Faced with increased demand, systemic shop capacity restraints, fallout from supply chain problems, competition from low-cost countries, slipping profit margins, and labor shortages, some shops are setting up robotic welding to take up the slack.
As with any new technology, there are questions to ask and issues to consider before making a decision on the right system for a shop, including production mix, productivity and scrap rates, infrastructure needs, personnel, and expected return on investment. Threaded Reducing Flange
“The shop owner should consider how they are going to win more projects, and grow their business while facing shortages of parts and labor, especially in the context of a potential recession ahead,” said CEO Soroush Karimzadeh, of Novarc Technologies Inc. (North Vancouver, B.C.), a full-stack robotics company specializing in the design and commercialization of cobots and A.I. systems for robotic welding applications.
What robotic welding won’t do is eliminate the need for all welders. It also won’t fix problems encountered with manual setups.
Those limitations aside, it’s essential to identify and appoint the right champion in the shop to make the effort a success, according to industry experts.
“From our experience, the most important attributes are more character related and kind of the softer attributes of someone who’s willing to take ownership of it, and he’s eager to see it work well,” said Josh Pawley, vice president of business development and co-founder, Vectis Automation LLC in Loveland, Colo. “A lot of the technical knowledge can be learned, you know, from the internet—or from our training materials or from other folks in the shop.”
Where Vectis sees struggles occur is when the wrong champion is chosen—someone who’s likely to throw up his hands and quit when problems crop up.
The right champion is eager, willing to take ownership and undergo learning on the job, and flexible enough to accommodate some trial-and-error, according to Vectis. If he’s lacking in welding expertise, he can be paired up with a more experienced operator.
That’s what happened with Sam Guinn, a grinder/finisher at large-scale structure fabricator Mohawk Northeast Inc., Groton, Conn. Until he first saw a welding cobot, Guinn used to spend his days grinding down welds. He wasn’t a great manual welder, but he was motivated enough to self-teach himself robotic welding. The company subsequently paired him with one of its experienced welders, and now Guinn is an American Welding Society D 1.5. bridge welder.
What happened at Mohawk is that Guinn was essentially upskilled within the company due to his championing of new technology.
Guinn might have also made the dull job he left behind obsolete. Robotic welding can be of such consistent and high quality that it eliminates the need for post-process grinding like Guinn used to do because it’s already smooth and looks like it’s an extension of the part itself, Pawley said.
For parts that don’t come out looking as finished as Pawley described, shops that implement robotic welding and see an uptick in productivity can add services such as finishing to clean up surface contaminants and spatter or to smooth out metal burs, noted Josh Leath, senior product manager at the Motoman Robotics Division of Yaskawa America Inc. (Miamisburg, Ohio).
“Usually after someone implements their first robot, it overcomes any fears they had about automation, and they are ready to see what additional robots can do,” said Leath. “I heard someone in the industry recently saying, ‘If the job looks like a robot should be doing it, then a robot should be doing it.’ It could be looking at your top production parts, or those you commonly have issues with … or simply looking around the shop at jobs that seem extremely redundant or time consuming.”
His comment highlights a tip offered by many experts: When getting started with automation, delegate the most basic, highest volume or least desirable welding jobs to the robot, while reserving more complicated, challenging tasks to the expertise of human welders.
Fears about programming a robot may drop with the knowledge that some robotics manufacturers have been on a multi-year quest to simplify coding requirements and even make their human-machine interface look inviting.
“Programming your robot is not as difficult as most people perceive,” said Ryan Lizotte, technical service manager, Tregaskiss (Windsor, Ont.), a subsidiary of Illinois Tool Works Inc. “And especially with younger generations who have grown up playing video games and using controllers to do things, they pick it up very, very easily.”
For experienced manual welders, the weld positions and parameters should already be familiar, Lizotte noted. Help is available through training courses, online videos, and from the automation integrator or supplier, he added.
“They do a good job of showing you the basics and giving you a hotline to call when you need someone to walk you through how to use their products,” said Lizotte. “The intention is to set you up to succeed.”
In addition, many welding robots have software that includes preprogrammed welding procedures that allow the operator to tune weld and motion parameters, enabling her to scale up quickly and start increasing production, Novarc’s Karimzadeh said. Novarc’s Spool Welding Robot (SWR) is designed specifically for pipe, small pressure vessel, and other types of roll welding.
The decrease in time to weld pipes using automation technology from Novarc is a “game changer,” proclaimed David Ray, fabrication shop foreman at Jacksonville, Fla.-based W. W. Gay Mechanical Contractor Inc.
“Our old way of welding stainless steel would be to TIG weld with an inside diameter purge,” Ray said. “A 16” [406.4-mm], schedule 10 weld would take two-and-a-half hours or so to get a purge set up and the weld out. We now can do a 16”, schedule 10 weld in 12 minutes that will pass radiographic testing.”
With that production time increase, shops can offer improved, more predictable delivery times, be more competitive, and outbid competitors vying for the same work.
Software that controls the robot can also feature computer vision-supported artificial intelligence (A.I.) to generate an automatic weld path. A.I. offers potentially “faster programming of new parts using a variety of technologies—usually some sort of ‘vision’ to see a part, a computer and algorithm programmed to identified welds and generate path, and the cell itself with robots to accept that path and weld the part,” Leath explained. “All of these can appear to be expensive, but any of them can quickly prove their investment.”
A case study of Vectis customer Specialty Rim Supply (SRS), a Terre Haute, Ind.-based provider of precision, spun-forged aluminum wheel rims, shows the value of automated weld-path generation. Some of the custom rims SRS produces have a welded flange assembly to meet a specific, unnamed market need.
“Our batch size can often be as small as four rims,” noted SRS co-owner Rich Cuvelier. “The flange’s axial location will vary based on customer needs, so I needed an automation solution that could handle that product spread. It just wouldn’t be economical for us to hand-program every batch.
In this case, Vectis leveraged its force-based Touch Sensing software. The application allows the robot to gather tactical feedback by searching a surface and then offset a weld path based on the results. An added custom touch probe allows the robot to reach and reliably search various conditions of the flange assembly location at SRS.
Novarc also has adaptive weld control technology with its flagship SWR robotic welding system.
With Novarc’s vision-based weld control software, NovEye, A.I. is used to learn the weld conditions and develop control strategies to fully automate the pipe welding process. The need for human interaction with the cobot is eliminated with the software to help minimize human error and the associated costs for repairs and rework, Karimzadeh said.
Productivity like the Novarc customer Ray described is increased even more with offline programming, a feature most robot manufacturers offer, according to Lizotte.
“You could be anywhere else,” he said. “You know what the part is since you have the CAD model and you have the offline programming package for the robot. In theory, you’re driving the robot to a point, saving it, and moving on. It’s just like if you’re holding the pendant in your hand, but you can do it from a laptop.
In Lizotte’s scenario, an operator could be welding while someone else is writing new programs for other parts. The program writer uploads the algorithms into the automated weld cell, then the operator can make a dry run through the program. Tregaskiss’ Lizotte recommends such testing for all new programs, to ensure all the points are good or make fixes as needed.
“It saves a ton of time (because) you’re not stopping production while driving the robot around and figuring out how I’m going to weld this thing,” he said.
“Someone’s done all the offline programming and you’re just stepping through, making sure that it’s pointing right in the joint, and moving on to the next point.”
Yaskawa’s Leath would agree that a skilled welder would be the ideal person to champion a new robotic welding cell. But, he added, no robotics experience is necessary.
“(If a shop owner) can’t find labor or is having a hard time retaining labor, we’d focus on an easy-to-use system maybe with tools like cobots, offline programming software, and teach pendant applications,” he noted. “These tools can make it easy to learn robotics and increase flexibility for applications that may need more adjustments for different parts over time and increase job satisfaction.”
Once a shop owner is ready to add a robot welder to the floor, he or she needs to think about infrastructure requirements, personnel and their safety, and return on investment.
A robotic welding station or cell can be configured to a shop’s needs, within space requirements. If there already is a manual welding operation in-house, no additional space may be needed for a cobot system.
“If it’s a cobot going into a space a human would normally work, it will barely take up more room than the human did,” Leath said. “It also depends on the size of the parts, how the parts are reaching the cell; coming off an overhead or belt conveyor, versus a forklift or crane. We make cells for parts as large as shipping containers and full tractor trailers, and cells as small as a refrigerator for smaller parts.”
The footprint for Novarc’s SWR system is 4’ x 4’ (1.22 x 1.22 m). Vectis Automation’s robotic arm sits atop a rolling cart that measures 6’ x 3’ (1.83 x 0.91 m). Tregaskiss’ welding guns for robotic applications need to be configured with a robotic arm—while everything the company offers is for automated operations, it doesn’t make cells or systems. Yaskawa’s ArcWorld HC is cobot-mounted on a 4’ x 8’ (1.22 x 2.44 m) perforated welding table: The entire top can be used as a welding work surface.
Lizotte said shops also have to consider their workflow as parts come in that need to be welded and then be transported somewhere else once the job is done.
“If you automate the weld process, but now you’re running all over to find parts to load and then finding a place to put the finished parts, you’re not saving yourself a whole lot of time,” he said.
The shop needs to account for storage of welding consumables, too.
“You’ve got things as simple as wire. Typically, in the shops, you’re going to see the smaller spools of wire that are either in the wire feeder or in the welder itself, Lizotte pointed out. “And in the automation space, there are even bigger drums, 1,500-2,000-lb (0.75-1 short ton) package sizes, of wire.
So those take up a lot of space, too. Just make sure you have a place for everything.”
While robotic welding can help offset worker shortages by taking up the slack in production, automating the process doesn’t completely eliminate the need for personnel, especially those with domain knowledge. Robotic or not, the cell or system needs workers for machine tending, quality checks, and repairs. It may even help with the shortage in another way—as a recruitment and retention tool for workers who see robotic operations as cutting-edge and the type of job that would not only interest but also challenge them, Pawley said.
He also stressed the need to keep workers safe. Operators should wear full personal protective equipment, and other workers should be shielded from welding operations with a curtain or other forms of protection.
“It’s possible that work can be awarded to fabrication shops that have a clean record and adhere to strict safety standards, but more importantly, shops can be at risk of violating health and safety standards (if they fail to protect workers),” Karimzadeh cautioned.
Perceptions about what constitutes a return on investment are changing.
“For a lot of folks, a lot of the ways that ROI has been calculated for automation for decades is, I make 100 parts a day manually,” said Pawley. “If I get a cobot or a robot or some piece of automation, maybe I can do 200, 300, 400 or 500 a day, right? Or do 100 in a fifth of the time. So that’s the weld-labor savings.”
Traditionally, shops add up the labor savings realized in a year and multiply the figure by the hourly “burden rate” to get an estimated payback period.
“But what we see now is there are way more facets to the true value of automation to business beyond just the weld labor savings,” Pawley continued.
One of them is reducing post-weld operations. Others include reducing or eliminating dangerous operations that can injure workers, improving business continuity due to reduced lead times and on-time production, and cutting employee recruitment and retention costs, Pawley said.
“I think those are multiple facets to ROI now that are being considered,” he added.
Whichever aspects are considered in the formula for ROI, Tregaskiss’ Lizotte said it’s necessary to factor in the cost of consumables. With robotic welding, a shop that sees an uptick in production as a result may buy larger spools of wire and use more gas, both of which bring down the per-unit cost. However, because the robot doesn’t know how to compensate for the extra space from a contact tip’s wear, the shop may also see the need for new tips increase. The slightly worn tips, however, can still be used for manual welding because human operators will be able to adjust their movements to account for the wear on the tip.
ROI influences the decision on which automation to implement, Leath said.
“If increasing profitability is their ‘why’ (for adopting automation), with a goal of fast ROI, then we would point them to looking at our pre-engineered systems,” he said. “It might not be the solution to hyper-maximize their throughput, but it still increases it while quickly paying for itself, allowing the business owner to become more competitive in their marketplace while maintaining control of production.”
Leath generally sets customers’ expectation for ROI at an 18-month payback. But, he noted: “On the short side, we’ve had more than one customer recoup their entire cost in less than a month for extremely long arc-on times on parts they previously had a lot of scrap and defects on.”
1: One robot is the equivalent of 2.5 full-time employees.
70-plus: The cost of labor accounts for 70-85 percent of semi-automatic welding.
1 x 1 x 4: One operator can be expected to oversee one cell, but the cell can have as many as four robots. If arc-on time is longer, it’s possible to have one operator for two to three cells.
15: The life of an industrial robotic welding cell is >15 years.
18: The ROI for a robotic welder averages about 18 months depending on the robot, its application, and other factors.
Flanged Spigot Pipe $50,000: The average salary for a welder in the United States is about $50,000. But, in some states, experienced pipe welders can demand up to $125/hour.