The fabrication and manufacturing welding industry is facing an ominous challenge. The National Association of Manufacturers reports that 81 percent of manufacturers are unable to hire the number of welders they need to satisfy demand. And the American Welding Society predicts a shortage of 314,000 welders in the United States by next year. With the average age of a welder in the United States now 55 years, the shortage is expected to increase significantly over the next five to 10 years.
To address this challenge, the industry is rapidly implementing welding automation systems. The number of installed industrial welding robots now stands at 400,000 units and is growing rapidly, with an additional 45,000 units being shipped annually. Automotive and transportation markets account for six out of every 10 units, with large robotic companies such as ABB, Yaskawa, Fanuc and Kuka supplying 50 percent of those units. (Asia represents 70 percent of the total arc welding market.)
But welding robots only provide the hardware; they lack the welding logic and cognition for practically meeting the requirements in welding applications. The next step forward, therefore, is an autonomous and adaptive welding solution that uses artificial intelligence (AI) and computer vision: Welding 4.0. This wave of robotic welding can address the shortage of highly skilled welders and dramatically improve the quality of end products.
The market for welding cobots is also expanding. Around 1,500 welding cobots are currently installed globally, but with more than 550 units shipped last year, this emerging technology platform is enjoying an annual growth rate of more than 30 percent with Universal Robots accounting for half the market share.
Companies such as Novarc Technologies are currently delivering pipe welding automation systems that enhance performance and productivity and allow fabrication shops to target bigger projects with improved delivery timelines, increased capacity and greater profit margins.
The Novarc spool welding robot (SWR), which is a cobot welder, allows a less-experienced operator to consistently produce high-quality welds and improve shop productivity while reducing the operator’s exposure to the health and safety risks inherent in the welding process.
AI allows the cobot to optimize highmix, low-volume manufacturing applications. Time-consuming and, at times, difficult programming is eliminated. Improved weld quality and throughput result in significantly reduced repair costs. In the case of SWR, customers have historically seen fabrication shop repair rates (typically 3 to 5 percent) dip down to less than 1 percent. The automation technology produces consistent, high-quality welds that improve every single time by machine learning from ongoing welding dynamics.
While industrial welding robots and cobots are growing quickly, there is still a need for more autonomy and intelligence. The welding robot provides the hardware, but lacks the welding logic and cognition for practically meeting the requirements in welding applications.
Some of the challenges inherent in robotic welding are:
Fit-up variation: Robotic cells need to deal with fit-up variations such as inconsistent gap opening, prep angle, filling volume and misalignments that lead to rejected welds. It is important to minimize fit-up variation by using accurate cutting and fabrication techniques, carefully controlling the assembly process, and using appropriate fixturing and clamping to hold the components in place during welding.
Still, industrial robotic cells experience failures due to fit-up variation even after all these mitigation techniques. For example, the root pass still remains manual and relies on the dexterity of the welder to avoid the risk of blow-through with robots.
Tack detection and fusion: Tack welds are a common part of fabrication to ensure accurate positioning and prevent distortion, but they must be properly removed or fused into the final weld to ensure a strong and continuous joint. Tack welding is a manual process and has inevitable randomness. Therefore, welding robots and the autonomous system cannot be pre-programmed to properly detect and fuse the tacks. This problem turns into a major challenge for aluminum welding.
Non-straightforward seam tracking: Through-arc seam tracking and laser are two common technologies for weld seam tracking. The technology fails, however, for nonstraightforward seam tracking like high-curvature seams, thin weldments, non-groove joints, aluminum materials, no-weave applications, and finding start and endpoints.
Adaptive fill: Sizes, shapes and volumes of the weld joint opening are not constant in real production, especially in multi-phase welding scenarios on thick plates. A pre-defined welding recipe can easily lead to underfill or overfill and eventually weld rejection. There is always labor-intensive work before robotic welding to tailor a predefined recipe to each part.
The next stage
Novarc is leveraging advanced AI to deliver solutions to address current welding challenges like gap adaptation, tack detection and fusion, complicated seam tracking and adaptive fill for multi-pass, multi-layer welding.
Novarc’s NovEye welding vision system uses AI to process vast amounts of data to mimic what an experienced welder would do to accommodate the welding issues and variations described. The NovEye vision system delivers a solution that converts weld insights through deep neural algorithms to fully automate the pipe welding process, resulting in better adaptation and more predictability during the welding process.
The next stage of development is the “smartization” of the welding robot, transforming data into intelligent systems and offering the ability to provide an autonomous adaptive welding solution for welding variations. Many of these solutions are rooted in computer vision AI like NovEye.
With Novarc’s broad set of capabilities in AI, computer vision and robotics, the company is at the forefront of high-productivity, low-impact welding technology. SWR is opening new markets in a variety of industries, enhancing the performance of welding robots.
After all, when the expansion and maintenance of the world’s infrastructure is one of the demands of the welding industry, the future is a challenge. But AI will provide the industry with essential automation solutions and welding smartization is growing faster than society’s perception of progress. We are lucky to be able to embrace it as we join in the communities responsible for the future of welding.