Master welders teach robots their skills

The skills gap for welders is widening as the workforce gets older and retires. To address this problem, companies are turning to their master welders to teach robots how to weld because robots are more efficient than humans.

By Tanya M. Anandan, RIA October 5, 2017

In 2010, the American Welding Society (AWS) reported a shortage of 200,000 welders. Many senior welders are retiring or leaving the industry for other pursuits and they are taking that welding expertise with them. The skills gap will continue to widen because there are fewer welders. As the gap widens, the AWS estimates a projected shortage of 372,000 welders by 2026.

What’s not clear is whether this is a true shortage. Will we really miss all those welders, especially as more welding is automated? Will other technologies replace welding? The answers to those questions remain unclear right now.

What is clear is that the mastery of the process will be missed. Without master welders, how will the next generation of welders fare and how will robots improve? Experts will tell you that, fortunately, the best welders can be taught to be the best programmers of welding robots. The precision, consistency, and control demonstrated by master welders are the same qualities that are needed from welding robots.

Zane Michael, CWI/CWE, is director of thermal business development for robot manufacturer Yaskawa Motoman. Michael has nearly 40 years of experience in welding and is an AWS Certified Welding Inspector (CWI) and a Certified Welding Educator (CWE). Michael began his career in 1979 teaching at Hobart Institute of Welding Technology in Troy, Ohio.

Michael often travels to his customers’ facilities in the construction, agriculture, and mining equipment industries and he sees firsthand the challenges manufacturers contend with on a daily basis.

"When I visit customers, I ask them why they’re interested in robots. What problem were you trying to solve? Nine times out of 10 they tell me, I can’t get qualified welders, or I can’t keep welders," he said. This is the sentiment across the nation. The solution to this dilemma is often coming from automation.

The best welders make the best robot programmers

Using flexible welding automation with robots is like cloning the welding knowledge and expertise of your best welders. Those in the know will say that it’s much easier to teach someone to program a robot than it is to teach someone to weld. Michael said that it requires nine months of full-time training to become a certified, qualified welder in multiple processes and multiple positions. By contrast, he said it’s much easier to teach an experienced welder how to program a robot than it is to teach a programmer the nuances of welding.

"In less than two weeks, I can have you programming a robot. But when I give you a weldment and say program this robot to put the welds on like the print shows, if you don’t understand the welding process, if you can’t do it by hand correctly, then you don’t have a good chance of being successful with the robot. Anybody’s robot is capable of holding a welding torch in a joint, turning the arc on, and making a weld. That’s easy," Michael said. "But understanding all the critical welding variables to produce a quality weld is not. The burden on the programmer is to have a very good understanding of that welding process." Michael currently teaches a class in welding processes at the University of Dayton.

"I tell these students that welding is like making an apple pie. You have a recipe you have to follow. You have so many apples, so much sugar, you bake it at a certain temperature, and you’re guaranteed that apple pie is going to taste the same every time. Welding is no different, except there are many more variables involved, like travel angles, amperage, and stick-out. All these critical variables have to line up to produce the expected quality outcome, which is called the Welding Procedure Specification (WPS)."

For gas metal arc welding (GMAW), or what’s commonly called MIG welding, those five critical variables are: 

  • Electrode size
  • Current
  • Arc length or voltage
  • Travel speed
  • Electrode angle.

"As they’re welding, manual welders will read that puddle," Michael said. "They can change their stick-out, as an example, to increase or decrease the welding current to help control the puddle and produce a quality weld."

Stick-out is the proper tip to work distance for the MIG welding process.

"These are all factors that a robot programmer has to know. I tell my customers it’s a recipe for disaster if you send me one of your mechanical or electrical engineers and expect them to be the programmer of your welding robot when this individual has never welded." 

Learning the welding process

Welding, whether manual or automated, is process-specific. Understanding that process is vital in order to clone it for a robot.

"Robots—I don’t care whose brand it is—are easy to program. It’s the process," Michael said. "Even when we automate non-welding jobs, like an operator sanding a casting or forging, where they are running that across the belt sander, we will study the operator’s motions and angles because we basically have to duplicate that process with the robot."

How the process is done is critical. Expertise that can only come from hands-on experience.

"Anytime that we start a new process, or approach a new part, a lot of the time it is something that we’re taking over from a manual process," said Brendan Brown, virtual solutions engineer at Genesis Systems Group, LLC. "Those are some of our best advocates. The guys who have been hand-welding these parts over the course of years. How they approach the sequencing of the welding, what they weld first, and the appropriate angle, and all the push and the travel speeds. That’s information we always want to gather. Who would know that part better than the guy who’s been welding it?"

Brown, an offline programmer with almost 20 years of robotic welding experience at Genesis, agrees with Michael about seeing the whole process.

"One thing we see a lot with first-time robot users is they have automated this process, but you have to look down the line to see how you’re making the part, how you’re tacking the part together. We often get customers that build their own tooling or outsource the fixture. They are not necessarily controlling the appropriate datums or securing the part properly. Even the best welder has to look at the entire process to get a good weld."

Genesis’ headquarters in Davenport has a training lab and two private classrooms where customers’ programmers and machine operators can gain hands-on experience and instruction. Brown helps teach the basic and advanced training classes.

"The majority of our students have some kind of welding background, whether it’s manually welding the parts they are getting ready to automate, or just the company’s weld engineer that will be overseeing the machine and will need to understand how to touch things up. When they come into the class, we say here’s the robot, your new tool," Brown said. "We don’t need to teach proper push and work angle."

The best welders understand the subtle nuances of the process.

"That’s why it is much easier and more successful to take an experienced welder and teach them the robot," Michael said. "When my customers ask, ‘Who do we send for training?’ I say, I want your best welder."

Efficiency doesn’t always mean faster

A common myth about robotic welding is that people think you’re going to weld much faster. That’s not true.

"A robot will give you more parts in the bucket at the end of the day, but it is not physically going to be welding faster," Michael said. "When a manual welder puts his helmet down and strikes the arc, let’s say his forward speed as the weld is being made is 20 inches a minute. The robot will basically weld at the same speed. That’s the recipe. Just like mom’s apple pie recipe. Bake for 450 degrees for 30 minutes. You wouldn’t bake your apple pie at 900 degrees for 15 minutes. To produce a quality weld, one of the essential variables is travel speed. You put too much heat into a part, you can create issues."

Welding robots are not necessarily faster. They’re more efficient.

"A manual welder is about 20% efficient," says Michael. "What that means is that out of the 8 hours you pay a manual welder, the arc is only on for about 20% of those 8 hours. The welder lifts his helmet up, repositions his chair, or cleans the weld off. There’s a lot of in-between, getting-comfortable time before I make the next weldment."

A robot, however, is about 85% efficient.

"The robot will get from one arc to the next weld and the next weld much faster. If it takes the welder 15 minutes to produce that part, the robot will do it in approximately less than 4 minutes.

"When I walk through a plant and I see three to four welders, I think that could be one robot," Michael said. "Not in the sense that one robot is going to take their jobs, but I can use those welders on more critical jobs that a robot can’t do." 

Adaptive control benefits

Onken Inc., an Illinois-based maker of bulk grease-collection systems and oil storage tanks, had relied for decades on a team of experienced welders to manufacture its leak-proof tanks. With the rising welder shortage, it was harder for them to find quality welders. This is where the robots came in.

With a two-zone workcell, an extended-reach robot, and some training for its welders, Onken was able to eliminate up to 20 hours of production time and complete the same amount of work with three welders instead of five. The two freed-up welders could then work on other projects, reducing the need for contract welders and further increasing cost savings.

The MIG welding cell created for Onken’s tank production was a custom-engineered workcell with two zones. This allows the operator to load and unload tanks in one zone, while the robot is welding tanks in the other zone. The welding robots have built-in adaptive control, so they’re able to adjust travel speed, stick-out distance, and torch and travel angles as needed. Advanced weld seam finding and tracking tools were used to accurately locate and track the various types of weld joints on these 120- to 500-gal tanks.

"Robots are very repeatable," Michael said. "You program a path from point A to point B and a robot will follow that path all day long, plus or minus five thousandths of an inch depending on the robot spec. To make a quality weld, the weld joint has to be positioned within roughly plus or minus one-half the diameter of the weld wire. If I’m using .045 diameter MIG wire, that’s a little over plus or minus 20 thousandths of an inch for the weld joint repeatability requirements." Welding often comes with distortion, which is a challenge that robots can be programmed to overcome.

"We’re heating up the metal, so it wants to move on us. In larger weldments (like Onken’s tanks), distortion is more predominant. If the weld joint repeatability is not satisfactory, we have to use sensors for the robot, so it can alter its program path to match the location of the welds for that part during welding."

Touch sensing is used to locate a large circular flange and several smaller threaded couplings on the top of the tank. Once each component is located, the programmed paths are automatically shifted to match the location of the weld joint for that one tank. The thru-arc seam tracking is designed to enable the robot to alter its path and keep the weld puddle in the seam of the weld joint during welding.

As the robot welds the joints, which include contoured rolled corners, a laser sensor provides real-time seam finding and tracking just ahead of the weld torch. This enables the robot to alter its path to compensate for any variations in part fit-up. Accurately locating and tracking the weld joint also eliminates the need for costly complex tooling.

"We recently invited a customer to visit our factory," said operations manager J.R. Onken. "He said he had 10 minutes to spare for a tour. Once he saw our robotic welding cell in operation, he studied it for 30 minutes. He grew up as a welder and couldn’t believe the quality we were getting from automation. Automating our welding process has ensured we’ll be around as a company for another two or three decades," Onken said. "It’s secured our future." 

Harnessing knowledge

It’s been said artificial intelligence (AI) will someday allow us to download our experiences, to digitize them for future generations, and clone our memories. Or at least our limitless brain power, in spite of our finite bodies. But that’s still a long way off.

For now, we need to capture and disseminate that knowledge or forever lose it. This is not theory or textbook smarts. This is hands-on experience accumulated over decades. Know-how that is difficult to demonstrate with words and diagrams. It varies from process to process.

Until AI reaches the pinnacle, it’s just us and the robots, master and tool. It would be a waste not to put this opportunity to good use.

Tanya M. Anandan is contributing editor for the Robotic Industries Association (RIA) and Robotics Online. RIA is a not-for-profit trade association dedicated to improving the regional, national, and global competitiveness of the North American manufacturing and service sectors through robotics and related automation. This article originally appeared on the RIA website. The RIA is a part of the Association for Advancing Automation (A3), a CFE Media content partner. Edited by Chris Vavra, production editor, Control Engineering, CFE Media,


Key Concepts 

  • Companies are turning to robots to learn welding from master welders to compensate for the growing skills gap.
  • Robots are more efficient than humans at welding and can improve a company’s productivity overall.
  • Robots can also be taught to compensate for potential issues such as distortion and other variations during the welding process.

Consider this

What other industries facing a skills gap shortage could benefit from robots learning the processes required to be efficient?