How IIoT, AI can help make future-proof manufacturing a reality
The Industrial Internet of Things (IIoT) offers manufacturers greater productivity and better information insights through artificial intelligence (AI) as companies look toward the factory of the future.
The Industrial Internet of Things (IIoT) sets the foundation for the factory of the future – a smart, connected factory ready to adapt to anything industry can throw at it. IIoT requires a new way of thinking, new tech investments, collaboration between information technology (IT) and operational technology (OT), faith in data and trust in the artificial intelligence (AI) that will aggregate, analyze and then act on that data.
IIoT requires much, but promises more. The promise of greater efficiency and productivity, more agility and flexibility, increased overall equipment effectiveness and return on investment (ROI), enhanced product quality, reduced waste and lower costs. It starts with a connection.
The automotive OEMs have been plugged into Industry 4.0 for some time now. The rest of the manufacturing industry, especially small to midsize manufacturers that make up most of the manufacturing pool, will take longer to come online. But it will happen. The resulting insights will help factories make decisions faster and with more confidence than ever before.
From remote monitoring and predictive maintenance to predictive quality and virtual commissioning, there are steps every manufacturer can take now to future-proof your factory. Harness the right data, start small and think smart.
Connectivity and the right data
Technology leaders such as ABB Inc. are lighting the way for digital transformation.
“Connectivity and data acquisition are the key enablers of future manufacturing,” said Václav Švub, global head of the digital and innovations business line automotive, robotics and discrete automation business for ABB. “We see more and more companies that have a clearly defined digital journey. They want to be data-driven companies. Two or three years ago, everybody was struggling with the question, what does digitalization mean for me? And now, we are slowing moving from pilots and POCs toward real projects with proven benefits.”
Švub comes from the IT realm. He was head of ABB’s SAP team for ERP systems in Eastern Europe before he moved to the business side to helm initiatives in digitalization and factory of the future.
He said we still have a way to go before data-driven companies with adaptive manufacturing or autonomous production are commonplace. He sees AI in use, but on a smaller scale, in isolated islands. Predictive maintenance and quality prediction are here, but it’s still isolated and it’s not running on top of an entire production line or entire factory.
“Manufacturers have to understand that there is no one-size-fits-all solution we can universally apply to every company, Švub said. “You have to use and benefit from digitalization in a way that very much depends on where you are now and what you would like to achieve. Some are attempting to digitalize everything at once. But it’s better to split it into smaller steps, to tackle easier tasks. Start with the low-hanging fruit before moving to more complicated things or projects with larger scale.”
Patrick Matthews, automotive global assembly group manager for ABB, said sorting through all the data and extracting specific pieces that will lead to meaningful decisions can be a challenge. Yet, it’s imperative.
“You have to step back and determine what you will do with that data,” he said. “How do you analyze it? Who analyzes it? Can you make an algorithm or app to analyze it? Because at the end of the day, what our customers really want is to produce more parts with less non-added-value tasks performed by their manpower to reduce their cost. That’s the ultimate goal for them. Data collection is huge, but it’s about collecting the right data.”
Matthews makes the case for digging deeper into manufacturing processes to seize the right data.
“Sometimes you have to look at capturing data from places you weren’t thinking about. For example, we have a very complicated process in axle manufacturing we’ve been doing for years. When a customer has a problem, we can send a guy into the plant and he can just look at a few parts of the process and very quickly analyze what’s going on,” he said. “What we want to do is have the machine automatically correct itself. That’s the stage we’re moving into now.
“Now we can take it to another level by adding a camera to emulate how this guy looks at certain parts to solve the problem. There’s the data you need to purely run the machine, but there could be a subset or even a different level of data you need to take it to that next step. It’s key to collect the right data, understand the data, and then start to apply applications or apps that can analyze that data and create meaningful change for our customer.”
Digital application solutions for manufacturing
One of those digital application solutions is the Robotic 3D Quality Inspection (3DQI) system. This 3D metrology solution uses IIoT connected, sensor-equipped robots and deep learning algorithms for automatic quality control of manufactured parts on both inline and offline inspection cells. Using a 3D white-light optical sensor to scan millions of points per shot, the system rapidly records and compares highly detailed geometric and surface data against digital CAD models. Comprehensive data analysis is processed in real time, allowing for rapid feedback on production process variation. Digital records support traceability and enable users to adapt their processes to prevent failures and improve overall quality and productivity.
The 3D inspection system is part of the ABB Ability™ portfolio of IIoT solutions that leverage connectivity, data analytics and artificial intelligence to enable better decisions. During the pandemic, the platform’s remote-connectivity solutions have helped keep manufacturers’ operations humming.
RobotStudio, ABB’s simulation and offline programming software, can be used for virtual commissioning and running a digital twin of the production line in a virtual environment. RobotStudio is useful in all product life cycle stages, helping to introduce and program new products and variants offline without having to interrupt production.
“Every quarter, we are adding new functionality to RobotStudio,” Švub said. “Last year, we improved the ability to do virtual commissioning. We integrated more standard protocols so you’re able to hook up over OPC UA to your equipment, to your PLC, both the physical one and virtual one. You’re able to get the data back to RobotStudio for not only the robots, but also other equipment running through the OPC UA (a cross-platform, open-source standard for data exchange). That means you’re able to simulate, test and validate the complete behavior of the robotic cell or robotic line.
“Especially now, we see more companies pushing us to do virtual commissioning because of the pandemic and problems with travel. This is a great solution to be able to connect remotely rather than move our engineers around the world all the time. We see this trend growing and customers more open to running factory acceptance tests (FATs) even partially in these virtual environments because they start to trust in these technologies.”
That trust will be very important to future innovation. For Švub, to be ready for the future means to be digitally enabled.
“We will have data collection not only from the robots, but also from the processes. This will provide more value for the customer. For example, we’re working on a pilot with one of our customers collecting data from a robotic welding process and predicting the quality of weld seams. There will be data collection from the welding cell, from the robot, from the PLC, and from the welding gun or welding source. There could also be more data collection from monitoring the condition of the roll-up doors, which by the way, is quite tricky for our customers.
“Data collection first, then build value-add applications on top of it. We believe this is our role in the digitalization ecosystem, to help companies understand how to apply these digital technologies and processes.”
Sensors with smarts
Demonstrating how to apply IIoT technologies is a passion for one self-proclaimed automation evangelist. He literally can’t stop talking about its benefits, even from 30,000 feet up.
For Will Healy III, marketing manager-Americas at Balluff, digitalization can’t come soon enough. “I’ve been giving presentations since 2012 and they used to be called, ‘You need to put more things on Ethernet.’ It was basically a call to action on IIoT. There just wasn’t a name for it yet.”
Balluff, a fourth-generation family-owned company founded in 1920s Germany, provides sensors, RFID devices, software, vision cameras and industrial networking products. Healy draws an interesting comparison.
“Using the human body as an analogy, Balluff is the senses and the nervous system,” Healy said. “We’re not the brain or the outputs of the muscles. We’re helping collect data from the machines. We help you get the data you need to do analytics.”
The network is the nervous system of IIoT technologies. Nothing happens without that connectivity. For small to midsize companies, digital transformation can be daunting. This is especially true with capital constraints and a shrinking talent pool across all of industry. Healy offers some advice.
“If I was a small or midsize company, I would start by looking at local integrators and automation distributors that can help me implement IIoT with technologies I already have installed. Or reach out to suppliers of the equipment you already own. They may have packages available. I’m amazed by the number of machines that have IIoT capabilities that just need to be turned on.”
He said it helps to partner with someone who really knows the technology. They can help you advance your business on a scale you could never do alone.
“I’ve seen some really cool companies coming out that offer automation-as-a-service and robots-as-a-service. This allows people to use operational funds to get automation into their plant and then justify the automation over time. Then they can afford more automation. That’s a real powerful tool.”
IO-Link for the future
With many intelligent, IIoT-enabled sensors on the market today, it doesn’t require a big investment or advanced technical knowledge to get started.
“For small to midsize companies, the place to start is condition monitoring. It’s so approachable and understandable,” Healy said. “Condition monitoring is taking a part of the machine and monitoring it for different states.
“Vibration is a really powerful pre-indicator of failure. Temperature as well. We have a bolt-on condition monitoring sensor. It monitors vibration in three axes, X, Y and Z, and it monitors temperature and humidity.
Think about what equipment in your plant is so vital that if it fails, it would shut your production down. What’s the lynchpin in your factory that has long spare parts lead times? What are the parts of the machine that are most likely to fail? What parts, if they fail, would be catastrophic? How important is that machine to the company’s success?
“If you have one stamping press and you’re a metal fabricator, and everything you do is based upon that stamping press, if it breaks, no one can do any work. Maybe it’s a laser cutter or a waterjet. Condition monitoring is about reducing those unplanned downtime moments.”
Healy cited an example of an automotive structural Tier 1 supplier with a robotic welding operation. Where you have welding, you have exhaust fumes, and those fumes need to be removed.
“They have a central fume scrubbing fan. Inside those fumes are tiny little particulates of metal from the welding process, which can build up on the fan blades over time,” he said. “This causes the fan to go off balance and then they must do maintenance on the fan. This can be a big problem and create a lot of downtime. It’s also hazardous to go inside that fan and clean off the blades.
“If the fan is broken down, they can’t do any welding. Downtime in the automotive industry costs tens of thousands of dollars a minute. They have multiple sensors on that fan. One on the fan, two on the bearings and one on the motor. They’re not messing around. They want a very detailed profile of all the vibrations in that system.”
The most basic condition monitoring system checks whether a condition variable is within threshold, and creates an alert on a PLC, or sends a text message or an email, if it exceeds that threshold.
“You can then take that data and do predictive maintenance or other kinds of analytics and AI-type projects with that data,” Healy said. “If you already have automation in place, the threshold to add things like condition monitoring is very low. If you don’t have the talent, then you have to start looking at things like ‘black boxes’ or cloud solutions from third parties.”
More on those “black boxes” in a minute. First, Healy has some more advice for small to midsize manufacturers looking to future-proof their operations. When selecting automation components, he suggests you choose wisely.
Don’t go with the “most dirt cheap, simplest thing” you can get. Consider the total cost of ownership. What do you want this device or system to do 5 or 10 years from now? Buy a little more than you need in the moment. Set yourself up for success in the future.
“When you’re selecting devices, especially sensors, if you pick smart sensors, you get not just the on and offs, but also more diagnostics about the function of the machine, about the quality of your process, about what’s going on in your process,” says Healy. “You get a lot more information about your process by selecting smart sensors and using open standards like IO-Link.”
Universal, smart, easy and IIoT-ready, that’s IO-Link, an industrial communications networking standard for connecting digital sensors and actuators.
From smart pneumatic valves to grippers, over 300 suppliers (many of them A3 members) offer smart technologies that speak IO-Link. Healy stresses the importance of interoperability and standards like IO-Link that make it easier for manufacturers and machine builders to implement new, future-proof technologies.
Data analytics for actionable insights
If Balluff represents the sensory input and nervous system, where’s the brain to crunch all the data and make sense of it? Healy suggests we check out companies like MachineMetrics.
“To collect data from your machines, you can buy a little black box and bolt it onto your machine. Then all the AI, all the aggregation, and any analytics can be processed and provide reports that help you make better decisions.”
In this case, that “black box” is actually green, and it’s an edge device, just one part of an industrial data platform with a lot of brains behind it. MachineMetrics has grown quickly since its inception in 2015. The Series B startup got a $20 million boost in 2021 led by automated test equipment and industrial automation powerhouse Teradyne, the parent company to A3 Platinum suppliers Universal Robots and Mobile Industrial Robots (MiR).
“We’ve made it really easy for manufacturers to capture data from their equipment with an edge device that’s connected to the factory floor network or to the machine itself,” said Bill Bither, CEO and cofounder of Massachusetts-based MachineMetrics. “We’ll connect to sensors, like Balluff sensors, or we’ll pull data right from the machine’s control. Then we’ll automatically transform that into a common data structure, so that all your machines, even though you may have many different types of machines, basically speak the same language. From there, we provide tools to analyze the data, to generate insights from it. For more of the simpler insights, like understanding the capacity of your machines and throughput, that’s all out of the box within the MachineMetrics industrial data platform.
“But we also allow our customers to enrich the data themselves. That’s where you can be very configurable based on the types of manufacturing that you’re doing. We can provide a whole workflow automation tool that allows you to build out workflows to automate processes around the machine. You can also stream that data into other factory systems, like your production systems and maintenance systems.”
The startup has an entire data science team figuring out how to capture the right data and display actionable insights like cutting tool wear, for example.
“On a metal cutting machine, those tools often wear out. They will start to create scrapped parts, or they’ll break. We’ve been able to enrich that data so our customers can understand when they need to change their tools. Once the data is contextualized, they can use our platform to run custom analytics.”
MachineMetrics cut its chops in the computer numerical control (CNC) sector, but has since expanded to all discrete manufacturing. Industries served include automotive, medical device, heavy industry and aerospace. The industrial data platform is available as software-as-a-service (SaaS) for an annual subscription. That includes all support, access to the customer success team and software upgrades. The power of IIoT means anywhere, anytime access.
“It’s cloud-based, so you can access that data anywhere, as long as you’re authenticated,” Bither said. “The high-frequency calculations are done on the edge, but then that data is sent to a cloud platform where you can bring in context from other places.
“You can run reports to look at what happened over the past three years. That processing is stored in the cloud. It’s built on AWS (Amazon Web Services). Only our customers have access to that data, in a secure way, and they also control what data is sent to the cloud.”
Bither says it’s plug and play: “We have out-of-the-box applications that help our customers drive value immediately. For example, as soon as they connect their machine, they can configure a workflow that will send a notification if that machine is down for more than a certain amount of time or experiences a certain alarm condition. All that is self-service.”
Where it really gets interesting is in the advanced analytics. Remember all those data scientists?
“Advanced algorithms allow us to go a little deeper than what you can see, like trying to emulate the experienced operator who can hear when something is going wrong with a machine. You can emulate that in an algorithm, which is essentially AI, and then deploy that at scale. That’s what we’re doing by making the data accessible and then providing the tools to emulate the human in those cases,” Bither said.
The company conducted its own benchmarking study in 2020, with opt-in by MachineMetrics customers to aggregate the utilization data for all the machines they have connected. The results showed a surprising average machine utilization rate of only 24%. Bither says this was much lower than what customers expected.
“Without visibility, you’re unable to make positive change. With MachineMetrics, we provide real-time visibility into everything that’s happening with your equipment. With that, you can continue to improve your processes and eventually move to more automation. That’s really our mission. What we’re building is that operating system for the autonomous factory.”
Autonomous production in sight
Real-time adaptive control still has a way to go, but advanced technologies like these could enable it.
“As soon as you try to move to autonomous production, you have to trust in the technology,” Švub said. “That means machines will adjust their behavior by themselves, without any manual inputs.”
That trust will be very important as we take the next step toward the factory of the future.
Tanya M. Anandan is contributing editor for the Association for Advancing Automation (A3), a CFE Media content partner. This article originally appeared on the A3 website. Edited by Chris Vavra, web content manager, Control Engineering, CFE Media and Technology, email@example.com.
Original content can be found at Control Engineering.