Customer Showcase: Reproplast

00:12
Speaker 1
Next up we have from Reproplast we have Michael Phillips, factory engineer and Damien Dearuda from Afrilec. Guys, can you join us on the stage to share your story? Thank you. All right, so again it is always a bit daunting being on the stage, especially if you’re not used to it. But we kind of insist, I think we want to get people to tell their stories and you know we’re not professional presenters but we want to make sure that people tell you what it was like and what it was been like. So it is quite important. So maybe I can give you the little intro. So Reproplast is a leading force in circular economy for poly-elephants, polyphenols, polyolefins exclusively. Yes, big industries. So they would be maybe I’m trying to understand there would be kind of ice cream tin technology, plastic tubs, buckets.

01:18
Speaker 1
All right, so you’ve integrated Michael in a digital solution that kind of streamlines your production processes. In doing so you have gained a deeper insight into operational performance with data driven approach helping you track production metrics and optimize resource management. So that’s the little intro and I think all we’re going to see today speaks exactly to that data driven element which is quite different. So Michael, thank you very much. Welcome. Thank you for joining us. You are out from your site is Cato Ridge kind of area in KZN. I hope the weather is better here than what it is on that side. It should be. And then Damien De Arruda, Damian is one of our good partners with one of our good partners, Afrolik. Damien, you’re the business unit manager for Smart Manufacturing. Welcome. Thank you for telling the story.

02:10
Speaker 2
Thank you.

02:10
Speaker 1
So maybe we have a lot to show before we get into the show. The context is always important. So maybe let’s kick it off with on theme of digital infrastructure kind of what were the requirements from your side Michael? What was the birth, the genesis of this project? Where did it start?

02:31
Speaker 3
Yeah, so we very new. Well the company’s been around for some time but our factory almost brand new about two and a bit years ago up in Cato Ridge. You can see it there on the slide. And so we recycle polyolefin plastic from post consumer waste to the final product. It’ll then go back to the injection moulding gas and so quite a big investment for the company was to build this plant and that’s when I joined the day we started commissioning in fact and kind of a factory engineer’s Dream to start in a plant that’s brand new. I come from the sugar industry, which, you know, those mills have been running for years and years.

03:13
Speaker 3
And so really here we had a clean slate to start off and we could really think about what do we need to get this place running to get it humming and, you know, what do we need to control and monitor. And so we commissioned the plant we started up and then thereafter started seeing what do we need to monitor. So this plant didn’t have much in terms of feedback to us as a.

03:38
Speaker 1
Management team and kind of like decision making capability.

03:43
Speaker 3
Yeah, almost zero. You kind of know the day after. And were working with paper sheets every day. We were guys operating. So just to give you a bit of a overview of our structure, we’re quite a thin structure, so there’s maybe three or four of us kind of in a management type role, but we are fully engaged on the floor. So the next level down is kind of the operators on shift. And so we don’t have a major base of skills. And so our focus is to digitalize and to automate as much as we can. So we started off this plant and it was like I said, paper forms every day, what came into the plants, what came out the plants late data.

04:24
Speaker 1
So your production meeting would focus on the previous days and not what we’re doing today.

04:29
Speaker 3
Correct. Yeah. And, and even, not even. We couldn’t even get that right because you’d have papers getting lost, you get stuff getting rained on, you guys losing things.

04:37
Speaker 1
So there’s this amazing thing called Excel.

04:42
Speaker 3
That’s where we started.

04:43
Speaker 1
It’s apparently very popular for production data until the file size gets to like 150 meg and then people start looking at other stuff. No, I’m teasing.

04:51
Speaker 3
Yeah, no, but I thought I was quite like not knowing the world that Damien and Afrilik come from. I thought I was quite sharp on Excel. And so I had this little production like monitoring what were doing every day. But still so far, the delay between the information we get and what actually happened was such a massive gap and often not accurate at all. Like it was very hard to make decisions. What was going wrong, what went down, when did it stop, who stopped it, and basic things like throughput, you know, so we can imagine this plant is running and there’s no information coming back. Very hard to make decisions. So that was a year into the plant running when we met Afrilec and learned about ignition.

05:37
Speaker 1
So if we can maybe summarise kind of the pains, if we can Call it that as real time visibility or just visibility. Accurate data and contextualised accurate data that will help you make some decisions as opposed to looking at late data the day after.

05:54
Speaker 3
Correct? Yeah. And I think things like stock take were nightmares because you didn’t know how much stock has moved from here to there. Breakdowns. We didn’t know what was the majority of our breakdowns. We couldn’t make decisions on what improvements we needed to make. We knew we needed to make improvements. We had like, you know, gut feels of what is going wrong. But yeah, nothing streamlined enough to really get it going.

06:14
Speaker 1
Okay, Damien, you maybe want to talk through how your observation of the summaries and how did you surmise that and kind of figure out what architecture and what solutions to put together based on that?

06:25
Speaker 2
Yeah, so basically the. The architect that we put down is pretty stock standard. There’s nothing really special to it. But essentially we adopted the full element 8 stack here because they each played an important role for different elements of their process. So the plant itself was running in fully isolated environments. Every machine isolated and not networked, and so on and so forth. So they first and foremost needed a platform to try and centralize that control. Then obviously we needed a place to store the data. So here comes Canary.

07:03
Speaker 1
Long term.

07:04
Speaker 2
Data long term, exactly. Because the journey is not what we’re not necessarily trying to solve the journey or the problem of today. We’re also then trying to put down the platforms and so on that will solve the problems of tomorrow and the day after and next year and so on and so forth.

07:22
Speaker 1
And the more kind of historical data we have, the longer period we have contextualized we have, the more valuable. Today’s KPIs are 100% cool.

07:33
Speaker 2
Yeah. So we then also chose to deploy flow to sort of centralize a lot of that information.

07:41
Speaker 1
Okay.

07:43
Speaker 2
Obviously. Okay, the hierarchy that we’ve got here is basically represented as a square. Obviously it’s not quite as a kind of plant floor.

07:51
Speaker 1
Different type devices, different brands, different.

07:54
Speaker 2
Yeah, so yeah, essentially we’ve got this basic architecture. We’ve got a SQL database that we’ve connected to ignition and flow. This is where we are restoring some of our information, which we’ll get to a little bit later. We’ve got our standard flow SQL database where all the configurations and stuff are stored. And then we’ve also got flow connected to Reproplast TRP system.

08:17
Speaker 1
Okay, cool. And then there we go. When Damian sent me this picture, I thought you sent me the wrong picture. Maybe. Michael, you can talk us through what that is?

08:32
Speaker 3
Sure. So as I said, you know, data forms on paper for this is the raw material section. As we load our factory up, how we process what comes into our factory. You’ve got. We don’t have too many products that come in, but there are a variety of them. Eight or so I think.

08:51
Speaker 1
And you essentially colour code them.

08:53
Speaker 3
Yes. So now with the ignition and the way we’ve been able to do forms on the inlet of our plant, we’ve got a like, that’s the mobile view on the right hand side and it’s showing how we load the process. And we’ve got to load our materials at a specific ratios. We call them like a recipe. So it’s not too complicated. It’s literally as you see it there, this many kilos of buckets or pails, this many kilos of pellets, some post industrial waste, HDPE, all different types of polymers and they’ve been colour coded because. Yeah. So the guy on the floor actually loading the shredder looks around the plant, he knows the blue one’s coming up next. He loads 200 kilos of blue, which is HDPE. So this form has really helped us a lot.

09:40
Speaker 3
And just in terms of usability for the end user, they know exactly what to do with very minimal training. The thing we’ve done, we put up a board by the shredder so that all the colours are there and a pie chart showing the pictures of the materials like you see here. And so it’s quite self explanatory. And so we get a pretty accurate input data to what we consume. And again this is where the stock kind of sorts itself out because now we are adding the right stuff and the right boms into our process. It’s got timestamps and the material type and the weight and that’s pretty much the extent of the form.

10:19
Speaker 1
Got a rudimentary tracking capability. Yeah. And this form lives on an HMI operator loaded HMI.

10:26
Speaker 2
This form lives wherever they need it to be. I think primarily it’s being used on a tablet, if I’m not mistaken.

10:33
Speaker 1
It’s a front end perspective ignition perspective.

10:35
Speaker 2
Correct. Yeah. So the form loads. It’s obviously capable of dealing with the different device sizes and all of that kind of stuff. But fundamentally they’re just taking this thing with them. They’ve got the recipe information with them, they go, they fetch, they load and then they move on to the next loading process.

10:56
Speaker 1
Okay. And this, and so this loading process report, sorry, this would be primarily for a stock Controller the Schrader loading events.

11:06
Speaker 3
So yeah, this the report. That’s. This is the daily report we receive for the past 24 hours at what timestamp, what material was loaded and what quantity. So this is where we get to see if there are some anomalies and we can kind of vet the data Instead of just seeing a total. You’ll see later on Ops report, we have a total per product. And you can now see a breakdown. What type of masses were they loading? What type of products were they loading? What times were they loaded?

11:33
Speaker 1
And Michael, I’m guessing this kind of replaced your process book, log book, whatever, the manual.

11:37
Speaker 3
Yes. All those sheets of paper that were using.

11:40
Speaker 1
Nice digitization. Yeah. All right. Oh, here’s the.

11:45
Speaker 3
Yeah, this is essentially just a summary so we can understand overall what the recipe was, you know, how many, what percentage of each product did we load for that day and the total kilos. So yeah, just so we can monitor daily. We know what shifts ran, we know if they’re sticking to the recipe, if they veering off. So we had plans for a bit more complicated system. Maybe in the future we’ll have where it kind of monitors performance against the recipe and tells us when we’re getting off. But at the moment, keeping it simple. We just look at these reports every day.

12:20
Speaker 1
Something like a tadr, something like a target actual. A difference that. That kind of view would be nice. Did you. Do you find that just having the kind of real time visibility. Just having the real time visibility already changed a lot of things already affected, you know, stock controller operator kind of performance.

12:39
Speaker 3
Yeah. So not so. Not so much with the loading and the input process, but we’ll see just now we’ll go through the. The flow dashboards where we see real time throughput through the plant. And that’s completely automated feedback on how many kgs are going through this plant at any given time. Here’s the slide here. So we got this really. I must say one thing, we must commend these products in Afrolik on is how we gave them a colour scheme, we gave them a design and even, you know, Damian told us can do anything on these things. And we thought so when you say.

13:15
Speaker 1
Design, it’s interesting you provided them with your CI guide, essentially your corporate identity guide.

13:20
Speaker 3
Yes. Yeah. So we’ve got a colour scheme as a company. And also we literally just drew out the report in Illustrator or whatever. It was exactly how we wanted it, thinking, okay, maybe they can get something close or maybe they can get more or less, but they matched it, you know, line for line basically in the ignition reports and the flow dashboards, they come out really well. So this dashboard here is something that’s presented in our plant. We’ve got a TV up there and it’s just connected to the.

13:50
Speaker 1
Over your display.

13:51
Speaker 3
Yeah. Okay, so this is for the operator. This is the only thing the operators see at the moment, I mean besides the input tablet. But yeah, it’s amazing how you put this up. We hadn’t been running, we’ve been running for two years without this kind of information. And the other day the TV goes down and the operators come running to you and they say, how are we supposed to know what we’re doing?

14:12
Speaker 1
The process didn’t stop but they.

14:14
Speaker 3
Yeah, you haven’t had this for two years.

14:17
Speaker 1
You’ve been doing it like this for two years. Yeah.

14:19
Speaker 3
And the good part is, you know, we didn’t train them on what this dashboard saying. It’s kind of, it’s so user friendly and self explanatory.

14:26
Speaker 1
Yeah.

14:27
Speaker 3
They look at it, understand what they’ve done today, what they’ve done yesterday, where they’re tracking on the target in the graph you can see there. And then you know, the past month’s performance. We can, we can then edit our targets in ignition, sorry, inflow. And it’s all, yeah, so well updated and easy to follow.

14:45
Speaker 1
Damien, any kind of thoughts around adding some kind of projection capability or view?

14:50
Speaker 2
So I actually have a, just a comment on perhaps how this sort of came about because Michael’s saying, you know, he’s describing the scenario where we’ve got this idea of what the data should look like and so on. And I think perhaps at the time the ideas were sort of, I want to say, I want to use the word immature. You know, we’ve essentially got an idea we’re just going to have one dashboard and it’s got everything on there and it’s showing stock controllers information and operating information and it’s just everything all at once. And this is the most effective dashboard. So essentially getting to this point was an iterative process where we saying, okay, what type of information does an operator need? They don’t need to know what last.

15:36
Speaker 1
Month’S performance, they must know what the current OEE is that is so important for an operator. You know, the OEE is currently 78%.

15:43
Speaker 2
Exactly. Yeah, exactly. So you know what, so essentially went through this exercise of trying to identify what’s important to each stakeholder and then we present information in ways that are applicable to them so that you can feel it’s the correct target audience.

15:59
Speaker 1
You’ve actually got a cool process of doing that. You kind of identify the person within the business, the kind of KPIs or the data points that they would be interested in, how they want to receive it as on email, is it on, you know, when should they be? You’ve kind of mapped that out and based on that you kind of figure out what to build for which person in the business.

16:17
Speaker 2
Exactly. And this is exactly why this, for example, isn’t on an ignition report, for example, and in a report that’s distributed daily. It doesn’t make sense for the operator because they need to know more real time information. So the information that’s here is typically very, you know, it’s much more high frequency. This dashboard updates frequently and so on and so forth. So the information that’s here is important to them. And I think that sort of speaks to why they’re saying, oh, we don’t know what we’re doing because the TV’s now down. Okay. It just sort of goes to show that by having effective information people sort of become, you know, you can build like a reliance on this and you can like project forward and understand what to do next based on this information.

17:03
Speaker 3
Just to add on that, like it’s so critical that when you get this for myself, when you get this information, what information you can have at your fingertips, you want to just add everything. And I like, I genuinely thought it was going to be a good idea like let’s put this here, put the downtime categories and per machine and put this here and all those things. And yeah, it’s an iterative process but ultimately, yeah, you need to exit, you need to think exactly who this stuff is for. You need as a manager you need different information to what the guy needs. And so yeah, it takes a bit of discipline to decide that and cut things off and put that there.

17:39
Speaker 1
And yeah, so the downtime events that you mentioned, so here’s some downtime events and the summary. So the kind of based on machine state downtime is, you know, can be done in many ways essentially. How did you do this?

17:55
Speaker 3
Yeah, maybe I can just start off with how we got this. It was actually the first implementation of how we get our data. So we before we had the input material, form and all these type of things, we had one area where we could have installed a little machine that actually counts every so many kilos and we put one proxy there and that’s the only thing and we had A PLC tied up that single proxy and just with that one tag we pulled out almost all the information we have now just with counting the throughput.

18:31
Speaker 3
So this machine counts the throughput with ignition, with flow and then when the machine is down or there hasn’t seen throughput for I think it’s five minutes or so, it logs the downtime start point and obviously when it’s up and running again, downtime end and then to that our supervisors can log exactly. While the plant is down, we’ve got per section, per machine. And what type of downtime was it?

18:58
Speaker 1
Very useful for your maintenance folks?

19:00
Speaker 3
Yeah, for maintenance and for now, you know, trending over a month or three months of what do we need capex for? What do we want to improve? What are we lacking in certain areas? So it’s, yeah, it’s. This is very helpful information to have, especially when you trend it over time. Again, something I didn’t have at all prior to that.

19:19
Speaker 1
And again, so much scope to add so much more going forward. Yeah, you’ve got to try and sort of cherry pick what do you want to add next.

19:25
Speaker 2
Yeah, it was a unique experience for me for this project where we put down this big infrastructure as a proof of concept and everything. And we had one single tag.

19:35
Speaker 1
I was about to say yes, one.

19:38
Speaker 2
Tag in the indignation, one tag in the historian. And this is where flow started, really started to shine because essentially from that were able to Michael’s point, calculate things like throughput to calibrate that device put down targets and we. So that is basically that entire dashboard that you saw was stemmed from one single data point.

20:03
Speaker 1
And it’s such an incredible point because you think about, you know, in stark contrast to, for example, one of Deon’s typical applications, the ability to have a data point, which to you is valuable, to me it means nothing. What is the unit of measure? And I think Gary is actually in track two later today, going to talk a little bit more about that. How do we essentially take these TVQs or TQVs, whatever you want to name them, take a time value, quality, add context to it and comparison and how you could very easily get 20, 30 KPIs from less than 5 data points just by adding the context and doing the comparisons. And this is an amazing example of that.

20:41
Speaker 1
You know, the underlying data that supports these reports just for some contextualization and linear regressions and all sorts of stuff, aggregations and stuff like that. So it’s amazing. Really, really cool. This is a daily operations report. So this is your entire production process reports.

21:00
Speaker 3
Yeah. This is the first page showing pretty much what comes into the plant, as I said, per product. That’s from that input tablet report or the form that we use there. The second page is the throughput through the plant. There’s a whole bunch of other calculations, scrap rates, average throughput utilisations and hours downtime. And there’s a bit of downtime events. Another interesting thing is the stock on hand. I don’t know. Damien, you can take us if you want to take us to how we started now trying to get into our ERP system for stock on hand.

21:34
Speaker 2
Yeah, so that one. That one I sort of like. It’s a fairly rudimentary integration or, you know, touch point with the ERP at this point. Basically it’s being used as a way to plan what type of production we need to perform for the day. So you would have noticed that it had like several colours that showed as green and red. In this case, there is. There are more colours to this. It’s essentially basically like minimum stocks on hand, that they require maximums and so on. And based on those different thresholds, it then obviously changes colour based on that. And it’s sort of a good way to just visually understand like. Okay, you know, there’s. There’s clearly a bit of a stock problem with these. With these four stock codes.

22:19
Speaker 2
So we need to place an order for that raw material, which is sort of where, you know, I would say probably the next phases of the project is going to go where we’re going to start integrating with warehouse management systems and so on, where we then starting to receive from the stockyards and. And so on.

22:37
Speaker 1
Is that kind of the next phase that you envisaging is maybe more integration with other systems?

22:42
Speaker 3
Definitely. Are definitely linking what we actually process to our plant and has as a final product to be updating our stock on hand end of the day so we know where we sit. And then of course, buying raw materials is just as important. Yeah, definitely. There’s a focus to time.

23:02
Speaker 1
We’re out of time. Any maybe some closing thoughts on any lessons learned? Any. You know, this was a very new process to you, Michael and Damian. Maybe a very different to your typical kind of process.

23:17
Speaker 3
Yeah, I think from our side, again, it’s the point of how simple the collection of data really needs to be to get the insights that you can really action some stuff with. So we’ve. And yeah, we’ve done a very phased approach We’ve tried to as reproblast, we try to do like little bite sized chunks every time and it’s actually worked really well for us. It gives us the confidence to move ahead and it gives us confidence to kind of grow to the next stage. To the next stage. And as you bring in the data it’s just the ideas come flooding in. So I think just to be patient with the process but also to take on manageable chunks at a time and get it dialled in so that you’re going to use it has been very.

24:01
Speaker 1
Important kind of timeframe. I think the very first engagement was sort of last year November. So just under a year or about a year ago.

24:10
Speaker 3
Yeah, it’s just on a year. Yeah.

24:12
Speaker 1
Okay. All right. Any thoughts?

24:17
Speaker 2
No. I like to Michael’s point, the manageable chunks is quite an important part of this. I think the tendency is to just do this big bang approach. We’re going to do everything all at once and what it ends up doing is causing confusion and there’s very limited buy in from the people who are going to use this. Suddenly they’re being, you know, it’s like huge disruptor. So I think these small manageable chunks does help in building the culture and sort of changing things I would say from the inside and almost like a mechanism to train everyone to say okay now this is what our journey is and this is what the value that we’re able to derive from it.

25:00
Speaker 1
Yeah.

25:01
Speaker 2
So yeah, I think that phased approach is actually quite an important way of approaching these types of transformational.

25:09
Speaker 1
Projects also not leaving people behind in the process. Yeah. Lovely guys. Thank you very much for sharing. Are there any questions for Damian Michael? No questions. Cool. Appreciate your time, James. I think it was lovely. That’s really good application. Appreciate it.