The Hurdles and Horizons of Industry 4.0
Industry 4.0 has given birth to operational efficiencies in the present generation of manufacturing that no predecessor of production could have imagined. The technologies behind the Internet of Things, which evolved into I4.0, now enable asset utilization, cost reduction, and superior productivity for the factories that have embraced and invested in necessary systems.
Yet for all its good progress and promise, Industry 4.0 has hit a wall; well, a speed bump, and the problem isn’t about adoption, which some might readily assume. The future of manufacturing is still comprised of automation, data, and connectivity, and individual resistance is going to prove irrelevant to the industry’s next destination because the drivers of I4.0 advancement are too strong. However, the ETA of full I4.0 potential has been temporarily delayed.
Apparently, the manufacturing sector has fallen prey to shiny object syndrome, according to Will Sobel––a standards cultivator, entrepreneur, and software architect––and won’t get back on the road to true I4.0 fulfillment till it collectively stops the pursuit of buzzwords and sexy-sounding technologies that don’t deliver.
The so-called digital thread is frayed, and industry tends to chase after all things new and exciting to solve the problem of tying disparate information together, Sobel believes. Whether it’s machine learning or digital twin, these concepts can’t solve the gap that’s holding manufacturing back. The solutions, he says, lie in standards to bridge the information silos—as well as rethinking what to do with information. Currently, manufacturing can’t connect its data systems and thereby derive as much value as it could.
What’s Coming and What’s In It For You?
An important concept to consider is how machines will assist in advancing manufacturing toward full data connectivity, which will facilitate reasoning and decision-making around geometries––aka, the parts and products factories make.
In the not-too-distant future, the evolution of machine technology will lead to self-aware systems, machines that can self-diagnose, that know what they’re capable of and that can “talk” to each other. At its outermost imagined level, these future electronic partners in the workforce will comprise what Sobel calls dynamic manufacturing systems that, in theory, can bid on work and coordinate job schedules among themselves.
“You can then get into agile, dynamic flows,” he said. “And if you have these dynamic manufacturing systems … it’s more similar to human interaction than orchestration, which is just pure top down. The machines negotiate and discuss tasks related to what needs to be done and find the optimal path through these systems to achieve an outcome.”
Machines, he said, will bring new reasoning about getting from the beginning state to the end state and determining the best set of intermediary processes to achieve the designed end product.
“It’s going to be a slow progression to get to this level of autonomous systems that can actually reason on that level of complexity,” Sobel said.
The Carrot and the Stick
The vision for agility of workflow coupled with the A to Z process traceability will continue to drive manufacturing forward, and standards such as those Sobel develops at MTConnect Institute are serving as the stair-steps. He calls these future outcomes and competitive advantages the “carrot” behind all the work it’ll take to get there. But the “stick” that will advance the next phases of I4.0 will come from regulatory/end-user origins.
“From a supply chain perspective, OEMS are asking, ‘Did everything I originally required happen?’” said Sobel. “And that goes all the way back to sourcing––was the material sourced correctly? Was the machining done correctly? Was inspection done correctly, and heat treat, and other processes?
“So, there’s a regulatory process where they would like to get more introspection in exactly what happened. Because the flip side of it is, when something goes wrong, the question becomes, ‘Why did something go wrong?’ If a part fails, why did it fail? Right now, suppliers such as foundries can’t answer the questions through data. It’s a very manual process, and there just isn’t the sufficient information.”
But that doesn’t mean foundries won’t still be on the hook to measure up to OEM standards, which is why Sobel issues a serious call to action for all in the manufacturing supply chain: Don’t wait.
“You need to start instrumenting the systems you have right now; you need to start doing the data collection, you’ve got to build a data collection infrastructure,” he advised. “There’s value that can be achieved from utilization data, and availability is a valuable metric that you can use and drive pretty decent ROI from Day 1. It’s not going to do all the fancy stuff I was talking about, but it’s going to get you something.
“This is an incremental process,” he added. “You’re not going to achieve everything in a day. The problem is, once this stuff hits, and once we start going to the next level, if you’re not even at the first level of execution, if you haven’t even taken that first step, there’s going to be a pretty big learning curve to get from zero to one while everybody else is going for level six or seven. We’re trying to help enable it through some tools, but each step is going to be a jump, and you’re going to need to implement more and more technologies to be able to take each one of these steps.
“Like we’ve seen with CNC and other scenarios, shops that embrace technology move forward quickly, and the other ones are left behind.”
As Sobel asserted, the end game of I4.0 is full traceability of a part throughout its entire lifecycle—the ability to discern causation of every flaw, however small or hidden, from every single process in the production chain, and this, once achieved, will predictively erase both process and product error. His vision is that I4.0 will empower manufacturers and their suppliers, including metalcasters, to understand, with an agile workflow, whether a piece of equipment did what it was supposed to do and evaluate if the outcome of a finished product met the designed intention of expectation.
“The real blocker, and this is an area that I’m focused on in my research right now, is to understand from the design to process engineering, simulation, and execution, how all of this data is related together,” Sobel said.
“That’s been one of the main reasons why we haven’t achieved a lot of the promise of Industry 4.0––it’s because we still have the siloed mentality of how we treat information across the entire enterprise going all the way down the supply chain. If you don’t use standards, you’re building more silos. You’re just building more ways where somebody says, ‘Here’s my proprietary data set, and here’s the way I talk about this,’ while everyone else is saying the same thing. You end up spending all your time connecting dots.”
Today, factories are harnessing good data with sensors and increasingly tying their data to dashboards and ERPs for operational understanding and decision-making, yet this progress falls short, according to Sobel. Siloed information is like cracks in a sidewalk: Take a few steps back to assess where something went wrong and you stumble. “We can’t determine where the anomaly was introduced because the machines don’t know about geometry.”
Taking Shape With Geometry
Geometry, you might say, is the central sun around which orbs of isolated data revolve––or should revolve. Shapes of one sort or another are the whole point of manufacturing in the first place yet this tends to get lost in discussions about I4.0. But geometry is actually the central axis upon which to hang everything IIoT has acquired.
“Geometry is the only thing that’s really in common with all these phases of the process,” Sobel said. “We have some shape we’re making from various materials and we need to get it to a geometry.
“We’re manufacturing something physical that has a shape in space, right? The problem is that we deal with IoT data and sensor data, but what we’re really trying to say is, how does this particular sensor and IoT data relate to the shapes I was making? I have to understand what happened while I was making that casting––just the sensor data by itself is pretty meaningless. If you look at data coming off a machine, and you give it absolutely no context, then it’s going to be just basically noise.”
What’s happened, consequently, he added, is companies have created great “data swamps,” loads of unstructured data from which they hope to find meaning. Sobel’s research is focusing on geometry because, as he states, it’s common to almost all manufacturing processes––you’re always dealing with geometry, he said, so all data should be tied into what’s happening to that shape at every juncture, be it material removal or addition, casting, forging … everything.
“Whatever you’ve done with ingredients, properties, thermals, there’s going to be some geometry coming out––how is the process manifested in that geometry? And how is that tied to the original design? There’s a lot of assumptions going on about how this is going to behave, and it goes to simulation, we do finite element analysis, and we then we make the part.
“In all these different phases, you’re always talking about something physical, something that happens. And I think that’s one thing that really fell apart in our Industry 4.0 adventure.
Bottom line, Sobel asserts that by associating all data from all processes to the end product and how it works in the field, data takes on purpose and can be harmonized around a central priority.
“It allows us to be able to reason on the data because it acts as a common identity for various aspects of this thing that we’re making. It allows us to connect the data together across all the lifecycle phases.” •