Enabling Automation Podcast: S2 E9

We’re excited to bring you our first-ever podcast series, Enabling Automation. This monthly podcast series will bring together industry leaders from across ATS Automation to discuss the latest industry trends, new innovations and more!

In our ninth episode of season 2, host Simon Drexler is joined by Christian Debus to discuss The digital layer of automation.

What we discuss:

  • How do you start using data from automation to drive optimization
  • Best practices for dealing with data
  • Automation layer vs digitization layer

Host: Simon Drexler, ATS Corporation (ATS Products Group)

Simon has been in the automation industry for approximately 15 years in a variety of roles, ranging from application engineering to business leadership, as well as serving several different industries and phases of the automation lifecycle.

Guest: Christian Debus, ATS Corporation (Process Automation Solutions)

Christian Debus leads the Process Automation solutions group at ATS which helps customers optimize their production processes utilizing automation and digitalization to make use of their integrated data. Prior to this, he worked for many years in the automotive industry and lead a global filtration business.

——Full Transcript of Enabling Automation: S2, E9——

SD: Welcome everyone to the Enabling Automation Podcast, where we bring experts from across the ATS group of companies to talk about topics that are important to building automation inside of your operation. My name is Simon Drexler and I’ll be your host. I’ve been in the automation industry for about 15 years, doing a variety of roles, delivering both growth and innovation to companies that are large and small. Very happy to have an expert from inside the ATS group of companies to talk about our topic for episode nine, the digital layer of automation. So we’re joined by Christian. Christian, do you want to take a moment and give the listeners an introduction to yourself?

CD: Would be happy to do so. Thank you very much, Simon, I feel super honored having the opportunity being your guest here. I’m Christian Debus. I’m leading the PA group, which is coming from a to being a traditional automation integrator to now a company which is driving value for our customers by helping them to control and continuously optimize their production processes with using automation and digitalization to make use of their integrated data. And before that, I was working many years in the automotive industry, and right before I joined ATS, I was responsible for a global filtration business.

SD: Just a fantastically varied background, Christian, we’re so fortunate to have you. So thank you for joining us. Our topic today, the digital layer and the digital layer and Industry 4.0 has been a primary talking point for our industry for quite some time. And how you use data to optimize inside of this world that’s been transforming right before our eyes, industrial automation. And then on top of that, you just talked about PA and how you’ve driven a lot of transformation inside of PA to keep up with all of that change. How do you start when you’re talking with a customer and how they can use the data that comes from their automation to drive optimization inside of their operation? Where’s the starting point?

CD: So the starting point is always customer value. So the starting point is so what? So what, what can we do with that? What, what can we, what can we improve with that? What does that digitalization help us. So and we, we notice actually that many customers are super eager to talk about digitalization and they are kind of contacted by many companies who offer them some digitalization solutions. But what we figure out is what very often remains unclear to our customers. So digitization is a cool topic. We need to do something with that. But what exactly are we going to do with that? So where’s exactly the value souls or what is exactly the benefit I get from digitization? And that’s exactly where we start. So we do not sell digitalization and we don’t sell automation, we sell value to our customers, which is in the end having better production processes, more productive, better quality, higher flexibility. So and here automation is helping and digitalization is helping us well. So that’s basically the entrance point.

CD: The second point where we basically get into and this is where our core competency is and what we are doing in PA since 37 years in the meantime is really connecting O.T. and I.T systems because that’s also something we noticed with our customers when they say, okay, I see the value, I see that I can make use of my data, of my integrated data to drive productivity improvements in my production processes. But the holdup usually is then how do we get the data out of the systems? So usually you don’t have a greenfield industry 4.0. You build from the scratch. The customers we are talking to, they usually have existing plants, existing factories, existing equipment, some new equipment, some older equipment, some very old equipment, some equipment from ATS, some equipment from, you might wonder about this, from companies which are not ATS. And all that needs to be brought together. And that’s where the big hold up for digitization is usually so really getting the different O.T. and I.T standards connected, get the data integrated, get the data contextualized to bring the data even to a point that you can drive value out of the data. Many companies sell analytics, artificial intelligence, machine learning. But first you need to have the data, and that’s basically where we are coming from. And that’s why we said, you know, I mean, we are doing this since 37 years, so we feel qualified for driving the next level of automation, means making honest use of your integrated data.


SD: Right. And I’m so happy our conversation started there because without that, you can’t drive your analytics solution. And that’s commonly a cause of concern for the customers and partners that I work with as well is you have all this variation and there’s no consistent way to pull the data from all these different pieces of equipment. Do you have a best practice or best place to get started for those that are listening and haven’t started dealing with this problem?

CD: So most companies currently do it like we used to do it, do it really from scratch for every customer. Again, basically 100% manual. So integrate every single equipment manually into a platform or into a database, which costs a lot of effort. What we figured out is with a certain focus on industries and we are focusing on the pharma biotech industry, on the chemicals industry, and in the food industry you have some equipment which is repeating. So that means you can build a certain standard, connect this to some standard equipment. Nevertheless, every customer has its own set up. It’s specific specifications. So the last mile you always need to do manually. But once you have done the last mile manually, then basically you can standardize a lot in terms of data integration and data contextualization.

SD: Right. So you’ve standardized where you can. To be able to extract this data. But then every customer’s problem is a little bit unique and it takes a little bit of finesse based on experience to be able to pull it out.

CD: That is exactly the point. Plus, I mean, what also plays a certain role here is having a domain knowledge. So we figured out, you know, with understanding the industries, we also have a certain understanding what data are relevant and what data drive action so you can kill yourself by collecting all data and pulling. I don’t know how many data in the cloud. This makes your interfaces slow, this makes your response time slow. But by connecting directly the right data, you save implementation time and basically you save implementation costs and you drive the right impact and the right insights from the data. So having the industry knowledge gives you a certain efficiency and gives you also effectiveness in the data integration.

SD: Right. Because at some point there’s a path of diminishing returns, from the data being extracted. You can only provide so much analytics that the team is going to do something with. As you were talking through that you very deliberately separated the automation and digitization layers. Was that done deliberately? Do you see those as being two separate layers of the implementation at a customer site?

CD: I would say automation for me is, is more like the O.T. integration. And what does O.T. stand for? Operating Technology. The operating technology integration and automation is then doing the operating technology integration for control systems. So what we usually do. So just to make it very practical, an example, if you have a chemical process, then our standard automation project or a standard automation project we are doing is build a process control system for that chemical process. So that means installing the PLCs, connecting the PLCs, building the control room, and enables the operator to control that chemical process. So that is to me, automation integration. So if you put basically digitalization on top of on top of that, then you use the data with any kind of insights and analytics to proactively help the operator to set the right parameters, to manage the process in a way that he’s increasing his the efficiency and the productivity of that process proactively. So to me is the automation there, basically establishing the control system them for the process and establish the operational execution of the process, making use of the integrated data to proactively optimize the process. That’s to me the digitalization.

SD: That’s a great definition. And so they are closely related but not the same.

CD: Absolutely. It’s for me, it’s like, I mean, automation is a technology which is existing since decades actually. And I think automation will definitely not lose its relevance going forward. It is still the way to make physical, physical production more productive. I think this is automation and this will remain automation. Digitalization, using the data out of the out of the physical manufacturing processes, contextualize the data to drive value out of the data. This is something which is, from my point of view, improving the productivity of automation. So automation is improving the productivity of physical production and digitalization is improving the productivity of automation. So to me, digitalization is on top of automation to drive automation beyond its limits.

SD: I’ve never heard it framed so clearly, Christian. And that’s such a good separation of the two  because you have this automation, that’s the physical representation of technology, but it’s driving all of this data in the background. And if we can use that to drive continuous improvement, to drive optimization, you’re using that data to provide more ROI out of the physical pieces.

CD: Yeah, exactly Simon. If I, if I can make one comment to that one. To me, we have figured out with our customers what is the big hold up to drive for really customer value in terms of productivity improvements out of digitization. We were talking about the topic number one, which is having all that various equipment being integrated and getting the data out of the out of the yeah, various operating technology layers, getting the data integrated, getting data contextualized, getting the data kind of coordinated in a way that you can make use of that. So that’s one thing which is actually hold up. And the second hold up is closing the loop, you know, of having the data integrated, having the data visualize, having the data analyze, having some insights from the data. But if you stop there, you didn’t drive any single impact. You need to close the loop because you need to drive the insights from the data into the physical manufacturing process. So this is in a visionary set up, maybe in, I don’t know, ten or 15 years. This can be done by artificial intelligence. So that’s artificial intelligence can be kind of visionary, something like autonomous driving. So it could be kind of an autonomous process control. But today I do not think that any customer would allow a complicated and dangerous chemical process or a quality critical of pharmaceutical process being controlled by an artificial intelligence. So you can use artificial intelligence to generate insights, how you can set parameters better, how you can drive productivity. But to find decision to really improve the process needs to be done by a person, by a domain expert who really knows how to interpret the insights from the artificial intelligence and is really doing something in the physical process. And if this step is not done, then all digitalization is in the end nice. But impactless so, and that’s exactly why we PA invest a lot into, number one, getting the different equipment connected. And number two, build the domain knowledge to help our customers, to work with the digital insights and really drive performance improvements in the process. And that’s what we are seeing in the market and that’s basically our differentiator because you find a lot of platforms in the market, you find a lot of artificial intelligence apps and tools in the market, but having the connectivity and driving the impact with the tools you are providing, that is something you hardly see in the market actually, and that’s a differentiator we bring on the table because of our automation integration knowledge and because of our domain knowledge that we really understand the customer process and can work with the customer on what does this insight mean and what do we need to change in the process to basically use that insight to really save money.

SD: Right. And so that’s using the impact to drive action and physical change in the physical world. And that still requires an SME today.

CD: And that still requires an SME today. And that’s where many processes stop to do the final step, to have the SME and say, So what are we going to do now?

SD: I think we’ve been using the word insight a lot, but that’s a very valuable piece of information for those listening is there are platforms, there are ways that this can be done and it will drive information or data or charts, but you still need to interpret that. You still need to take that and translate that into real action to be able to drive the ROI.

CD: Exactly.  And this needs to be done by somebody who is understanding the process, who is understanding the industry, who is even understanding the customer that you can say, well, this are this data, this and this, and insight would require this change and this adjustment in the physical process. And then it’s going to drive value. So and this is important and this is some capabilities. And we have invested actually also in in regards of M&A, we have invested a lot into exactly that step so that we can actively use the artificial intelligence to improve the process, that we can do advanced process control, that we had proactively set parameters with the operators in a way that going forward, the process is running more effectively.


SD: Wow. Is that where you see the primary focus of innovation or long term vision in the digital layer of automation or the digital digitization layer? Where the application of AI, at least in the short term, provides insights quicker.  And in the long term may have the ability to  physically adjust itself.

CD: Exactly. So that’s from my point of view, the big area of automation. And that’s a good thing. You know, if you have a machine learning tools and you combine the machine learning tools with domain experts in all with the domain experts working with the machine learning tools, you know, the machine learning tools build their knowledge. So the more you have the expert interacting with the machine learning tools, you know, the better you build knowledge, the more advanced you are also compared to competitors. And that’s exactly what we think. We think we need the analytics, but at the moment we also need the domain knowledge. So because the analytics cannot do it by themselves, plus I would not say I would say customers are not ready yet to give even for uncritical processes to control to a machine. But if we work with the analytics and if we work with the domain experts and we improve continuously with that into the analytics, at a certain point of time, maybe the customer’s ready and we are ready to.

SD: Yeah, and I think we’ve seen similar transformations even in the last five years of access to data in the cloud. There’s been a number of partners who a decade ago that was a nonstarter of a conversation.) But now we’re actively asking how we can do that for preventative maintenance and, you know, preventative optimization or proactive engagement.

CD: And that’s exactly the point, Simon, it all it comes back to the to the beginning of our conversation. It all starts with customer value. You know, as soon as the customer sees the value, the customer is basically moving away from paradigms. So, I mean, we see this with everybody. I mean, you surf every day in the Internet, you know, and you always have to click, you know, do you allow cookies to you allow your data to be stored? You know, I mean, everybody tends to say yes, yes, yes, yes, yes. Because I’m getting some value of from doing that, from providing my data. So that’s why I do that. And we even see this in the pharma industry, you know, which is the highest productive industry we even see in the pharma industry, you know, that they tend to move their data into the cloud and tend to use some analytics in the cloud. So because they see they can solve mission critical problems with that. I mean, if they have a comparison of process and they have just with benchmarking in some advanced analytics, if they I mean the pharma industry has 30% of their manufacturing costs is cost for bad quality. And if you can improve the cost for bad quality in the pharma industry just by 10%, you might, you know, you saved them millions. So and I mean not speaking about what does a quality error or quality problem in the pharma industry, what does this mean for the end customer? So I mean, this is really mission critical. And if you basically move the needle here by pulling some data in the cloud, then this is a conversation companies tend to be more open to recently because they see, okay, it’s going to provide me a huge, big benefit on other than that, I mean, if you see companies are working with Microsoft Outlook so and have all the kind of critical information in Microsoft Outlook, which is basically also in the cloud as well, I don’t think it’s more critical to have some running manufacturing data in the cloud.

SD: I completely agree with you, but definitely has been a recent change.

CD: Yeah. Yeah. No, absolutely. Absolutely.

SD: We’ve talked a lot even already around it all starts with the data and finding the relevant data. Can you help the listener base again who may be doing this for the first time? It can be very overwhelming  the amount of information that exists out there or the amount of information that you can pull out of automation. But just because you can do something doesn’t mean you should. How do they arrive at what that what is the relevant and right data for them.

CD: And that’s exactly the point where our process experts come into the game. So we do not have digitalization experts setting up the solution for our customers. We have process experts setting up the solutions for our customers. So we usually talk to the operations people and we have operations people talking to the operations people to understand the process, to understand also their pain points. So what is exactly the point where they feel their biggest lever, a lever for improvement is so and then our process experts basically figure out, okay, what kind of data would we need out of that process in order to drive these improvements? If you have an I.T. person setting up this, that the platform and the analytics, then basically you pull all data you can pull and then you see patterns wherever you can find patterns, but you do not work in a targeted way. If you do this from an operations expert, from a process expert, you can be more targeted in putting in the right sensors, pulling out the right data, and putting also on the right analytics to drive really the meaningful insights out of the process and out of the data. And that, again, like we said before, that makes it faster, that makes it cheaper for the customer and it makes it in the end more impactful. So the keyword is having process experts setting up the solution instead of having IT experts setting up the solution.

SD: So that’s an interesting line of talking as well because one of the pain points at least for me, as someone who’s passionate about innovation, I like to see technology applied to problems. You can run into situations where you walk through a facility and you see kind of a robot pushed off into the corner and you start to ask why and where that may have gone wrong. And it’s because invariably it comes back to the process domain and that whomever was implementing that technology didn’t really understand the process and then put technology at it. And if you know, you apply automation to a bad process, all you get is a faster, bad process.

CD: And Simon, this is it’s very funny what you say because this we have exactly been in these situations. We have we have been put in from customers where they said, so a COO, you know, who has taken over a plant and he said, I look at this plant and this plant is fully auto- is fully automated, but I think it’s over automated. So I don’t think we work effectively. We have a lot of investment there. We have lots of capital spending around, but I don’t feel that we are working efficiently. So you guys please help me to make better use of my automation. And then we basically get in and we say, well, how can we set up the process in a way that we use the automation in effective way? And this could happen to a situation that you pull out a robot out of the process and put it in the corner. It could also lead to the situation that we say, Well, here is exactly the bottleneck where some additional automation could help and this is the first step we need to do before we start with any kind of digitalization and before we start with any kind of automation, you know, understand the process, set up the right level of automation. And once you have set up the right level of automation, then use digitalization to continuously improve.

SD: It’s interesting in the separation of two layers, the physical automation and then the information layer that you can treat them as separate. But the unifying piece, you know, the first step is process understanding. And trying to lean out the process and optimize and improve the process.

CD:  But even here, you can you can use data. And that’s actually where we are super successful in the market currently. So it’s called a digital lean assessment. So that that we go into the customers and we because of our automation integration knowledge, because we are able to connect the systems very quickly, because we are able to have within hours, you know, the relevant data, out of the relevant systems.  We do not basically go there and buy our, you know, train knowledge. We observe a process and we see how the process is going and we make some ideas and make some suggestions. We make a data drip, we go in there with our process experts and this process expert, they’ve pulled the data out of the machines. They pull the data out of the manufacturing, they analyze the data and they exactly identify, you know, where is the bottleneck, where is the holdup, where do we need to set up for improvement and where is maybe, you know, if we if we have a process with just step one, two, you know, being on a on a certain level of productivity and then step four is completely over optimal, over automated, and then you have step five and six is again, you know, on a lower level. Then it’s maybe better short term to scale down the productivity that the automation of step number three of step number five was over automated because you basically balance the line in a better way. And that’s exactly what we figure out, you know, by pulling the data out of the lines and out of the machines that we say, you know, where is exactly the overall process improvements or having a holistic view by using all the data and looking on how can we make the overall process better.

SD: That’s a really interesting example of where scaling down one step in the process may actually provide more value when you look at the facility holistically.

CD:  Exactly. So that’s exactly the point. And that’s and that’s also the point why in the end, you know, the integration of the different standards is so necessary to avoid optimizing islands. So and we have seen so many customers who had so many optimized islands in the manufacturing, but holistically, the process was completely inefficient and then go in there and see, you know, how do we connect these optimized islands in a way that we maybe scale up a little bit here, set up and scale down a little bit there and balance it in a better way. So that is usually driving much more value than going into a machine and say, how can we drive the OEE of this particular machine from 80% to 85%? So maybe the machine cannot even handle the volume, which is now already coming to the machine. And we had we did lots of analysis of different machines, of different customers, and we figured out usually the productivity hold up is rarely in the single machine. The productivity hold up is usually in the process feeding the machine or in the process basically transporting the finished goods away from the machine. That’s basically where usually the productivity mainly has to hold up. So if you if you then go in and you continuously optimize and drive the productivity of the machine, you don’t necessarily drive the productivity of the process. And we are applying a few that we say, but let’s look holistically on the process and then basically let’s optimize where the holdup is. And this is usually the connection between the individual steps.

SD: That’s such a succinct way of talking about the system as a whole. And I think I’ve done a lot of speaking on the industry 4.0, the sort of broad topic of this. That’s a very practical definition of why the whole thought process and ideology is important, because you have to look at the whole system in order to optimize the whole system.  Looking at bits and pieces. And you have a little bit of data here and a little bit of data there. It doesn’t it might not help you to optimize bits and pieces.

CD: Exactly. And this is exactly the point. You know, I mean, if you really come in from a holistic process point of view, from a holistic, holistic view on that, then number one, you can figure out, you know, what exactly data you need. So don’t take just the data you are having. I mean, look, from a holistic process point of view on what exactly all the data you are needing, and this could be, you know, that you might need to install some additional sensors, that you might need to install some additional cameras, that you might need to pull some additional data out of particular PLCs that you are able to really analyze the entire process in a substantial way. So and then basically figure out where the bottleneck is and then work, particularly on the bottleneck. So that’s basically the approach we are having. So doing it holistically and really looking from it from a holistic point of view, instead of just jumping on to the obvious because the obvious is usually not where the problem really is.

SD: Right. And it’s great advice. One of the goals of these discussions and the podcast as a whole is to take lessons learned, to take approaches from industry leaders and those that have been doing this for a long time and translate that down into those that may be doing it for the first time. And so if I’m a listener and I’m looking at starting into the automation journey, how do I make sure that I’m taking the right first step to not only the automation and the O.T., but also the digitization and the I.T.

CD: You are mentioning actually a very important points also that one strong lessons learned. We are having is you need to have the operations people and you need to have the IT people being engaged and being involved. So that’s a critical enabler to really drive improvements. So we are talking very often to the operations people. So and the operations people are usually the ones which are having the pain and are usually the ones which are having in the end the benefit and the improvement and the impact. And you can get to quick results with the operations people. But once you start scale, you definitely need to have the I.T. people involved because if you if you talk beyond the single line, if you talk beyond a single plant, then I.T. needs to get involved because you have to work somehow with global I.T. And they have a different view on the topics because they have different criteria, how they evaluate solutions. So having them from the beginning in from the beginning in the loop is essentially necessary so that you implement a solution where I.T. is fine with and it is willing to support to leverage that solution. On the other hand, if you just talked to the I.T people, then usually they are blocked by the operations people. And so because the operations people, they know their plans, they know their pain points, they know basically what they are doing. You know, if they basically get presented the solution from an I.T. department. So the first thing they say, oh, well, it’s not fitting for my plants. So you have you have a certain hesitation on the operations level. So having both parties involved from the very beginning, I think this is super essential. The second thing, which is super essential, is really having that holistic view. The third thing which is super essential is having really the capability to pull the data out of different systems. And the fourth thing is to close the loop really to drive impact, to reflect whatever you gain insights in really in adjustment in the process in real time.


SD: To drive impact to the physical world.

SD: Christian what a great conversation and addition to our podcast, I can’t thank you enough for sharing your insights in the digitization layer and having this discussion for me and extending your expertise to our listener base. Thank you very much for joining us today.

CD: It was great talking to you, Simon. Thank you very much.

SD: To those listening, thank you very much for joining us today for our ninth episode, The Digital Layer of Automation. I hope that you continue to take insights away from our podcast, where we bring experts from around ATS to talk about exciting topics and changing topics inside the world of industrial automation. And the last episode of our second season will be no different, where we zoom in on women in manufacturing and how that supports the enablement of automation. I look forward to the next discussion.