixture that allowed for part loading/unloading on one side while the robots welded on the other side, the customer increased output by 200 to 300% on one shift. But keep in mind that a robotic work cell costs the same to purchase whether the cell operates for one, two, or three shifts. Thus, while acquisition costs remain static, companies that run two or three shifts amplify their return on investment and significantly increase cash flow by selling six to nine times more product. In the case of the pallet rack manufacturer, the company will soon have a total of four robotic cells, each generating new income for the company. In addition to greater arc-on time, robots produce more parts per hour because they can increase travel speed, especially with an advanced process. For example, a typical manual GMAW speed is 11 to 13 in./min for a 5⁄16-in. fillet weld vs. 12 to 14 in./min for a robot. Here, the cycle time gains are minimal. However, some applications, particularly those with plate thicknesses from 0.08 to 0.6 in., may benefit from the advanced GMAW process that can take advantage of a robot’s speed and “endurance.” For example, ESAB’s Swift Arc Transfer, an enhanced spray transfer GMAW process, can produce travel speeds of 49.2 in./min when making a fillet weld on 0.4-in. plate (0.035-in.-diameter ER70S-6 wire, 92% argon/8% CO2 shielding gas). Qualifying Candidates Of course, not every component with a family notion of parts is a good candidate. One key for successful automation is to choose components that are truly well suited for the process. Secondly, consider modifying component designs (if allowable) to enhance them. For example, a clip angle on an I-beam typically has four welded joints. However, by eliminating one of the welds, it enables better fitup when connecting to a plate from another beam. In this case, the clip angle weld procedure could be modified to eliminate potential fitup issue. Small changes with big consequences should provide sufficient motivation to engage with an automation integrator. Even if a company has robotic and/or welding engineers on staff, those resources will not likely have the breadth or depth of experience of companies who deliver turnkey solutions every day. As welding industry trends indicate, the business model for automation works most effectively when it combines companies who excel at robotics and motion (integrators) with companies who excel at welding (the major welding equipment and technology providers). Put another way, it’s easy to be good at several things, but being excellent requires focus. Open to Change During the process of acquiring a robotic system, prospective customers should not necessarily expect their current welding procedures to be the most effective welding procedures. Several scenarios occur frequently enough that they are worth mentioning.
  • Shifting to the next smallest wire diameter while maintaining a similar wire feed speed and slightly reducing voltage. Using a smaller wire increases current density, which in turn enables an increase in travel speed and deposition rates while reducing heat input.
  • Shifting from solid to tubular wires, typically a metal cored wire. Metal cored wires can better compensate for gaps, catch both sides of the plate and provide good tie-in at the toes of the weld. In addition, they often reduce spatter issues to minimize postweld cleanup, and they tolerate mill scale much better.
  • Shifting to a high-speed welding process, often a modified spray transfer process. As noted above, these processes can weld at or in excess of 50 in./min. A human welder simply cannot move at this speed for an eight-hour shift and maintain consistent quality.
Caption for lead picture: The next generation of low-volume, high-mix automation welding stations can also include automatic positioning of components, such as this clip.  

2024 Platinum Sponsors

2024 Sponsors

Become a Sponsor