Still Waiting for the Robot Rapture

The popular press may lead you to believe that the Singularity is right around the corner. But a Fortaco welding engineer explains why robots used on big structures must dramatically improve before they’ll completely replace human welders.

by Jari Hakalahti, QHSE Manager & Welding Engineer

Manufacturing professionals who are not welding engineers — often those who have been justifiably amazed by functioning robot lines –sometimes talk about robotic welding as if it’s incredibly simple. Just throw the parts in the air, press a button, and voilà, your finished product is ready to ship!

At Fortaco, we enthusiastically use robots whenever they make real sense for our business, yet our behind-the-scenes vantage point forces us to think in sober terms. The robot revolution may one day arrive: robot welders which (who?) understand, learn, and adapt to changing conditions while they weld. But before this day comes, there are a host of issues which need to be sorted out.

Can you match this, bots?
At our factory in Kalajoki, Finland, our 40 welders produce around 7,000 tons of welded structures every year, among them 32-ton steel base plates on to which ship engines and generators weighing over 70 tons will be fastened. Given the millions of loading cycles and vibrations our welds will endure over the ship engine’s lifetime, there is not much room for error.

For a robot to match human quality, it will need to see air gaps and immediately adapt. It will have to be sensitive to environmental conditions and immediately compensate. And of course it will have to figure out how to get to those hard-to-reach places and corners where a bot isn’t currently up to the job, or can’t be repositioned without major human effort.
Another challenge is tack welding. Since robots can drive over tack welds and may cause defects underneath, humans are necessary to ensure even quality. Assuming a robot could tack weld, you’d also need another robot who could position the parts perfectly. Yes, this can be (and is) done for simple products and huge volumes, but it doesn’t make sense for Fortaco’s biggest structures over 15 meters long and weighing 32 tons.

A challenge to suppliers
The size of the structure presents other problems. If you want robots to weld them, you need perfectly-cut and pre-bent parts to avoid gaps and weld stress deformation in big structures.
Also, as any hobby welder knows, a large part of good welding is positioning the parts before you begin to weld. And even beyond the realm of robotization, we see that most new welding technologies demand very accurate parts and plate fitting. Parts manufacturers may wish to take this as a challenge, as the success of robots is partly in their hands.

It’s payback time
For robots to match human performance is not impossible, but it is expensive. Optical sensors, temperature sensors, cameras, sound sensors – all these are required to approximate the human welder. Not only are these items expensive, but they take up massive amounts of space.
Please don’t view me as a Luddite – I’m very much in favor of robotization. I love that robots don’t take coffee breaks and that they can turn a part in a second without the use of a crane. It’s just that I work with big structures and robots every day. I know their limitations.
In many Fortaco factories we have products that suit well to robot use, and we will no doubt continue to invest in robot welding in the future. [SD1] But since we’re also running a business, we have to be very careful about which robots we invite into our lives.
From an investment point of view, we cannot wait an eternity for a return on the investment. With robots, the investment is huge and the payback time is long. There’s always the risk of investing today in yesterday’s technology, and recouping only a fraction of your original investment. So before we invest, we need to be convinced the robot will make a meaningful contribution.

Sympathy for the bots
Perhaps it’s time we humans show some sympathy for robots. After all, sometimes it’s we humans who are holding them back. Robot manufacturers, interested in sales, often provide numbers that are too optimistic. For us, the best measure of efficiency is how much welding wire is used in one hour. A human welder uses approximately one-half to two kilograms per hour, depending on the welding process used. For the jobs robots can do, they use four to ten kilos per hour, depending on the set-up. It’s great efficiency, but the main issue (which is usually forgotten in comparisons) is that it’s the only part of the welding phase that can be automated — welding set-up and finalization must still be done manually. Therefore, the over-the-moon numbers some robot makers give you for the overall performance boost are not always accurate.

Also, robots can literally suffer from prejudice. Take welding around notches, for example. In some conditions, robots can do this work well, but some humans still oppose their use. Even when robots achieve the required state of development, humans may still make decisions about their use using outdated information. It takes a while for information about their proven track record to circulate. Poor robots. We’re lucky they don’t yet have emotions.

As professionals, the best thing we can do is to attempt to understand the real applications of robots, accept that they are not the universal fix-all in manufacturing, and not demand too much of them. From time to time, we might even offer a little bit of robot love.