Host Intro (Gene Kim)

All right. The first session for today is yet another group of leaders who are pushing yet another part of the frontier of performance. This time they're going to talk about the large problems that result from what they call information technology and OT, or operational technology.

This is a problem that I've seen firsthand in so many areas of critical infrastructure, where the consequences of this divide are the source of so many issues for things that are critical parts of modern society.

Up next are Willem van Lammeren, who is Technical Lead Industrial IT for Solvay, and David Ariens, Manager for Analytics for Industry and author of The IT/OT Insider. They'll talk about what I believe is one of the most significant challenges in the manufacturing industry, and their experiences trying to bridge the worlds of OT and IT -- I'm sorry, IT and OT -- using principles that are familiar to all of us. Willem and David.

---

David Ariens

Amazing. Thank you, Gene. Thank you very much. Thank you for the introduction. We are really excited to be opening day two at the conference.

We are kicking off today by talking about industrial digital transformation. It's a domain we both live and breathe. I'm David, and my former colleague and good friend is Willem. The best way to describe what we have been doing for fifteen-odd years is that we digitize manufacturing and critical infrastructure. We basically do IT for factories.

We both worked at BASF, one of the biggest chemical companies in the world, with sites the size of a city and manufacturing plants as far as the eye can see. For some, those plants and factories might look traditional, but on the inside it is an industry that is constantly evolving.

We work in that twilight zone where digital meets operations. But bringing digital solutions to production workers, lab technicians, or maintenance staff is not really an easy task.

A couple of words about myself. At BASF, I was the industrial digitalization director of one of their major sites, the one in Antwerp, Belgium. Globally, BASF had more than 100,000 employees; on our site, just above 3,000. Besides that role, I was also leading the EMEA industrial cybersecurity program for BASF for quite some years.

Our job was to build and maintain products to support and optimize production processes, supply chain, laboratories, and so on. Two years ago, I switched to the dark side: I went from customer to supplier. I'm now working within the AVEVA ecosystem. AVEVA is one of the world's largest suppliers of industrial digital software and digital transformation software. I'm mainly working within critical infrastructure and food and beverage. With my company, Analytics for Industry, we help our customers become data-driven.

---

Willem van Lammeren

I'm Willem. It's a hard name. I'm Belgian, and as a good Belgian I started my career also in chocolate and beer. But as a chemical engineer, I also came to BASF, where I initially joined IT and then moved to Antwerp, where together with David we worked on digitalization in manufacturing.

Recently I moved to Solvay, which is a Belgian chemical company. Not as big as BASF, but we're still talking 5 billion turnover, 9,000 employees, 50 plants worldwide, so it's not a small player.

Since I wanted to continue working with David, we created The IT/OT Insider as an excuse to continue exchanging ideas about work after work, because we're nerds like that, and also as an excuse to connect with interesting people in this field outside of our official job descriptions.

---

David Ariens

It's a perfect excuse. Definitely.

IT and manufacturing have a lot in common, but at the same time they're really different. It starts with the lingo, the terminology we use. These are things we encounter quite a lot in our jobs.

When IT people talk about engineering a pipeline, you probably mean you want to get your code into production in the best possible way. In the manufacturing world, we also want to get something somewhere in the best possible way. But instead of DevOps engineers, we have pipeline engineers.

If we talk about working with containers, you probably have been doing container orchestration for the last ten years or so. We've been handling shipping containers since the 1950s.

Finally, there are differences that have a very big impact on how you and we think and act. Moving fast and failing fast, maybe breaking some things, is something we sometimes dare to do when we do IT projects. But in manufacturing, we are slightly less adventurous. We are a bit more conservative. Doing a rollback of a plant is not really that easy.

Why should you care about this niche field of digitalization in manufacturing? Haven't we fixed our problems by now? Isn't this old economy? We like to disagree, because while in IT things are mostly virtual, the real world we all live in is a physical one. Cars, houses, electricity, clean water, data centers -- all these things are being made.

Being best at making those things is hugely important: making them in the best possible way at the lowest possible cost, creating interesting jobs. These impacts also go outside an individual company. The only way to tackle global problems like climate, affordable housing, healthcare, doing less with more, and so on, is by actually creating those things.

Digitalization can help us with that. To design new materials, to build a new plant -- what we call a greenfield plant -- or even to optimize or rebuild existing plants, what we call a brownfield installation, we can make use of digital tools. The big problem is that a lifecycle of 20, 30, 40 years is not really an exception.

Here is a thought. In the '80s and '90s, I dare to state that manufacturing was really on the forefront of digitalization. Plants around the world evolved from being mostly manual -- analog meters, manual work -- to one of automation. Instead of a bunch of relays switching things on and off, we had fine-grained control, sensors with sentinels to control systems, PLCs for example. Multiple sensors and multiple PLCs would form a network, which we would call a SCADA system or a DCS, allowing us to control an entire plant from just one control room.

Although we kept on improving, the people in this very nice marketing picture seem to look really in sync with their data and their processes. But in reality, even in the biggest companies, paper and Microsoft Excel still rule the world, and the technologies we apply are extremely mature and therefore also rather old. We also see a lot of resistance from people when we try to change things, for example when we try to implement a machine learning algorithm, when we introduce cloud technologies, and so on.

We would like to start our talk by looking back: where are we? As a baseline, I've taken the numbers of overall U.S. labor productivity. On this graph you can see an overall steady growth in productivity, here in U.S. dollars per work hour. That's good, because that's the basis of economic growth.

If we take a look at the last fifteen years alone, and instead of taking overall productivity growth we focus on manufacturing alone, we actually see that the graph seems to stall. Productivity seems to stall. There are probably a magnitude of reasons, but as we are spending quite some money on digitalization, we have to ask ourselves the question: does it actually make sense that we are introducing all these new buzzwords? Does it make sense that we are introducing IoT technology and 5G and cloud computing and AI, et cetera? Are we actually delivering the promises of digitalization? And if not, why?

---

Willem van Lammeren

David, you and I have been working for fifteen years in this field, and it's a bit annoying to say, but we're stuck. We have a problem when it comes to scaling, to bringing those solutions onto the shop floor. Like you said, it's not because we're not trying. There have been loads of projects and initiatives. I'm sure billions have been spent on this.

But go to an average plant and you're transported back to the '80s. I know of people writing stuff from one computer in a notebook to go to some other device and type it in again. It's not really moving as fast as we would want.

What's holding us back? We've been going to several plants, talking to several people, and we see two big observations.

First, we have a technical scaling problem. Each plant is unique in its landscape, which, while optimal for its own local solution, really makes it hard to scale solutions. We're stuck in an internal loop of pilots. It usually goes like this: you have a promising idea, some funding and attention from management, your vendors are there, and you can do a pilot. Usually the pilot succeeds, because we're not doing crazy stuff. The next step is rollout, and that's when usually the problems start popping up.

You go from incompatible systems to makeshift solutions, like some complex network of Excel macros that are linked to some AS/400 print server under somebody's desk. I'm not making this up. You slow down. You need to adapt your solutions to the reality of the shop floors. With that slow progress, enthusiasm goes down and attention goes away. Usually by the time you're one third of the way, something else has caught management attention and your project gets shelved, and probably hangs around in your architecture for the coming years.

Those issues aside, we also have organizational challenges, and we dare say they are even more challenging. You probably know Dev and Ops and the divide between them. The divide between IT and OT engineering is even wider than that. Our common point of reporting is normally the CEO.

In companies that are not very mature when it comes to digitization in manufacturing, we see a loop that brings them even further apart. In the beginning, everybody is happy in their silo, each working in their own world. Then one side usually starts some Industry 4.0 initiative. It could be IT: they built ERPs, they have data lakes, they introduced cloud. How hard can it be to create a dashboard showing some IoT sensors on a global scale? It could also be OT engineering saying, hey, we build huge plants, hundreds of millions of dollars in investments; we can handle a couple of dashboards or digitization projects. We're smart people.

Usually those kinds of initiatives fail. In the end, one side feels vindicated because they were never involved in the project and now the other side knows how hard their work is. The side that initiated it usually pretends it never happened and doesn't touch that field again for a couple of years, until somebody new comes into that situation.

It doesn't look very positive. We can't scale, our silos are moving away from each other, and we haven't really seen any real evolution in the past fifteen years.

But there's hope. David and I talk to people and look at different sites in different sectors to see where they manage to make a difference, where they bring digitalization to the shop floor solving real problems. What we've seen is it's not linked to technology. It's happening in very diverse fields and sectors. One thing they did have in common was the way they managed to cooperate. Much like Dev and Ops had to learn to work together, they found ways that IT and OT managed to work together.

Technology is easy. We love it. Now AI has to be everywhere. But cooperation is boring, and it is not just something you do. We also didn't have a language for cooperation. Around the early 2010s, we took some ideas from DevOps -- apologies. If Dev and Ops is talking about silos, why not use it for IT and OT?

We started drawing circles on our own, with colleagues, with customers, and it really caught on. We found that just using those little circles, we could talk in a much more concrete and detailed way about what cooperation meant. It unlocked those discussions, and we moved away from statements like "we need to cooperate better" to actually devising something that works for us.

Here you can see the eight patterns that we saw in the wild, with some description. If you want to read more, there's more on our blog. That started a path for us where we started to experiment with these concepts within our own organizations. From that, we distilled a three-step plan. It's very simple. It is just a lot of work.

Step one is you need to bring those two groups together, but not too close. Often when you step in, there's mistrust, misunderstanding, history. You don't start a mega-project where everybody needs to cooperate flawlessly. You build a basis of trust. The easiest one is through some facilitation, where you have smaller projects, more limited in scope, and on each organization you have a person or group of people who will be the interface that will build common understanding, build some trust, and build relations across those organizations.

If you're a bit more daring, or the relation is not that bad, you could even have a temporary 4.0 team, just like you could have a temporary DevOps team. Just make sure you dissolve them, because if you don't, it gets ugly. We've seen it happen.

Once you have those initial successes, you can go to the second step. That's where the brunt of the work is. It is where you go from a more silo-based organization, which is very common in manufacturing, to cross-functional teams where you put IT and OT people together working on a common problem. Those problems usually look like maintenance, sustainability, energy management, and those kinds of things where you need both.

The source of inspiration for us there came from Team Topologies, which unlocked those ideas. With them, we started experimenting in our own organizations, previously at BASF and now I'm using it. The really interesting part for me is that suddenly organization and how we work is not linked to HR. It's not something done secretly by management or management consultants. It's an effort that you do with the people who are doing the work, with the teams, with the stakeholders.

Once you get this up and running, that's where you really make that step change. You need that first step of cooperation and understanding, but once you start working in those cross-functional teams, you are going to go from isolated pilots to really fast delivery of global-scale solutions. It's always surprising when you see that speed suddenly going up.

I just want to be clear: this step is really hard for manufacturing. At BASF, for example, you had whole departments with only heat exchanger engineers. Efficiency and expertise are at the core of their DNA. To start with this is quite a step.

Imagine you're well on your way with this. This is a conference also about Agile. We haven't even talked about it. There's a good reason. If you don't have the right teams in place, it makes no sense to introduce those methodologies to those teams or your organization. On top of that, you also don't want too much change. It's already a big shift to go to cross-functional teams. If you then take away the Gantt charts and yearly budget cycles, they're going to freak out and your change halts.

I'm not going to go too deep into this. I'm sure there are lots of very interesting speakers at the conference who can bring more insights into this one. But those are basically the three steps.

In summary, David and I think that digital transformation in manufacturing is only starting now. We're now seeing traction of those concepts, and we're noticing more and more success for those who implement those concepts. It's taken a while, but we're not so pessimistic anymore as we were a couple of years ago.

We've been talking abstractly about circles and cooperation, but to make it more concrete we added some quotes from the field, from people. I'm not going to go over all of them, but there is one that really stuck with me. We had invested heavily in getting a good team together, working together with IT and the plants to create a data platform and tooling that really helped the plants.

That's all nice on slides, but one day we got a plant manager telling us they had an intern who finished in three weeks a dashboard that we knew a year ago would have taken at least four to five months with different departments. It's very visible. It's not very flashy, but you will suddenly see that solutions get delivered faster, customers are happier, and in the end that's what you want. You want the plants to really manage their work and come up with solutions on their own based from a platform.

---

David Ariens

Absolutely, Willem. Absolutely. Time flies, unfortunately. To wrap up the presentation, of course we'd love to interact with you on the Slack channel or in person.

A couple of triggering questions: for those working in manufacturing, how does collaboration between IT and OT look in your environment? Is it a cold war all over again? Do you collaborate? What works, what doesn't? Maybe use our eight patterns.

As a second question, we believe a lot of people here at this conference will have an IT background. If you are curious about exploring the intersection between IT and manufacturing, we'd love to get in touch with you and discuss that further.

Lastly, please help us spread our word. We try to bring the word to the world as fast as we can, and we really hope -- and that's why we love talking here at this conference -- as Willem said, a lot of our inspiration, most of our inspiration actually, came from the IT Revolution books. Thank you very much for inviting us, Gene and colleagues, and we look forward to further discussions in the future.

---

Host Outro (Gene Kim)

Fantastic. Thank you, David and Willem. This is just great. As you have more success stories, we look forward to having you share those experience reports with this community. Fantastic. Thank you.