Kellogg's improves quality with sensor-based data

Guests: Emelio Anglés, Information Technology Engineer, Kellogg Company
Ted Combs, Industry Principal for Food, Beverage, and Consumer Packaged Goods (CPG), AVEVA


Information technology engineer Emelio Anglés of Kellogg Company is part of the Smart Factory project to digitalize all of Kellogg's manufacturing plants. He describes the different stages of this multi-year project to move from siloed PLCs using 5 different protocols to a single source of all data using a single protocol. Among the benefits he describes are a 50% reduction in mean time between failure on packing lines, a reduction of energy use over 15 years, and shortening production meetings from 45 minutes to as little as 15 minutes. Kellogg's can compare the efficiency of cereal plants from around the world using a single source of data.

Subscribe to our biweekly newsletter. Visit our website. Follow us on LinkedIn.

Transcript

0:09

Well welcome everybody to Radio PI. I'm Nick Dorazio. I'm a media producer here at OSIsoft, which is now part of AVEVA, if you haven't heard. You know, a topic of great interest in operations is the the cooperation, the collaboration between IT, information technology, and OT, operations technology. And we have a guest here who spent 25 years in--his training is in information technology, his educational background, and he spent 25 years in operations doing projects, major projects, where he managed large things like the installations of compressors, and entire packaging lines. And he is our guest today. So we're really interested in talking to him about that collaboration. So welcome today, Emilio Angles. He is a an information technology engineer at Kellogg. Good morning, Emilio. Morning, and thank you very much for inviting me. Great, thanks for being here. Also, today, co hosting is Ted Combs. Ted is the industry principal in food beverage and CPG here, hello, Ted. Hello, thank you for having me. All right. And let's get started by--if you could tell us Emilio tell us a little bit about Kellogg and what your job is. So, well and I am responsible for [inaudible] that Kellogg Manufacturing Valls in Spain. So adopts the right technologies in order to achieve the business goals that we have in safety, food safety, delivery cost and continuous improvement. Basically, keeping our operation technologies reliable and cost effective. My background is in IT engineering Information Technology I am an Information Technology engineer. In 1999 when I started in Kellogg's I started in the in the corporate it working in different IT projects related with financial marketing area of [inaudible]. And later on in 2005, I was moved to operations to engineering department and currently I am the manager of a department that we call here in in Kellogg's power control and Information Systems and we are a team of six people plus some contractors. Emilio Can you tell us how you made the move from IT into operations, I was part of the team that implement and install on OSIsoft PI in Kellogg's Manufacturing Valls. In order to count cases that we produce in our packing lines and send this information to SAP. So, after this project, the company decided that they needed someone in the in the in the plant in order to absorb the outrage of system in the in the plant and I started in the engineering department in manufacturing. So and since then I've been working there and now the last 15 years mainly apart from doing different improvements in the in all the network on the control PLCs and all this stuff needs work okay. And the front end project in order to create all PLC is that we have in that that we had in the in the plan. So, now I am working integrating the two worlds The it with what is called a patron technologies IoT. So on and also now I am part of what we call as mud factory project, which is a project that we are implementing in Europe and offshore around the world in order to digitalize all our manufacturing plant. So how do you think your background in it has helped as you're working on these these very engineering heavy ot projects it makes me my life easier here in the in the plan because during the last 15 years or so we have been upgrading or altering structure. In terms of PLCs connections, we're now we have a big deployment of optical fiber in the plan, okay. And all this connectivity has been done along with over the IT department in order to to create a solid infrastructure that allow us to have a reliable network in order to be able to access to all our wide PLC network that we have Here on access the data that now escape in order to implement our smart factory project, that at the end is digitalization of manufacturing Blanche, Amelie, that's really interesting when you talk about the move from the IT side over to the OT side, and in that discussion of creating a data infrastructure on the OT side, and I'd be really curious to know, once you are building that data infrastructure, was it something that started kind of, from the ground up? Or were you borrowing things from the IT side? Did you, for example, just work with the Master Data Management and the Master Data Management governance that was already in place? Can you maybe Are there any insights that you can share there? Let's say that, when I joined the engineering department, I found a plant that it was what is called now industry, three digital, sure, it was just aisle of automation, I found a lot of PLC is what that was just alarm, nylon shell and then what we did was to start a project, we did a project plan of five years, in order to invest in different areas of the plan, and start to first of all, great PLCs great, great connect on once we have to get the connectivity, we connect to our data infrastructure that in this case was OSIsoft PI. But during during this this day's show, I remember that we had almost five different protocols in the plant. And we had five different autosurf PI Interface in order to connect to connect to all this data. So and then when we in step by step, so as I said, during this five, six year plan, show, we were replacing the old PLC is connecting. And, and in this project, during this project, we take advantage of the project in order to remove the old we have communication boozers, okay, so once we have all the plan with DC, TCP and IP connectivity, then we did a project in 2013. In order to replace all the external, the different communication interface for just one communication interface, the one that we have right now, which is OPC DA using keep workshop where we use keep server, we just have one communication protocol. At the end, we collect all our data with just one page of interface, which is OCS OPC da, and I'll show the last year that we have installed a new PLCs We are ready, we are already connecting using OPC ua during this process the to to have a good relationship between our IT department corporate IT department and on the OT side, it was very bad without it because it allows us to to create understand the network that allows allow us to escalate quickly and now we have even we have a system that allow us to implement some cybersecurity stuff around over plant because we have a good design and the design has been done between OT and it not just ot ot or so, we we have been working together during this this year. Probably because I would my background was from it and I had some contacts there. Sure. I know very well what are the next steps that I have to do having that data available to you what kind of results have you been able to get Is there anything you could point to as a Yeah, you have improvement from having that data? Yeah, for instance. So we are using this data as I said before, for for the outdoor maintenance project and unfolding. And one of the KPIs that we measure daily basis on our project basis is the mean time between failures. On for instance, we have increase over mean time between failures in from buggy lines. In more than 2030, and even 50%, now we have data in real time, and we are able to, to see which kind of action we have to do in the lines in order to improve this. In this case, this mtbf. So you're expanding this to multiple plants, and you have this visibility, and you're doing some modeling, and it's opening up communications, or is there been a significant change in almost how you conduct the business or approach the business? You know, for example, maybe somebody in a different country is having a problem. And now by virtue of the fact that everybody's approaching it the same way has the same tools and practices, GIS? Is there kind of an improvement in collaboration and communication and problem solving? Yes, in fact, Shaw, I've been named data Shama matter expert for faceoff PI here in Europe. And I am helping the other serial plant in this case in order to to standardize to put the information in the in the in the right in a writer manner in order to, to have in the all the plant, okay, the same show of true show, that if we if we check the mtbf for one burger show, this burger can be compared. So, now, so we are able to compare the Spanish baggers with this with the English Bible, you know, the meaning started there, because we have data that is our that we are measuring the same data. And you are we are using the same kind of Yes, sort of information. Okay. Well, that's great. That's the perennial story. We've heard that over and over again, Pat Kennedy, our founder and president likes to tell the story of how well you know a lot of times your morning shift meeting will consist of two people bringing two sets of data and then you know, spending two thirds of the meeting arguing about whose set of data is right, have you experienced any of that where one source of the truth fixes that? Yeah, yeah, sure, this is this is what So, for instance, in every morning as all the manufacturing plant in the world, I think, so, we do the overproduction meeting and now that 70% of the information that is coming from from the in the in the pollution meeting is coming from our parking line analytic system on all the people of the different areas and I'll show the maintenance people show you is using this information on and this is very, very helpful because we can concentrate not analyzing the past Okay, we can focus in depression and also in the future, because we have real data all the people is using the same three as you said before social through so on, on on on for instance, in position meeting that I remember maybe a pollution meeting could take usually 45 minutes, one hour. Now, daily, sometimes the pollution meeting that is 30 minutes, sometimes we finish in less than 15 minutes. So, these are important addition, because we focus in the present the future on basically because there are no discussions about the the consistency of the data that we are that we are dealing in the production meeting. So I'm curious Have you been able to reduce energy usage just this increased visibility Have you had any any like easy wins like that just because people have better visibility? Yes, or in fact, so we started a project in 2005 where we installer will deploy a lot of electrical and gas compressor estimated around the plant. So we connect all these matters to our choice of PI, okay. So and then we start to measure where we are spending more energy okay. And during this year, we have done different engineering project in order to to reduce our energy consumption. And, in fact, we measure the energy We say we use a KPI which is kalo but not on produce okay. So, you have to take into account and then in 2005 we had an energy usage of for every turn we spend, let me remember if I remember well is 1265 kilowatts okay. And right now, for instance last week our usage was that less than 1000 kilowatt baritone, okay. So, obviously, we have done different I should have said before engineer project, but, and currently we have a nice dice energy time water we called energy dashboards in in that is made with PI Vision or PI Processbook Okay, that Allah wash to control which are out of the plant in real-time are in, in production and energy consumption is is okay, or which ones are estop for instance, for maintenance for cleaning for any reason. On in that area, we have equipment that is that is started and is commissioned is consuming energy that is not producing. And we have to adapt some kind of alarm, okay, in the in the in the process book that allows the operation people to in this case is topic why men that I'm not in production, okay? It seems something a little bit silly, okay. But at the end at the end of the year, it helps us to that, to have information that is available for the not for the management for the operator, the operator that is in the line, it allows him to to make actions in order to reduce energy. And at the end, at the end, it is more things because the big project, it's a big project, okay, we are going to change a boiler that is more efficient. For instance, this is a bad engineering project. But there are a lot of more things that could have happened in the plant like okay, I go there, I have to clean some equipment I started so the equipment is not in production, I started equipment. Okay, clean something, but I will I finished my shift. And I forget to stop the appointment. Okay, this kind of things happen in the bath in a minimum in a manufacturing plant. And sometimes a woman is there running for hours and hours now with this system allow us that is when when did I finish the shift? The people that came in the shift after they are able to see if they are consuming or not an add in in energy in India, depending on the area of the plan. So, okay, well, I think it must be a rule of human nature once you make something more visible. Because we see this all the time when people have better visibility into the data, facilities, especially anybody who can control a thermostat all of a sudden people just respond better. So. So quick question about notifications, and email. Do you implement any kind of email notifications where people can be informed of, you know, automatically if there's a problem that you've defined? Yes, so we are able to send a notification to where maintenance people when the meal has run for intervention more than 10,000 hours. Or, for instance, when the mill has done more than 1 million of cycles, we tend to notification to the maintenance people because depending on the number of RPMs they have to do some preventive maintenance, damage, and other use cases. For instead we have a big turnover of raw material. So then, we set up an analysis in our PI if we have up if we set up the notifications, and now a daily basis, every 12 hours we send to our raw material supplier in this case for technium. chocolate cream, shall we send daily notification informing them on it allow us to have a better inventory control. Well, you can you just describe a little bit of the approach you took with the modeling for the packaging lines? Yes, of course. So when we started with the with this application that is mainly based in in SAP. Shawn. We found that there were some calculations that initially, we did using a typical Performance Equation that has a PI OSIsoft phase of PI. So and we found that it was And that we were not able to, to have to have their currency in the calculations, for instance, the status of the line or for instance, they stopped region of the line, or the operation of one machine, or the status of one machine. Okay. So then goes, we have to take into account different PI tax, okay, and different conditions in the line show. And then we ask this Punisher OSIsoft, guys, because because we have our data support from here, they recommend that to use PI AF, and analysis in order to do this kind of complex algorithms in order to calculate the state tool of the state to run the test operation, shown on in order to do that, so we create a model of our packing lines using PI if and need help a lot, in order to have now a very accurate information on and I'll show a contrast and information. So it helped a lot for the modeling. And also now it's very easy, because when we started the project, we just have six parking lines in March. So and during the way the parking lot in analytic depreciate, so we started a new parking line, which one then, in order to set up the parking, the new parking line in the system in the backing line on latex shares was very easy, because it's just, we created, we use the PI AF template in order to just create line borrar cartoner, or whatever requirement that we have in the line just in one minute. Because it's we have a good model of the of the line of a packing line, I'm not sure of the different machines that we have, or different equipment that are involved in the packing line. So essentially, you're able to create, refine the model to make it just like you want it and then once you wanted to replicate that it was as simple as copying that template over pasting and going. So very efficient and effect. Nice. Thank you. That's helpful. Okay. Well, that's, that's it, we're at the end of our hour. I just wanted to ask you some quick, quick lightning round questions, because I know, you know, we're in operations, right? There's some things in operations that only people in operation see people it kills me that people think everything's just smokestacks, and you know, waste treatment, right? What's the prettiest view on any of the sites you've ever been at, at at Kellogg, I imagine you've been to a lot of forums instance in in a poser plan. So when we have a big box of powder of different show, when sometimes the big bucket, it goes down in the in the floor, and it's brought, okay, and then the operator gets a lot with a lot of power in there in de cloud and Shawn, and is completely white, for instance. So I've told me I saw this DC, my car in there, or even covered with chocolate, you know, Chuck, let's go ahead with all the all the all the all the all the glows with dark chocolate, that's pretty cool. Things happen in a food company. Okay, so especially since you've got an IT background, what's the what's the coolest spreadsheet or calculation or even a bit of code that you've ever done in your career? What's the neatest thing you've done? Is that difficult question because for sure, I have some examples. But for me, I'm one of the calculation calculations that I am more proud of that is the answer an Excel spreadsheet that we have, in order to calculate as if basis, the raw material control and consumption from our extrusion process, you have to take into account that extrusion process is the 40% of overproduction. And then to have in fact, this chief basis, the raw material consumption, and this is key is key for us in order to control the cost, Shawn and I remember that when we did this, this Excel spreadsheet that behind that there are some some mygovid Visual Basic causes gold as well. And so I saw for me is the one of the things that are even more satisfied. And that's nice. That's nice, sir now, since you come from an IT background I'm just curious and you've gone into ot. Are you? Like most engineers, you know, chemical, mechanical, electrical engineers and ot. Do you have some kind of a broken piece of equipment or maybe a fried motherboard or something that you keep on your desk? as a memento of something that went wrong at one point? Yeah, to fuck off a PLC. So no PLC that we had a problem an electrical problem, but one barn. Okay, got it. Well, I've really appreciate your time. Thank you so much for sitting down with us. Again, we've been talking to Emilio angles from Kellogg. Thanks so much for joining us. You're welcome show. It has been a pleasure. And thank you very much for my mean to this podcast. Well, thank you Ted combs for joining us as well. It's my pleasure. Thank you. Okay. Well, we'll see y'all in two weeks. Thanks again for joining us. Bye bye.