ARC’s Andy Chatha Sheds Some Light on IT/OT/ET Convergence

Can you shed some light on what IT/OT/ET convergence is and how today’s industrial organizations are dealing with it?

For the last several years, we have been talking about how more and more IT (information technology) is making its way into plants. Take analytics, for example. Many companies today use analytics in their maintenance operations for predictive maintenance and so forth. Now we also have cloud; every company is trying to move into the Cloud. It's all IT. Ten years ago, pretty much all the exhibitors at this ARC Industry Forum would have been OT [operational technology] companies.  Now, we have more IT companies at this conference than OT companies. The reason being that more and more IT is making its way into the plants.

So now the challenge for OT people is to make sure all that IT is safe and secure to use in their plants. At this year’s Forum, we had a unique joint keynote presentation from two Dow executives.  Dow’s vice president of manufacturing, Peter Holicki, and CIO, Melanie Kalmar, talked about how they've been collaborating for the past 10 years to get their respective OT and IT organizations to work together better in the company’s plants.  Peter and Melanie talked about how they built a digital operations center in which they can try out new information technologies to make sure that they are robust and secure before they deploy the IT into their plants.

Another area that’s also converging is engineering technology (ET). So now it's not only just IT and OT, but also ET that are converging, adding yet another dimension to this beneficial technological convergence.  At this year’s Forum, for example, we also have engineering-related companies, like AVEVA, Bentley Systems, and Hexagon participating.  I think this convergence of IT, OT, and ET can benefit all manufacturers and many other industrial organizations, but – of course - they must be fully aware of the potential cybersecurity-related challenges associated with this convergence.

Watch on YouTube

Many manufacturing executives appear to be confused by edge devices.  Can you explain what these are and how they can be used in the plants?

Yes, much of the confusion stems from the fact that edge devices, or “edgeware,” have become buzzwords.  It might be easier to just think of anything on-premise as a potential edge device.  So, these could range from the network routers, switches, and other edge devices that IT people are so familiar with, to all kinds of field devices and controllers that the people in our plants have been using for years.  The private data centers that many companies connect their plant data to is a higher level on-premise or cloud layer.  And on top of that, we have the highly scalable cloud platforms like AWS and Microsoft Azure.

So, you see, we actually have three layers for data and functionality to reside in: edge devices in the plant, private corporate data centers, and then the Cloud. And one of the biggest challenges that companies face right now is deciding how to balance the load. What should they put in their field devices or control rooms?  What should they put into their data centers?  What should they put in the Cloud?

Take an MES (manufacturing execution system) solution, for example.  When you move MES-related data and functionality to your data center or the Cloud, it becomes easier to share it across multiple plants.  But at the same time, some of that functionality needs to reside on-premise, because if the connection with the cloud is broken you still need to run the plant and make product.  So, you need to strike a balance between what you put on-premise in the plant (so your plants can run smoothly) and what you put in your private data center or the Cloud to be able to share this functionality across multiple plants and learn and benchmark across plants, which many companies can really benefit from.    Most of the application software suppliers are trying to work with their customers to decide what functions they should put on-premise, what they should move to the cloud, and so forth.

Artificial intelligence and machine learning seem to be everywhere as well. What are the major applications for manufacturing?

Well, many companies have been certainly using some of these algorithms for a long time.  If you talk to someone from a chemical or oil company, for example, they’re likely to tell you they've been using algorithms to optimize their operations for a long time. But, clearly, now we have more powerful and easy-to-use analytics software.  So, you don’t necessarily have to be a data scientist to gain value from it. Right now, there are literally hundreds of analytics companies out there. And so far, I would say the most successful application in the plants of analytics has been for predictive maintenance.  This enables companies to predict the failure of different machines and, ideally, fix it before the machine breaks. It's a popular application because it can bring ROI right away. If you can eliminate a plant failure or a machine failure, you can save a lot of money.

Right now, in addition to using AI for predictive analytics, more and more companies are starting to explore machine learning.  This enables you to start gaining new insights into your equipment and your plant based on collecting, cleansing, contextualizing, and then finally analyzing actual machine and other operational data.  This can go a long way toward helping you improve the performance of the machine going forward.   We’re seeing analytics algorithms being embedded into all kinds of machinery, from welding machines to robots to drones. Analytics is becoming pervasive, because you can derive a lot of benefits from it.

Digital twins are another hot topic discussed at the 2020 ARC Industry Forum. How can manufacturing companies benefit from these?

There’s so much data and information about just about everything out on the web these days.  But while many manufacturing companies have been developing, using, and benefitting from process models for a long time now, it’s really hard to develop digital twins of all the equipment and processes in their plants. You literally have dozens or even hundreds of different machinery types and literally thousands of pieces of machinery in many plants. 

If you start with the as-designed product model for these products and couple that with operational data that shows how the machine is performing over a period of time, you keep accumulating more and more data.  This can provide the basis for a powerful digital twin of that product.   Once you have that, then you can start analyzing the performance and start looking at how you can improve the performance of that particular machinery.  Companies are starting to see that there's a lot of value in it.  Digital twins can definitely help eliminate product failures, but they can also help companies start to improve the performance of their machinery, processes, and plants. So, I think we are going to see more and more digital twins over the coming years for all different kinds of products.