2025 - AVEVA World - San Francisco - AVEVA PI System
Bayer: Driving manufacturing excellence with standardized AVEVA™ PI System™ data and AI
Standardizing AVEVA™ PI System™ data throughout a global manufacturing network can be a major endeavor. At the same time, gathering data for production issues on the plant level is often a process of manual report compilations and time-consuming data contextualization. The presentation shows how Bayer leveraged the opportunity to standardize PI System data across 33 sites globally while creating immediate business value by utilizing OEE (Overall Equipment Effectiveness) as the driving force for standardization. It shows how process data from PI System is contextualized in Shiftconnector to help manufacturing teams drive continuous improvement in batch and continuous operations, from active ingredients manufacturing to filling and packing lines. It also illustrates how collecting and contextualizing performance data from PI System preserves operational knowledge, which then can be accessed through an AI copilot, supporting industrial workers in troubleshooting, root cause analysis, and onboarding the next generation of workforce to the plant.
Industry
Chemicals
Company
Bayer Corporation
Speaker
Kaarthi Govindaraju
Kaarthi Govindaraju is a seasoned technology leader with a passion for driving digital transformation through DevOps, automation, and continuous improvement. With more than 12 years of experience at Bayer Corporation, Kaarthi excels in managing and leading IT projects within the manufacturing sector. He has more than a decade working with PI data and integrations and has demonstrated expertise in project planning and a wide range of technical skills. Prior to joining Bayer, Kaarthi worked as an Consultant to Bayer where he designed, built, and implemented supply chain, MES, and data historian applications. Known for his ability to collaborate with cross-functional teams, Kaarthi consistently delivers successful IT solutions that enhance manufacturing processes.
Session Code
SESS-145
Transcript
Good morning, everyone.
My name is Karti Govindraju. I'm currently working as a unit lead in Bayer for global IT enabling functions.
Our team's main goal is to build applications and roll out applications for operational excellency to improve the production process in Bayer, different divisions in pharma, consumer health, and crop science. And my previous roles, I used to work to develop the MES applications and various applications using the PIE data.
So I have a lot of experience with the PIE before.
So that's where my expertise comes over here. Before I jump into the topic here, I just want to introduce our company, So as you guys know, Bayer is a life science company and a global leader in health care and nutrition.
So we have a mission, health for all, anger for none.
We aim to put an end to hunger and help everyone to lead an healthy life while protecting our ecosystems.
Just a glance of our data from last twenty twenty four.
So as you see, we have three different divisions, crop science, pharmaceuticals, and consumer health. And we are spread across like eighty different countries. And we have like ninety three employees around the globe, ninety three ks employees around the globe. And we quite spend a lot of money in R and D. And we increase every year like one billion dollars for R and D alone, so that we can innovate and help the world there.
And let's talk about our three divisions here. The crop science. The Crop Science is a leading company in the agriculture. We offer a broad portfolio of innovative high value seeds and improved traits and also the biological crop protection and various digital solutions for regenerative agriculture. And in the pharmaceutical divisions, we do a lot of concentrates on the prescription products, and especially for cardiologists and women's healthcare. And also we have a special therapeutics focused on areas of oncology and hematology and ophthalmology. And also in the midterm, we are also on the cell and gene therapy on that one.
So also in the consumer health, which is like our world leader in the supplier of non prescription drug, which is like we have a variety of portfolio comprised of nutritions and cardiovascular risk preventions also. And in the bottom, we have enabling functions. That's where I come from. So we support all the softwares and deployment for all the three divisions here.
Let's go to our journey, the transformation journey.
We did it in our Crop Science division.
So in Crop Science, as you see here, the Crop Science division plants are across all over the world right now. It's spread across in three different continents.
And having different plants, like an active ingredient plants, and like a discrete process with the formulated crop protection plants. So is quite challenging since the environment is so different in every site. So that's a challenge. That's the first challenge actually have in our entire bare production network there, right? And so when we started our journey, the first challenge for our business is to make sure we have a standardized OE.
That's how we calculate our stuff. We want to have a standard OE across our production plants in a different like active ingredients and other plants. We want to make sure a standardized OE. And also, we need to cultivate a culture for the shift team, how to get that OE stuff in an operational excellence way, because everybody is doing their own way, but we want to standardize globally across the Bayer Crop Science plant. So as I talked, the OE stuff, we want to use the time usage model as we move forward. That's one of the challenges we want to do it.
So the challenge for the IT and the OT team is, as you see this map, it's like the production is spread across the different continents, different countries. That means we have a different batch system, different equipment. So it's a very heterogeneous system. That means collecting the data and making a unified structure is another challenge for us, right? So once you don't have that one, then implementing any applications, any things, it's difficult for us to move forward. So this is both a challenge from the business, which was given. And then we have an IT challenges here, right?
Let's go to the next slide here.
So you can see the architecture, the solution architecture, how we build. This is the final architecture. But it took a long time for us. So the foundation for here is the PI System. Even though the PI System is available in different legacy Bayer systems, Bayer plants there, the problem it is across in different plants. I think I remember like eight years back or nine years back when we started a journey in crop science, everybody was talking about having a separate PIE system for each plant.
But at one point, we decided to have a one PIE system for the entire bare crops division to collect the data and to make sure it is unified in the structure so that it can be used for different application, for different data analytics also there. And then we have a shift connector, which is used for calculating our OEE stuff, and also to make sure they're also using it for shifter while landing over the shift so that they can handle the events and actions. So they are cultivating the culture. But initially, these two systems are separate apart.
That means we are building our foundation by one team. And then the Shift Connector implementation is going on to the different sites. The reason for that one is so we are trying to build that culture of OE to make sure we have a uniform OE culture across the plants, across the different countries. So that's the when we started our journey, that's how we started it.
But as we progressed it, once we had the uniform structure on our foundation system in VivaPi, and then once we have the shift connected, we started connecting to both the system together. That's how we got the final solution. Now the journey has not stopped yet. Still continuing it.
We are actually building our foundation. Still, I talked to the big team that's over like to the internal team. They're still adding more plants into the pig system right now. They didn't stop it.
The same with the Shift Connector. We are rolling out to the most sites. So we're building that culture across all the bare crop science sites here.
And let's take a quick look into our pig system. This is the one production system you have.
So we collect all the data from the different production plants here. And you can see there is a one DA server and then one AF server. And you can see how many tags and how many events. This data is like a very latest data.
We are collecting eighty million event frames. And then we have six ninety five ks tags, which is sitting in our PIX system right now. So we collect various different datas. We collect a process data, which is a normal data from your PIE data from PIE things.
And also, we collect batch data, which is also important, which gives the batch ID, production information, material information also there. And then events and alarms. And then on top of that, we will collect our quality data from our LIMS also. So that's a wide variety in one system.
So there is a big team working in Bayer to have this maintaining this entire system. And then we have a lot of implemented partners also working with us to make sure this system is running twenty four bars, seven year.
And then let me go back to one more.
Let's talk about very briefly how we structured our data, batch data, in a PI system right now. When we started it, we started with our raw batch data, which is coming from different batch systems like Symantec and Delta V systems. That's how we collected into our AF structure there. It's a raw data from event frames generated there.
And then the next step for us is to make sure we are organizing ourselves all of our asset frames and to make sure we are standardizing that one. That is the second step we did, that one. And the third step we did was we organized all of our batch data, especially the batch process production things. We organized in the ISA eighty eight model that this system is even though the dates looks like twenty twenty one, but this is like adding more and more together.
So this is like a overview of our batch system and AVEVA PI System, which is the foundation over there.
Let's get back to our story here, our digital transformation, especially for the operational excellence for the manufacturing team. As I said, the first step is backbone. Backbone was built. And the second step is the Shift Connector, which creates the context on the shifts.
And then the third one is the insights, which gives all the reports. And the fourth one is the intelligence. So this is like a full stack right now. Right now, it's a full stack.
The AI stuff is we implemented in one of the sites in Mutant in Switzerland. But that's also in work in progress. We are making sure that we will take it to the next step further on that one.
The important piece on this full stack is, as you know, we collect all the data, sometimes in paper, sometimes in Excel sheets.
So many things, we kept it in different ways. If there is a problem, so we all tend to spend a lot of time looking at the papers, looking at the spreadsheet to find the root cause, what happened on the plant, right?
But the good thing here is they have everything like a pie data, the equipment data, the time series data, and batch data, plus your shift events and shift actions are all in one place right now. That means it saves your time.
Actually, when we talked to one of our site colleagues, they said it actually saves two days of a process engineer's time there in the site because they don't need to scramble for all the spreadsheet and everything. I think two days in a process engineer's time, that means we can spend the two days to improve our production process and also to make sure that their time is utilized for the different production improvement process also there.
Let's look into the visualization, how we are looking into this one.
If you look into batch systems, we use normally in a control room.
Like you can see, we all monitor in a control room because the losses in a batch system may be acting one or two losses per shift.
But when you talk about the discrete, that's where a lot of minor stops, major stops is happening.
But they are looked into the small industrial PC. And also, they can look into the bigger monitors on the plant floor also there. And then you can see here different visualizations here.
If you have a batch, that's where we organized ourselves into ISA eighty eight models so you can see all the event frames there. And Shift Connectors provides that visualization, but it gets all the batch data and the event frames from the pie.
And the same with the continuous batch processing. You can see that our tag data is coming from the pie AF there. That means we can see the AF structures and AF data in the continuous process. For discrete, we get the data like our good parts, bad parts, any error codes. We get it from PIE. That's also displayed over here.
Let's look into our new I think I talked to few of the colleagues in Mutant's site. They are really excited about this product, about the Shift Connector AI stuff, what they are trying to use it in the Shift Connector here.
What they are trying to do here is it's AI this product is a part of the Shift Connector tool.
Since, as I said previously, we are collecting a lot of data from PIE. And then we're also collecting all the events and actions from the Shift Connector as a context data. So that means the people on the plant floor doesn't need to scramble for root cause analysis. You have everything.
So this tool, this AI module in Shift Connector, really helps the people to look into root cause analysis. Some of the operators on the plant flows are decayed. They have been like twenty years, twenty five years on the plant. They know in their mind what they are trying to do the root causes, right?
But the new beginners, they don't know what they are doing it, right? But we need to have a one stable place where they can find the root cause analysis, right? So that's the story here. If one person says it's a residues and the other operators says the color variation, but when they start using that with the chat, it's more like a chatting system with the AI stuff.
But it gives the same results, what would be the root cause analysis, what would be the steps they need to take care of to solve this problem here.
So this tool is like it's in one side, but we are trying to make it to the other side. You can see, like one of the supervisors says, it saves a lot of time because people doesn't need to go and find data from ShiftConnector or PIE to do analysis here.
And I'm not going to go through everything, but the main important thing is, as I said, we have everything in one place.
That saves a lot of time for process engineers.
One of the plants in Dormagen, Germany, they said they because of this full stack systems, they actually improved at least ten percent of their improvements. They've done it through the site because they can find the anomalies in the batch system and because why the production has stopped. They can see all the reason tree quotes. All these events and everything put together, that's actually they increased their improvement by ten percent. That's one of the straight statement came from one of our plants in Darmenheim, Germany.
