2- Juliano de Goes Arantes, Rockwell Automation (Switzerland) & Danijel Koren, NEXE dd (Croatia)

we're gonna move on now to our second presentation. we're gonna, it's a great pleasure actually to be able to introduce Giuliano, a long-term supporter of Cemtech from Rockwell Automation. Giuliano is digital account executive responsible for artificial intelligence and machine learning. at at Rockwell in the EMEA region, he has over 20 years experience in advanced process control solutions across mining, mineral processing, and the cement sector speaks Portuguese, English, German, and Spanish. But today English, I hope.

and, and later we're gonna introduce Daniel who is from Nex e cement in Croatia and responsible for all plant production, quarry product dispatch, operations, and plant maintenance. prior to that, he held a number of roles at Naiche Cement including head of production and head of Maintenance. So, over to you, Giuliano. Thanks, Thomas. I'm very happy to, to be presenting at the Cemtech again here in Istanbul. And last time I was here was nine 18 or so now. Yeah, very happy to see a lot of unknown face meet new people as well. Very, very honored to, to be sharing the stage with you. Thanks. So this is the agenda today.

I'll be giving some introduction of a recent project we have with Nexus event. Thanks, Daniel, for the, the introduction, so to warm them up. So what we are talking about, we are talking about the machine learning, artificial intelligence, digital transformations, all this buzzy words, right? this is not my number. So this is coming from from Gartner, from PTC. So what we want to understand is there's all this information, the digitalization, digital transformation, the data can allow you to do something with that. So that is the, the, the beauty, right? So we want to have, I I, I, I said that before, if you, if you guys saw me presenting, you cannot control what you cannot measure.

So first of all, you need to be able to measure something somehow. So it's a physical equipment, or it's a virtual equipment, it doesn't matter. You need to be able to measure. And then with the data, you can do something with that, right? If you're doing a backward calculation or trying to build up a modest rate of historical data, if you're doing step tests, that's another story. But that is the concept. So we want to take advantage of this data. We are having ton of data that is just storing there, and you cannot profit from this. But that is the, the concept. So what we want, we want to enable our clients, so help our customers to bring this data from the plant floor, right?

But in the contextualized way. So where you can really start to get the benefit out of this data. So we got data from the machines, from the, the, the various process control system. You might have on, on site. You might not have only one. You have multiple ones. You have lab, you have a manual input, you have ERP, you have many data and a lot of things you cannot measure as well. I've been working with, I worked in cement business since 96, so I have seen many things until now, and it's difficult, and I think forever it's gonna be difficult. but what can we do to mitigate these issues? So we need information. If we have information, we can do supply chain optimization.

So all these things, but first we need data. Then this data go up. Then we start to have dashboards, understanding measure and control, control optimize. That is the concept. When we have all this data together, then we can start to be more creative step by step. So go predictive analytics. we can do augmented reality, workforce enablement, so training, all these things. But it comes from the base, from the data first. So I'm part of the, the digital business inside of Rockwell. So I'm from the data science and AI team. I'm working, I would say 90% of my time with a model, predictive predictive control. But we have other colleagues inside of our team that are doing other things.

So predictive maintenance we have straight consultancy people agnostic technology. So we can do with say any kind more or less any kind of technology. And also manage services. So that's where we are with our partners. So we have partners with Cisco, Microsoft, PTC. So that's we created a factory toque innovation suite. So where we can bundle all those technologies together and start to build up applications for our customers. So that is we have in many industries, many verticals. So again, we bring up this, this data to, to one single location that could be on premise, can be on the cloud. So that depends off the, the customer of, let's say preference.

And then with the da, the data available can start to build up your applications according your needs. We believe in the open platform, scalable, step by step. And in the end of the day, it doesn't matter which supplier application belongs to you. You are the end user. You need to own the application. You need to understand what is inside. If it's a black box it's just a matter of time that it is gonna stop to work. And then it will be on the side. So you need to have ownership. So customer is touching the application. There is a special feeling. It's almost psychological. So there is, you have this attachment with the application. So you need to own the application.

Doesn't matter who is supplying, and this is what we believe, and this is what we are delivering. talking about the, the, the data science, the, the advanced process control team. This is, we are a global company. We operate in various regions. we have more than 30 years experience. And yeah, we cover many industries many verticals. Cement is not the biggest let's say vertical from, from Rockwell regarding advanced process control. but yeah, we are growing the business. And yeah, I think this is, this is good to have also a two key software solution that also works in other industries that gives you a lot of flexibility. some of the logos.

So here you can see various non, non logos from, from cement, mining, metals, pop and paper, ethanol, chemical, c, p, G. So can you imagine that if this kind of solution works in other industries, it may be even more complex than cement. No disrespect to the cement process, but you, it can have a very complex special application so that this, this solution works. And as I said, we are a global company, but we are divided in the regions. So that is a time to reaction from, from our team. It's a much faster, so you don't have all eggs in one basket. So we have people around Europe, around Middle East, we have people in Asia we have in in America. So we all, all well spread.

So if if a global customer wants to have a 24 7 support, we can offer that as well. solution overview. So we'll this today. So I will give him this introduction that will pass the, the stage to, to Daniel that we are gonna talk about project we are doing together. So we have one of the solutions we are deploying at the next as a model, predictive control. So this is well known. So deals with constraints, control and optimization. And the way we do, we also do, we utilize machine learning to build up those models. So we will take historical data, we'll do step test activities on site whenever it's possible.

And we also collect the process knowledge because according to my experience, it's not possible to do full application of cement kiln only with historical data. So that is the, the what I have seen. So you, you need to compliment because you don't have measurement. It's the same, like you say, oh, a scda can do autonomous driving, so Tesla can, because all the sensors and radar cameras and so on. So it's not about technology, it's about the condition, right? And what we have now, hopefully, hopefully this is gonna change in the, in the near future, we will have more information, more capability to do everything fully, let's say autonomous, right? But right now we have to mix things up.

So we do a little bit of using historical data. We do a little bit using step test data, and sometimes it's not possible. Then what you do, you get the process knowledge. So we combine the best of all the words, and we put in the model. And this model will control and optimize your plant. So we are do doing a k control and optimization meals raw mix preparation. So the, the majority of the areas in the cement plant, we have control and optimization from the data point of view, as I said. So we, we have a spar, we have data and different sources. So you need to bring this data in a single location where then you can start to build up. So your IOT application.

So bring all the data together, otherwise, just the data, no, no added value. So you need to bring data value to the data. So contextualize those those data, and then based on the data, and then you can do different mashups. So oes, KPIs drill down. So as I said, you need, you cannot control what you cannot measure. So here we are gonna help you to understand better how the plant is operating, knowing why things are happening. And then if you know, then you can take actions.

So the concept we have just to illustrate a little bit, so we have this our factory talk innovation suite powered by PTC, where we have thingworks collecting data from from different sources, putting this data together, data treatment, everything. And then based on this data, we will make dashboards, KPIs. That is the first step. You can have many, many more things. You can use ai, you can do anomaly detection, condition monitoring a lot of things including augmented reality on top of this platform. But that is the first step. So we are, we're doing like baby steps, because it, it, it is a vision, it's a journey. It takes time, but you need to do the first step.

And this is the, I'm very honored to have the chance to, to walk together with next in this, in this vision. And we were gonna have a lot of good things. I hope next year we can, we can share a bit more information with all the final results. but we, this is the beginning of the, the journey together. A little bit more information about the, the OE. So, as I said, get all the information. this is no news for you guys. so get information from different source, the lab and so on to calculate how, how the plant is operating. Then I hand over to Daniel. Thank you very much. So thank you Juliana, for this introduction. good afternoon to you all.

So I already met some friends over here and some technology providers that we are working with, but for those of you who are not familiar with Next group just a few information for, for a startup. So next group is basically business system that employs almost one 1800 employees in 15 companies in Republic of Croatia, Serbia, and Bosnian heads growing. We are regional producer of building materials, so starting from cement, concrete, concrete elements, bricks and roof tiles. We also operate Riverport. That is our gate for getting materials from Western Europe and also from Black Sea.

and also we are managing we have a company for waste management that we can use to optimize our alternative fuels supply. So three count countries, 15 companies, and like I said, a little bit less than 1800 employees. So our vision is basically to be the leading producer in the region of contract construction materials, and also to be recognized by all partners and social communi community that we are working in. Our mission is to build better future by invest, investing in stable and growing markets, of course, with overview in sustainable operations. So how we plan to do it, we build up our strategy for period of two, 2022 to 2030.

And it is based on four strategic pillars that will create some additional value for us. these are market orientation, implementation of new technologies, operation excellence, and of course development of our employees and organization. market orientation for us is especially crucial because we are operating in Euro in European Union, where we are impacted with ETS system and very close to countries that are non ETS. So cross border mechanism, we are some, this is something that we are expecting very, very eagerly. this strategy also includes projects that really enable us to go to energy transit transition, and also green transition.

and energy transition basically implies projects for reducing energy dependence of all of our plants by introducing plants that we can use for renewable sources. While green transition implies a reduction reduction of CO2 emissions by more than 50% in 2030 30. So where we are Daniel also mentioned in, in his presentation that having data is the crucial thing. So we actually have modern DS system, distributed control system that is capable to collect all the data coming, coming out from process. It is basically 1,500 analog values recorded in ten second period, and we are producing something about one terabyte of data per year.

this system actually covers standards start and stop sequences, interlocking equipment, protection devices, all alarms operate directions are also recorded. We have some, something about 100 PAD control loops, closed control loops, and of course, digital surveillance system that we can use to oversight our operations in the plant. from the reporting side, we, we are using system that we were developed by our own. So it is basically Microsoft reporting services where we can show process data, laboratory data. We have all data coming out from continuous submission and monitoring system, electrical energy and everything that is necessary to produce ESG report at nowadays.

So what was the project scope for this project that we are doing with Rockwell Automation at the moment? So we started to search in, to look into data. One terabyte of this data is really, really big. And we decided to start small steps like juliano already said. so from all the field equipment for our project, we choose to start with, with raw mill kil and cement mills, because we recognize those for faster return, fastest return of investment in the project. So we can get all the data for these field equipment through our PLCs. and re report it to, to, that is basically OPC server that is capable to, to acquire data from different sources.

So in our plant, we are using Alan Bradley from 1990s, but we have a base number of other, other systems. We have laboratory information management system that communicates from to CCSV files, excel files, and so on. So we will be able to do this conversion by using thingworks, and also we have a connection with our enterprise resource planning system. It is Microsoft Navision at this moment which we can use to get actual prices and stock values coming out from, from office or plant level. we can use this data to optimize our strategies, control strategies, not only to optimize process, but to optimize strategies. How to run our, our business.

thingworks is basically system, it's a platform that collects data that incorporates Iza 95 equipment hierarchy. so you can make it really visible to, to everybody who can, who who will use it. we can calculate KPIs and we can have different roles on, on the system. So for the operators, you will have one set of data, which they will use for daily operation. Leadership will probably use data for some other purposes, but you can do all of this on, on thing on thingworks platform. so why we used Roku, why we choose Roku software for our project. after we signed the NDA agreement we sent to them this history of one terabyte data. And data scientists basically did some investigation on, on data.

And only by looking into data, without knowing the process, they were able to recognize some crucial things that were bothering us and we were not maybe fully aware of it. So when we look into, into these pictures, I'm not sure if this is working, nevermind pyro process application, you can recognize this is basically line free line content in, in clinker. So data scientists were able to recognize, ah, there is something wrong because you are running different set points and different targets without any, any reason that data scientists can recognize. So we included guys from process from Rockwell also, and our process engineers to investigate a little bit truly into it.

And they found that these discrepancies came mostly from alternative fuels, because one time you are running with alternative fuel of one type. The other time, other time some others. And data scientists was able to, to tell us what we are actually doing wrong and what is possible to, to control at, at the moment. operators always run on the safe side, so they're using more energy than it is necessary to have a good quality of clinker. So if we keep low lowest possible energy to, to achieve good quality of clinker, we will for sure have some energy efficiency increase, also CO2 level reduction and the plant will for sure run more smoothly.

Rockwell could recognize at least 4% gain of energy efficiency only, only on this, on this part. on the next picture you can recognize, recognize maybe it's a little bit hard, but this is basically secondary and tertiary temperature of clinical cooler coming out into, into the process. So what data scientists recognized also is that when there are times when we are running with let's say 2.5 strokes of great cooler, we are getting the highest possible temperature. This is of course, again, energy efficiency increase.

so if we are running in this region and we, we are able to have some smaller standard deviation or some smaller fluctuation of measured data, we will for sure have another energy efficiency increase on raw mill throughput. what we recognized just looking into data, you can see these histogram with high bars. We recognize basically that our operators are running raw mill in steps of five tons per hour. So anybody, anybody who speaks with you will tell you this is for sure not opt optimal way to do it if you will recalculate all the time. If you will change set points more, more frequently and close to your targets, then for sure you will get some throughput increase.

This is, especially for us very important because we, we are limited or raw mill is basically our bottleneck in, in production. So if we will get some raw mill extra, then we could go higher, higher with the production also. So for the future we plan to do this project only on this three parts that, that I mentioned, raw milk kiln, and one of the cement mills. One of big parts of, of the project is training of our implant engineers and also knowledge transfer, how to do it by yourself. So our plant our engineers, our software engineers will be able to deploy this knowledge to all other plants. So we will cover everything from query to all of cement cement mills later on by ourselves.

Thingworks is basically industrial iot platform that behind has analysis tools that will help us also to investigate other plant process bottlenecks. it'll enable us to predict maybe some maintenance issues. Some are here already asked about maintenance. Yes, sensors with fast analysis can be connected to iot platform and give us some prediction of, of maintenance needs for, for the future after we will have this system running. We also plan to use it for all other companies in, in our group. So we will scale it out in, in the future. At the moment, we are starting with, with the system installed on premise.

So it'll be running on our service or our equipment, but in the future, when we would like to scale it out for other companies, like I said, we are operating in free countries. So we are looking into cloud solutions that Thingworks is also capable for. Thingworks also can do augmented reality. So we plan, we plan to install Googles so we can use it for augmented reality for our mechanics to, to go into operations and try to solve the problems very, very easily with the guidance coming out from the system with the standard operating procedures coming out of these systems. And also, of course, further CO2 reduction energy efficiency all the time.

And we are putting also our hopes in our big future project called esa, that will, will reduce our CO2 reductions on scale of zero in the future. So how to do it just in short phase one scope we will first produce minimum valuable product that will incorporate all measurements, formal formulas based analytics, high level site architecture, and including global principles. Later on, we will finalize all the screens. We will agree about plant specific consumption, customization rules, and later on, roll it out to all of our plants plants around the region.

So thank you very much for your attention and I'm really looking forward for some next Cemtech conference where we will be able to show our results. Thank you very much. Thank you. Thank you, Daniel.