Artificial intelligence boosts metal fabrication efficiency

I’ve gained a few airline miles over the past month. In March I flew to Stuttgart to attend the TRUMPF Intech event, visiting an area fab shop on the way; and in early April I flew to St. Louis for NASCC: The Steel Conference, hosted by the American Institute of Steel Construction.

In many respects the two events are at opposite ends of the metal fabrication world. Intech focuses on precision sheet metal fabrication; NASCC (formerly known as the North American Steel Construction Conference) focuses on commercial construction and the detailing, structural fabrication, and erection that goes with it. Both events, however, shared a common theme: The role of data in metal fabrication will become more important than ever.

Much of Intech focused on part tracking, machine monitoring, and machine learning. Connect automated machines to the cloud, and the machines learn how to perform better over time. Automation becomes autonomous. Automated denesting and material handling are prime examples. As machines lift dozens, thousands, and eventually millions of different parts out of a cut sheet, the system can learn how to grasp certain shapes in certain situations.

The challenge with machine learning is that it thrives as situations scale up. The more people use Google, the smarter it becomes. The more fabricators connect machines to a central server, the more effectively the machine can learn.

The challenge, of course, is that small businesses dominate metal fabrication. Few have dozens, let alone hundreds or thousands of a specific machine model or even machine type, be it a laser or press brake or anything else. Considering this, machine learning might be difficult to achieve in-house.

This opens the door for outside service providers. These services could come from machine tool providers, third party data analytics firms, or a combination of both. In the years to come, data analytics and machine learning services could become an integral part of metal fabrication.

And this won’t be limited to the precision sheet metal arena. At NASCC I visited Trimble’s booth and donned a helmet that showed the potential of augmented reality, overlaying a virtual building model over a real model of a building. A partnership between Trimble and Microsoft, this could be the next step in the evolution of building information modeling (BIM).

Although the technology around BIM has been around for decades, few construction projects have all parties—architects, engineers, detailers, fabricators, erectors, mechanical contractors, specialty contractors, and all the rest—working on a complete BIM platform. In fact, many NASCC sessions dealt with communication problems that remain, particularly between the engineer, detailer, fabricator, and erector. This includes basic stuff like weld symbols.

The skilled labor crisis presents a major hurdle too, of course. If workers don’t engage or even show up to work, it’s tough for any company to thrive.

Despite these challenges, it’s probably safe to say that machines, be it a fully automated sheet metal cutting system or a fully automated beam line, are no longer the constraint. So many bottlenecks arise from not having the right information at the right time. This is why data analytics and machine learning could be monumentally important in the years to come.