Join the Insider! Subscribe today to receive our weekly insights
David (00:00)
Welcome, you’re listening to the ITOT Insider Podcast. I’m your host David. Subscribe to get the latest insights shaping the world of Industry 4.0 and smart manufacturing. Today I’m pleased to welcome Gregory Grauwels. Gregory describes himself as an Industry 4.0 enthusiast and we’ll talk about that later in the podcast. He’s the Group OT Manager at Cloetta. Cloetta is founded in 1862.
is a leading confectionery company in the Nordic region and the Netherlands. So that’s here in Europe. They are a manufacturing and marketing company who are producing confectionery, chocolate products, nuts, and much, much, much more sweet stuff. Gregory, thank you for joining us today.
Gregory (00:47)
Yes, thank you for inviting me.
David (00:50)
My pleasure. So why don’t we kick things off by your introduction, your personal journey.
Gregory (00:57)
Yeah, who is Gregory? I live in Belgium. we have a little bit the same background that we come out of the petrochemical area where we worked. But maybe a bit before. And so I started working actually as a freelance music producer. So during my studies,
David (01:00)
Who’s Gregory?
Gregory (01:27)
I played a lot with some music, remixing existing music and that got out of hand a little bit and assigned some contracts and traveled the world for five years nonstop. So that was completely different. But there was a time that I came to a point I have to move back to my original roots, which was
David (01:42)
Wow.
Gregory (01:57)
the hardcore engineering part, just the mathematical, technical stuff and not the, how to say, the artistic side of my life at that point of time. So I started working at an automation company as an automation engineer. So I started programming PLCs, HMIs, SCADA systems.
all the typical Siemens products that we usually have in Europe. And then I progressed to a bigger company called Bayer, Monsanto at the time. And there I was a digital manufacturing plant manager. So focusing on all sorts of industry 4.0 initiatives. So digital maturity assessments, but also cybersecurity.
and the MES/MOM layer as well. So yeah, that was till I joined Cloeta, my current position. So I went from a petrochemical industry to a confectionary, which is completely different. So that was exciting.
David (03:18)
Yeah, I can imagine so. can imagine so. for I always like to explain a little bit, I would say the different industries in the world in this podcast, already had many. Actually, we didn’t do a podcast yet on the chemical industry itself. So we actually might want to do a retake between the two of us. But this one is about confectionary. So what does it take to produce?
Candy, how does a confectionary company look like?
Gregory (03:52)
So very simple. You put in raw materials, you process the materials, you mix it, you give it a color, you give it a smell, a taste, a form. And at the end, that’s an eatable product. And that eatable product, you package in all sorts of sizes and so on. So at the end, you go from raw material into a processing line to a sort of semi-finished conditioning.
and then from that on you go into the packaging lines where you package these products going on all sorts of packages and palettes and so on to all sorts of customers.
David (04:40)
So does that mean because you’re producing a big variety of sweets, does that mean that there are a lot of different production lines? Or is there some kind of standardization? Is it possible to use the same line for different products? Or is every plant, every line totally different? Can you give me some insights on, would say, let’s call it scaling possibilities or whatever you want to call it?
Gregory (05:09)
Yeah, of course, it really depends on what you’re producing, but we call it technology. And that’s a bit, how to say, fighting with my technology term in my mind. But technology that we have in our company is really the food technology. So how to make a certain product. So a technology to make wine gums versus chewing gum.
or versus jelly bean or chocolate is a complete different technology. So that comes with different machines, different assets, different lines. So it really depends. And that’s interesting because a winegum, I would say at the base, you mix a lot of materials and you give it, like I said, a taste and a color and so on and so on. But the end product can be completely different in a sense that you can put
sort of salt or an oil finishing or like a sugar finishing. You can have it in different forms and colors. So you have all these kind of different things based on the same kitchen slab, we call it. So you can make it like very lengthy, like we call it in the Netherlands, snore. But you can have it like stacked.
David (06:20)
Yeah.
Gregory (06:37)
like the English drop, the English flickeris. So you have all these kinds of different things and then machines start mingling with each other. So production lines we produce in one factory, ship it over and finish it in another factory. So it becomes quite a complex thing, I would say.
David (06:37)
Yeah.
I can assume that for an OT manager that creates quite some challenges.
Gregory (07:03)
Yeah, yeah, and it’s old and it’s new. we have, you said that we were founded in the 1800s, meaning that we had lines from that area almost in the sense that we have very old machines as well creating a particular product, which was, for example, I think in Swedish market, had like Mooms Mooms and all these kinds of things.
specialized sort of candy well known in that area. And that’s just like one old school machine just making that particular candy. And now we have also the very fancy brand new packaging machines that comes with all the bells and whistles. excites me, of course, as well.
David (07:55)
Can you imagine so? Can you imagine so? And yeah, if you have a machine, no matter how old it is, if it’s performing, if it’s doing what it has to do, yeah, then there is obviously no reason to replace the machine just for digitalization purposes, right?
Gregory (08:08)
No, of course not. And an asset delivers, yeah, you buy an engineer a certain asset to deliver a certain product. So if it’s delivering, then that’s it. So the data part and all the technology where I’m focused on, that’s like a second stage.
David (08:27)
Yeah. something which is all what I like very much in our domain is that we are sometimes talking about assets which have a certain life time and they’re already sometimes they are 10 years old or 20 years old or 30 years old. So now we’re talking about assets which are, which might be 100 years old.
Gregory (08:48)
Yeah, not anymore included, let’s call it that. That was a miss. But yeah, indeed, all sorts of life cycles.
David (08:53)
Okay, okay.
Hey, so you started as an OT manager at Clobetta. Now in your previous role at Bayer Monsanto, you were dealing with digitalization more from a smart manufacturing point of view. you weren’t, I would say, is it correct to say that you didn’t have the end responsibility around the entire OT domain within Bayer?
Gregory (09:27)
No, buyer is a really huge company. So it’s 120,000 people working for that company. And I was focusing on a particular segment. So Monsanto was focused on seats. And then Monsanto also has a petrochemical part, which is making the roundup as we know it. So you have also the pesticides.
different departments, different technologies at the end as well. So, yeah.
David (09:59)
I would say specialized focus probably. Now you changed over to becoming the group OT manager of a multinational company with multiple sites. I think for our audience, many of them will be in an OT role. Some might change from being an engineer to an OT manager. Some might…
that within the company they’re working in, some might join a new company. So I’d love to take, I would say, just one step back and go back to that original journey for you. What does it take to start as an OT manager within a company? In your case, Claretta, but let’s make it generic. what do you think it takes to when you start at day one, so to say?
What are the steps someone needs to take when you enter this role?
Gregory (11:00)
It’s a good question.
Indeed, I’ve been progressing from like a project engineer towards more a manager, I would say, and now entering a field that’s also sometimes, I think, unknown for companies, like what is OT in general? What is it? Because we are so used to have E &I and process automation in place, like PLCs and HMIs. But then if you talk about MIS, MOM,
and infrastructure, data infrastructure and databases and all that stuff, then it becomes complex or you come more into an IT area. So what is an OT manager responsible for? And I was lucky and that’s why I jumped to this position was that I could form that kind of position itself.
And in my mind, I have four domains to focus on. So the ENI and so the electrical instrumentation and the process automation side, that’s like, I think a certain given. you buy machines, it comes with the PLC and an HMI. You need to plug it in, in the power, but also in the internet with an internet cable. So that’s the baseline.
But then you start using data and you want to have applications on top of it. So more plant-wide SCADA systems, but also a mass manufacturing execution system and manufacturing execution system itself is very broad. If you talk about MES or MOM, yeah, you can go from left to right, from up to down, you can land everywhere, I would say. So what is then that area? And then…
David (13:00)
Mm-hmm.
Gregory (13:02)
A big part is also network blueprint. So what is the cybersecurity measures? How do you set up a proper network blueprint in your company, in your factory? Because it’s a completely different ballpark than the IT world, I would say. You have very specific needs. And for example, a big difference is that we highly focus on human safety. And that’s really key.
whether you work in petrochemical industry or in food and beverage, human safety is always key. And then you have your assets and you have your food safety. And that’s really interesting. So petrochemical, I think the real focus was on human safety because you’re dealing with a lot of crazy products. While with food and beverage, yeah.
A candy will not kill you, but it’s the machines and you have these kind of things. And of course you have metal pieces that can end up in your food. So it’s more the food safety that’s really crucial for also your customers. So yeah, very interesting. So for domains, OT network blueprint, cybersecurity part. So if a supplier comes with this machine and has an ethernet cable,
David (14:06)
Mm-hmm.
Gregory (14:24)
or even the devices with 5G. I really want to avoid that one in your factory. But how do you connect it properly in a safe way? How do you have your applications running, like printing machines? You have your Mesmom, like an OE application and all this stuff, and then your SCADA systems. So that’s like the four domains that I identified as a focus point. And then…
I would say if you enter a company, then you first need to do a sort of situational awareness. So how do you become aware of what is alive and kicking in your factory? And where are the gaps? Where are the opportunities? And most of the time, that’s proper way to do a digital maturity assessment.
So for example, understanding the maturity of PLCs, HMIs that you have in place. What do we do with the data? Do we have data historians in place? What kind of applications are there in a factory? And then if you have to manage multiple factories, then you have to see the red line across these multiple factories.
Are there common applications or is it like all silos and every factory is on its own? So you have all these kinds of complexities. And at the end, I think you start to form a sort of awareness and also feel the opportunities and the gaps that are there for the company. So then you can start building also your requirements and you see where the gaps and opportunities are in your domains that you define.
So are these opportunities in more the process automation level or more in the top down approach with networking and complicated ERP integrations are necessary. So that’s like that whole awareness creates a sort of requirement, also define the requirements, I would say, but also define sort of strategic steer. And I think as an OT manager,
I’m the end customer. So I have always been working as a sort of integrator, but working as an end customer and being in that position, you really need to define the right steer for the company. And that’s a certain pressure, I would say, but also certain, it gives you a certain edge to really try to find the best way forward for the company.
David (17:09)
If I, this might be a bit black and white now, but I’m gonna ask the question anyway. So I’ve seen a lot of maturity assessments on cybersecurity, on innovation, on whatever. Some of them are like three question, questionnaires. Others are like spreadsheets who seemingly they have no ending, right? You go like, another tab, another tab, another tab, another tab.
How do you, if you do such a maturity assessment, and whether or not you’re doing it in Excel in your heads or in something other, that doesn’t really matter, but how do you balance such an assessment that it’s not too complex, but it’s kind of asking the right questions, but not too much questions. You don’t really bother, I would say, the people.
who are working in the plants too much but they are still triggered by your questions. I always find it interesting to figure out the right balance here. Was it also trial and error for you? Did you change in how you do such assessments?
Gregory (18:26)
Yeah, good question. I was lucky, I think that I, during my time at Bayer, I was leading also the digital maturity assessment, but that was also with an external company and with a lot of knowledge coming from the company Bayer. So I got some insights in how to perform a digital maturity assessment very lengthy. And that’s like a full week.
identifying all the key stakeholders and then having a proper discussion with all the stakeholders. So at the end, I think you need to spend one week at the factory, just to understand how the factory is running, who are running it and how do they run it. So just have an open question and I think being on the shop floor and just follow a person and let them explain what they do and let them explain.
the good things, but also the bad things and where they see the opportunities and what would be their wishes if they would be in control and could change stuff. So I think that’s quite not going too lengthy and having all these kinds of standards in place and then knock someone dead with the 10,000 questions, but just try to understand what we are doing and why and are there improvements and gaps?
David (19:32)
Yeah. Yeah.
But that’s so important and I fully agree. First of all, the fact that you need to be there physically on the shop floor, talking to people. Like there is no way you can do such an assessment from an office building over just a remote teams session or from headquarters because then you’re missing out on so much, I would say, important insights.
And also what I also liked, and just want to repeat that because you also said like, yeah, open questions, walk with the people who have to work with the solutions on a day-to-day basis, figure out what really bothers them today, what is working well, what is not. Because here again, I’ve been in discussions where if you’re talking just to the management level, they would go like, yeah, we have to capture that and that type of data.
Why? Yeah, because we need to build KPIs, cetera, et cetera. But if you then go to the shop floor, the operators go like, yeah, but this is just utter bullshit. If I need to enter all these data points, first of all, why would I? And secondly, how is it actually helping me my job? Question mark.
Gregory (21:02)
Exactly.
Yeah. And I think that’s, that’s like you say, it’s really key. At the end OT manager, I’m really in service of manufacturing our factory plants. And my key customer is like an operator because he needs to do his job every day. And I try to provide him the right tools. And, and, and, and I can say this thousand times over and over again, like
David (21:22)
Mm-hmm.
Gregory (21:35)
Technology is not the solution. And so it’s only a tool. It helps you, but it’s only a tool. And for the maturity assessment, I really try to find the real issue. Not that the need is there or a wish list that you need to have a new car, but why is your current car not running well? So what is like the real reason? Is it like an organizational problem or is that like a process problem?
Do we do it differently in another factory? Why do you have the problem and that other factory not? So that’s what you try to find out. And then only then you find out really the functional requirements and the technical requirements to define the right tool for it.
David (22:22)
couldn’t agree more. then, but then there is still this problem we call building a business case. At a certain point in time, and I fully agree with you, Gregory, I fully agree, but at a certain point in time, you need to get money. So somehow you need to convince management about the fact that spending is needed in one way or another.
Gregory (22:30)
Yeah
David (22:52)
Do you have some insights to share on how you, so you’ve done the maturity assessments, you obviously then go back and you do your homework and you build some kind of a plan or some kind of a roadmap. Then at a certain point in time, you need to translate that roadmap into funding, into a project, into maybe in some organizations, they prefer this big bang type of projects which I’m not.
big fan of but I’d like to hear your opinion on that. Or you want to go for more a small-scale agile approach but with the drawback that you need to go typically back for funding multiple times or maybe some kind of a hybrid approach. What have you seen in your career?
Gregory (23:42)
both, I think. And that’s really tricky.
because.
I would say.
Technology is a tool and it’s an enabler, but it’s sometimes quite tricky to attach direct savings to an enabler because you enable the savings. So you need to believe in what you try to enable to get to the enablements. So that’s quite tricky. So there are a lot of ways how to define a business case and to do…
to define the return on investment and so on and so on. As you can, for example, if you, and probably we go into that topic later, but defining the proper infrastructure and having a cost avoidance, for example, is a major thing. Then you still need to spend money, but you can do it for less money or you can do much more for the same money. So it’s the way how you see it. So the same like a business case where we see
where we say we can avoid X amount of FTEs, so X amount of people. maybe you don’t want to reduce your people, you want to give them better tools and enhance them and make them more efficient so they can focus on other stuff. you can spin it in a lot of different ways. And sometimes it comes with a very big bang, like a bigger company, like where I worked previously.
That was a big bang. But my approach would be that you do it in a small scale pilot and really drive the benefits and then people really will see the benefits and then you get that pull from the factory and automatically you will drive that digital transformation that you’re seeking. So not enforcing digital transformation, but really
let people drive the digital transformation effort.
David (25:57)
It’s also bit rare, you you have the, I’m going to say old school. I don’t mean outdated, but I just mean old school, all the ways of working within an OT organization used to indeed procuring assets with a long life cycle, designing engineering, for example, PLCs, or you also talked about safety. we really want to make sure that the…
programming code within these PLCs or maybe even safety PLCs is running perfectly, so to say. And once it’s developed, typically it stays there for quite a while. Let’s put it that way. Now, on the other hand, you have influences from the IT world. IT embraced this agility, scaling DevOps type of mentality already many, many years ago. And we see that
IT technologies, but also I think IT mindsets is coming to the OT world. This is also something which is definitely driving industry 4.0 type of approaches. How have you seen these IT technologies, IT approaches coming into the OT world?
Gregory (27:21)
Yeah, I think that’s what’s really driving me in the first place. I come from the typical automation engineering where you have this pyramid. So in every layer, you have a problem to solve with a certain application. and probably you stick with the same brand as well. So you can buy the whole product stack from one company.
and I have an MES and then a PLC and an HMI and everything from the same brand. But at the end, you try to solve things and the old school way or the old way was that it was solving a problem at a time and then you got this tech solution. But now we want to use data everywhere and everyone needs to be accessing that data. so
What I really like is that image where we move away from that pyramid and go into a more networked structure architecture. Isa also shared that image. And I fully agree that you need to build the right data infrastructure. And for me, that’s like event-driven architecture. So how do we leverage real-time data in the factory? And then you have these unified namespace principles that comes around the corner.
but also like microservices. How do we become more vendor agnostic, create an open architecture, and every bit and piece we just can swap and replace and just play with it. And at the end, it will not impact the total, but still make it better. And having this really open infrastructure and driving real-time data, which is easy to maintain.
but also needs to be secure, reliable, performance, scalable, all these kinds of things. That’s what’s really keeps me going, I would say. So going from a PLC, HMI, installing a Windows server, installing Scala, spending like three days installing applications towards using Linux, Docker, one command, five seconds later, the whole architecture is there. So that’s like amazing. How will we…
go from a old school server installation tool, for example, Docker containers, installing software like ignition, capware, installing brokers, high bytes, all these kinds of functionalities, all these applications, all these technologies coming towards the industry. That’s really what I love. I think the industry always followed
the big brands and I feel a shift where the end user becomes the owner and decides what they need. Instead of buying the whole product stack, we’re gonna decide where your product ends up in our stack. And that’s what I really like.
David (30:17)
Thank
Yeah, more this open, interchangeable architecture approach. So you mentioned the unified namespace as a concept, as a design pattern to organize all these bits and bytes, so to say. It’s a concept I also like a lot. Now, I can imagine that your approach when you, I would say, approach
round field versus a green fields? Is that is that is that difference? How do you? How do you? I would say apply new concepts, new ways of working not only the unified namespace, but also new technologies to to a brownfield plant. How do you I would say separate the things which are already there from from what’s what’s new, for example, how would you do this type of projects?
Gregory (31:24)
It’s a really good question because that’s one of the key questions that I have to keep in mind as an OT manager. So if you want to form a strategy, you need to build a strategy that’s for green but also for brownfield. So starting fresh, that’s always, I would say, easy and nice. But having all the brown stuff in place and putting the same stuff on top of it or
integrated, that’s the real challenge. But at the end, think the, and that’s why I like really technology driven architecture. It’s like these microservices and sometimes in a brownfield, you already have pieces in place and it’s not that you have to swap it. You can just integrate it and you swap it with the one that you have in your strategy, I would say. And there are, and then you can make the case whether you swap.
should swap it or not. But at the end, the key question is, what are you seeking? So what are you trying to do? And the unified namespace is only, for me, it’s not the answer. And so it’s, maybe I gonna step on the tone, step on the feet of Walker ends, I would say, but unified namespace is not the answer.
It’s a tool as such, and I really believe in what a unified namespace is. But at the end, you try to seek something and connecting your machines, whether it’s old or new, connect them and collect the data, store it somewhere, visualize it, transform that data and publish that somewhere that everybody and everything can access to. So you have like an actual state of your business. That’s like for me, unified namespace.
David (33:20)
Yeah.
Gregory (33:21)
but then you have a lot of complexities and yeah, that’s really interesting. How do you set it up properly?
David (33:24)
course.
I’m sure Walker wouldn’t mind that you’re calling me out here. Obviously, there is not a single technology or a single concept which is solution to everything. That’s maybe exactly the thing we’re trying to solve here in this discussion is how do we make sure that we combine the best of the best concepts from different worlds and merge them into each other.
which I also like or I also see changing is that as you talked about more the traditional stacks, the pyramid, I don’t like the pyramid actually. I never liked the pyramid because the pyramid implies that somebody is on the base and somebody is on the top and the top is always very small and the base is always really wide. anyways, besides the visualization of the pyramids, there is another thing that’s, let’s call it.
open source versus closed source or open source versus very proprietary type of software. I do see a lot of open source software now coming in into the industry as well. And I see that I would say as it’s positive, but it also has some challenges. So why do I think it’s positive? It’s because well,
The fact that you see what is happening, the fact that you can maybe contribute, that you have a community which can contribute. The fact that if a vendor would go bankrupt or cease to exist, that you would still have the source codes to continue. And there are a lot of other reasons as well. But there are also some challenges because in many cases, you can be on your own. You might not have a service contract to rely on. You might want to, yeah, you also have this
Again, old school way of thinking versus let’s try and see what goes on or what happens. At this point in time, I still find it difficult to balance out the role of open source software within any context, but especially here in an automation context. And I’d love to, I also love to get your insights.
on this topic, how do you balance this out? How do you make sure that if you are not there, that’s something goes on that there is still somebody to support.
Gregory (36:09)
question. It’s a multi-layered answer, I would say. I’m going to give it to you. First of all, I think you need to have a solution in place that’s, like we said, it needs to be secure, reliable, performant, scalable, maintainable, all these kinds of things. And how you do it, that’s on the second question.
David (36:14)
I like multi-layered answers.
Gregory (36:39)
and the how is in what kind of tools do you use? So it’s not that an open source is less reliable than a closed one, like a monolithic solution. open source is, there is a difference between open source, open and more vendor agnostic, I would say, in a sense that I can fully agree with the monolithic
kind of system to buy if as long as it’s open. And that’s the problem what we see with a lot of systems that it’s so they create their own ecosystem, a closed ecosystem. And that’s a problem for me, because then either you have to pay a lot to get it open, or it’s just impossible. And if then a tool is open source and it’s
having the same from a risk management perspective, the same level, I would say why they’re not. It’s not that it’s open source that it’s bad. I do believe that. And that’s what I really like. For example, as a Walker Reynolds really pushing is the land and expand model. So you need to be able to have first a transparent cost structure because you have a lot of systems that you need to pay for a user.
David (37:56)
Mm-hmm.
Gregory (38:07)
for a client, for a data point, for a server. So you have all these kind of complex cost structures and you have no idea what you will pay in the end. And for me as an OT manager trying to build a strategy and then being called out at the board meeting to say, what will it cost at the end? I have no clue because.
David (38:23)
Mm-hmm.
Gregory (38:34)
even the vendor cannot give me the final price, I would say, or it’s like ridiculous expensive. So the point is that you need to be aware of the solutions, best fit for purpose, I would say. And especially with the pilot, you can really use open source very well and very fast. Like if you just want to solve something, you can use like…
free tools like a notary and whatnot, but also just having proper solutions in place with the trial license. So the land and expense. So you just trial the stuff and you maybe reset it every two hours or a week or whatever. But at least your point is proven at that time. And then I think the other layer is your organization. I think that’s really key.
David (39:21)
Mm-hmm.
Gregory (39:30)
because how are you organized for handling these technologies? So if you’re going to buy a new car, but you don’t have a driving license, then yeah, you spend a lot of money for a nice car. So you need to have, I think, in-house knowledge and expertise, but also outside your company, you need to have the right partners in place. So for me, that was really crucial to be knowledgeable.
David (39:35)
Yeah.
Yeah.
Gregory (39:57)
from the inside, but also have the right partners to rely on. And all that together comes then to an open infrastructure. That’s, think, really key. That’s the objective.
David (40:11)
Yeah, I like the lend and expand approach. I also have the feeling that sometimes we are talking so long about how to potentially solve a problem that either the problem becomes for it or somebody says like, you know what, let’s just keep on doing whatever we’ve done the last 10 years. So, but just trying it out.
as a caveat, what I’ve seen with a lot of these, of these projects is that, and I’ve, I’ve been guilty myself. So I’ve, I’ve done a similar thing. how long ago, like 12 years ago at BASF just, here is a very, here is a simple solution, which we’re gonna, we’re gonna use it just for a couple of months until we got something, we have something better. Now I know that thing is still running 12 years later.
Gregory (41:03)
Exactly.
David (41:07)
So I do believe that you need also within land and expense, you need a roadmap. You need to know like, okay, this has a lifetime of so long. And after that lifetime, either we need to kill it or we need to build a service model around it. And that might, I would say that you might wanna go for
a product which we did as an SLA or some kind of support model or you still accept to go for a product which you need to support yourself but then you also need to be the one who is on call 24-7 maybe. How do you build these types of Lend and Expand models within a multi-year roadmap? How do you make sure that
You can also build an organization around it. How do you make sure that you’re not just building loose elements?
Gregory (42:18)
step by step. I think we are always very eager to know the end game. A typical financial person will ask you what will it cost me at the end. But every time you have to say digital transformation is just an ongoing thing. So you’re transforming, it’s a moving part. putting like an end number on it, it’s just impossible.
David (42:20)
Ha ha ha.
Gregory (42:46)
So you have to just take step by step and prove it. And while it’s proven, then you get more traction. And that traction will solve then the financial part, et cetera, and the organizational part. Because if people see it and they really believe in it, then they will give you all the resources needed, whether that’s time, people, or money. So I think that’s what I really learned. Focusing a lot on these fancy slides and like this multi-factory rollout and timelines.
then you come into that waterfall kind of thing, which we don’t like. We need to be agile, we need to be flexible, and you don’t know what you need or want in three years time. So maybe that can be something completely different. But by just step-by-step solving something, it will actually also guide you to a certain direction, I would say. But of course, you need to think about it. But I think the…
David (43:32)
Yeah.
Gregory (43:44)
the vision I would say should not be the path itself.
David (43:49)
We mentioned unified namespace. We mentioned Walker, but we didn’t really define the unified namespace yet. Or at least your definition, because, I have mine. There are a couple on the internet as well. But I’d love to hear your take on what do you think a unified namespace is, and also, what does a unified namespace concept, I would say, bring as value, or why should you go for that kind of…
of design approach.
Gregory (44:21)
Good question again.
Again, there are multiple layers on a unified namespace. But for me, it opened my eyes that you need to start collecting data and putting it in an appropriate hierarchy, I would say. And we all know the ISO 95. So you have your site and your areas and your lines and then your units. But then the
It’s an intriguing question. How do you solve then that hierarchy itself? So you have also functional domains within a line. So you have like this route data that comes in from all sorts of sources. You have like this functional domain where a lot of MES metrics lives or like ERP master data that comes in. But you also want to collect status of your line or machine.
production metrics, whether it’s at the schedule or whatever you produced, consumed, wasted, scrapped, whatever. And also a lot of measurements, settings, maybe maintenance metrics and all that stuff. So at the end, you want to have all the information that is needed available in just a simple topology, I would say. And that’s, think, very key. And then
The backbone, the technical backbone is how do you make that data available? And that’s of course via a broker. So you publish that data and you consume data via broker. And a broker is like this kind of place where you just throw all the stuff at and it’s living there at the center. And at the end, what you want to create is that you have this single source of truth, they call it, but at the end, it’s really the actual state of the business.
real time because it’s not a data historian. It’s representing the data as it is now. So delivering an actual state of the business, I think that’s really key. And once you have that data available, then you can do whatever you want with it. So then it becomes interesting. What are you going to do with it, with that data? And I think that’s the key question. What are you going to do with the unified namespace? Because setting it up, I must say it’s quite easy.
David (46:43)
Ciao.
Gregory (46:49)
whether you use, for example, like an ignition, you have tag provider and you make this structure like UDTs or whatever, and you just publish it even if it’s open, if it’s OPC UA or it’s MQTT, whatever, the architecture is there, but with the broker and MQTT, you have disadvantages of publishing and consuming data. But then how do you really form your ISO 95?
Only that is, it’s like a massive discussion. I would say, what is the line? What is a unit or a cell? So it’s, it’s really, it’s stupid, but some companies it’s a given. I come from a better chemi chemical company where that’s a given. You have P and IDs and then you have like a flow transmitter and that’s all sorted out. But then sometimes you also come in a company and you don’t have that. So what is the line? Is that like.
David (47:25)
Yeah.
Okay. Yeah.
Gregory (47:47)
a process line and a packaging line at all at once because that’s the line that’s producing a product? Or do you say no, packaging is a packaging area and you have a process area and you have to shop maybe a direct line into two pieces? So that becomes a bit complex. And then defining all the data that’s needed. That’s like all these functional domains and how do you give that a certain topology? think that’s
Yeah, very interesting.
And if I may add maybe, for me, crucial is that this triggers the way you think about data. And that was for me really key. So instead of trying to find out what data is available, you come more into ownership and you define what data you need. And that’s really key. You’re going to define
David (48:33)
Yeah.
Gregory (48:52)
towards a supplier, what kind of data that person or that supplier needs to deliver. And that’s really key. So for example, OE is a typical metric that you use, but what is the source data to deliver you the OE metric? And that source data, that data point, that’s what you need. And that should be a given, that should be a requirement towards your supplier to give. And I think that’s the key of also unified namespace that you start thinking about.
David (49:01)
Mm-hmm.
Gregory (49:22)
What kind of data do I need to transform it really in information so I can use that, bring context with applications or whatever to really get the actual state of the business. Far more than just raw data points moving around.
David (49:37)
Yeah, I like the concept or I would say the brand UNS. But as you also said, in essence, it’s not new, right? We’ve been working around ISA95, around data modeling for many, years. And one of the things I always found very tricky indeed was to just…
come up with that data architecture. It’s actually a data governance problem in a sense. And it’s also something we’re not really used to in OT. think in an IT environment, can’t build an ERP system without having a data governance model. You have to define how an invoice looks like, and you’d have to define the fields on that invoice, and you have to define how an order looks like, et cetera, et cetera. It’s the, I would say it’s the…
normal thing to do when you’re building an ERP system. But for some reason, although data modeling has been around in the OT world for many, years, for some reason, it’s still something new and we don’t really feel too comfortable with that. Maybe a few tips on how to get started building such an hierarchy. do you, as I said, what…
Do you need to go for the full blown model? Do you wanna go for a very small one? Do you wanna involve everybody? A few tips.
Gregory (51:08)
Yeah. So I think you can try to approach it from different ways, I would say, and maybe come back to that OE metric. So at the end, the whole OE metric is also availability, performance, and quality. And at the end, you only need a few data points to deliver this kind of metrics.
And the point is that you have to decide what is really key for you. And so if that’s an approach where you say that’s a key metric and I go to the source of it and that’s like how I make a sort of template out of it. And that’s what I need to get out of every machine that can be done an approach. But I would suggest that you start small because I built a whole UNS kind of data model.
with all the bells and whistles. But then it’s almost, then you come into that brownfield greenfield thing. You come to a machine that’s maybe 60 years old. And if you’re very lucky, you can maybe get one sensor. Sometimes you have PLCs that are just not open. don’t have the source data anymore. You don’t have the program available. The supplier vanished. I don’t know.
It’s an old school thing. So it’s not always possible to get all the data out of it. At least for example, a state would be nice, whether it’s on or off, whether it’s running or not, that would be nice. And if it’s not running, why it’s not running? Can you get certain alarms out of it? And for example, a pack ML standard is a nice standard because it states like how you go from certain states and what do you need.
So you also need like a speed. You would like to know how fast it runs. But again, a PICML gives you a certain standard, but it’s quite empty in the sense a supplier can just fill it in how everyone wants. And that’s a bit, I think the pitfall of the standards these days. You have ISO 95, you have 88. I read them all and I, yeah, I get like a
how to say, print-hoeft in Dutch. We get like a very massive brain, a massive headache just by reading it. It’s made for doctors. I don’t know, but I’m maybe a simple guy, it’s it’s really nice to try to find a standard that fits everything, but at the end it fits nothing because it’s not practical. So really try to find the practical way.
David (53:40)
A massive headache or something like that, right?
Gregory (54:04)
having just a few states in, but like an OE, just have a sensor that sees whenever a box or whatever material is along and passes that sensor. That sensor alone, that one data point can give you the whole OE module you’re seeking for. And that’s, think, really crucial that you focus on these kinds of things and not go after meantime between failure and all this crazy stuff.
Because at the end, yeah, you need to give some priority on what you really want to get out of it. But at least building the model, I think it’s nice to.
David (54:42)
And again, to come back to your previous point, your number one customer is operations anyway. So that means that what are the things you can do to help the plant produce what they need to produce in the best possible way, in the most efficient possible way, et cetera, et cetera.
Gregory (54:59)
Yeah, exactly. And just knowing the state and already giving you very fast glimpse of what your line is doing and all these sorts of things that’s like question number one. What do you really want to see? And then we’ll figure out whether we can get that data. And if we can get that data, ideally, you also format that in a structured way. You make a template out of it. So we call it a UDT.
And then ideally you also make a visual component that’s linked to that unity. And all of a sudden you have like a template, a template kind of skater available. And then once you go to factory A, B or C, you just need to have a state or you just need to have an OE sensor. And all of a sudden you have this whole application running by a few data points. And that’s, think is key for a unified namespace that you identify these functional domains with a few data points.
And that’s how you start building your whole unified namespace for the company. But don’t try to already breach PLCs with the cloud. If that’s a need, fine. You can also have a business case that you just want to do predictive maintenance. But I think that will not solve the unified namespace principles for your factory or company. So you need to think bigger. It’s how do you build the whole infrastructure properly?
David (56:28)
Yeah, super interesting. There are a of things to remember from this talk. remember step by step. Absolutely. I land and expense. I really remember open architecture. doesn’t really matter whether you go for open source or closed source, but open architecture. That’s what the industry needs. I fully, fully agree with that.
So with these three highlights and there were many more, with these three highlights, Gergery, thank you very much for joining this podcast. I’m sure we will talk again in the future. And to our listeners, thank you for tuning in. Thank you for joining. Make sure to subscribe and we’ll be back soon with a new episode. Thank you.