Mentis Africa: Smart Manufacturing with Ignition & Sepasoft

00:10
Speaker 1
We have Cassie, Pietie, and Ernest from Mentis. Guys, welcome. Welcome to the stage.

00:16
Speaker 2
Thank you, thank you.

00:17
Speaker 1
You seem, you seem roaring and ready to go. Is it?

00:21
Speaker 3
Well, Cassie is sitting there. I think he’s going to pass out.

00:25
Speaker 4
I’m just waiting for the brandy. What’s going on here?

00:31
Speaker 1
Ernest? So we kick it off with you. You’re essentially the recipient of this application. You are involved; you’re the customer. Please introduce yourself, your role, and Mentis as the business where the journey began.

00:44
Speaker 2
Okay, thank you. My name is Ernest Mukuwari. I’m the Automation and Maintenance Manager for Mentis Africa. I joined Mentis at the end of 2022, and the main issue was that we wanted to automate the plant. And after some brief discussions, I realised that it’s not only automation, but also digitalisation.

01:10
Speaker 1
You asked us for uns.

01:13
Speaker 2
Yes. Yeah, so I did come through. Yes to Element8, and then I asked a couple of questions.

01:19
Speaker 1
But that was the vision. The vision was a unified namespace.

01:22
Speaker 2
Yes. Okay. So, Mentis, we are into the manufacturing of various products, including land drilling solutions, expanded metal, and metal and plastic gratings. Okay, okay. There are various other products as well that we offer. Okay, okay. Okay. Initially, when we arrived or started the project, we had virtually no real automation at all.

01:57
Speaker 1
Okay, so what do we stand alone? Isolated machines.

02:00
Speaker 2
Yeah, isolated machines which were actually hardwired. Okay. And then the idea came, okay, how do I collect information from hardwired systems? So, that’s when we said, ‘ Okay, this is where the genes start. ‘ We need to put up some bit of intelligence on the edge. And then, based on the intelligence, what IoT platform do we need to implement that can communicate with various other elements within our ecosystem? Okay. And this is where I think our engagements with the different stakeholders within and outside the organisation, including CSI and Element8, are. So this is what you. We believe we need to embark on.

02:44
Speaker 1
Okay.

02:45
Speaker 2
And what are some of those products that we have that can make this journey a reality?

02:52
Speaker 1
Okay, fantastic.

02:55
Speaker 2
So, a couple of specifications were put through, and I think, ‘Okay, I’ll not be talking through, but some minimal technical requirements.’ What exactly do we want? We also want to avoid data silos. It’s a problem. We also want to be edge-driven, with an open architecture and a lightweight approach. Everything should be reported by exception.

03:24
Speaker 1
Okay. Those are obviously all MQTT requirements or, at the very least, functions of MQTT that enable them.

03:31
Speaker 2
Yeah, in a way, yes. Yeah. So, it’s either okay with OPC UA or MQTT. But when we. Okay, the question was probably about OPC UA lightweight, but it doesn’t necessarily mean it doesn’t work. But that was the thinking at that specific point. Yes, yeah. So, what were the real business problems that you wanted to solve? The issue is that we want to automate both the machinery and the business processes. Okay. And then the two islands that are always not talking to each other as well as they should are Ot and It. Of course, yes. Yeah. And yeah, then we started having some discussions. I’ve got my colleagues here as well.

04:18
Speaker 1
Are the IT people here? Should we remove them? Yeah, no.

04:23
Speaker 2
Not really. The discussion is that we don’t engage with each other. That’s usually how I was coming from, OG and IT don’t engage that well in terms of. Unless it is required to inform each other, this is what is happening, this is what we need to do, and everybody becomes aware of precisely what is happening on the other side and what support is required on the other side as well.

04:49
Speaker 1
Okay, some automation is required, with specific requirements around security, lightweight reporting by exception, and collaboration, as well as improved integration between IT and OT systems.

05:01
Speaker 2
Yes, yeah. However, in our case, I don’t think it was a major hero because the issue is that when we drafted our digital transformation strategy, it was shared with everybody, and everybody went through it. And you understood, okay, this is the.

05:17
Speaker 1
The journey for which you documented the strategy.

05:20
Speaker 2
Yes. So, then, this is when we ended up having Ignition. For us, it became a reality that was easy to connect to. We also wanted to communicate with the ERP, in our case, historize the information, and ensure that it is structured in a certain way, thereby establishing a single source of truth for everything happening within the organisation.

05:51
Speaker 1
Fantastic. Let’s take a look at the architecture. Should we look at the architecture?

05:58
Speaker 2
Or.

05:58
Speaker 1
Should we cover these aspects first?

06:00
Speaker 3
No. Okay, that is that story—oh, perfect world.

06:04
Speaker 1
Oh, okay.

06:05
Speaker 3
So that is the inquiry that came through to CSI.

06:09
Speaker 1
And that was the requirement, basically what we’ve been through.

06:11
Speaker 3
So he’s only focusing on the end product and what he wants to see at the end. Okay, so if you go back to the second slide.

06:17
Speaker 1
Yeah, this one, that one.

06:20
Speaker 3
Okay. So the original design had hardwired controls for each line. So there was no feedback, nothing. It’s operating interface. So we had to convert that into a system so you can pull information from the line. Wow. With an operating interface. Therefore, downtime reasons for downtime operator T breaks need to be collected from each line and entered into each ideal system.

06:49
Speaker 1
With the downtime reason code.

06:51
Speaker 3
That was the first challenge: to standardise all these 12 lines that were supposed to be standard, but none of them would be standard because everything was hardwired. The engineering behind it was the first challenge, as it needed to ensure everything was standardised. These PLCs require standardised documentation, and the tagging from the PLC into Ignition needed to be standardised as well for each line. That was the first challenge I didn’t mention.

07:22
Speaker 1
It wasn’t a software or an application challenge.

07:24
Speaker 3
And I’m going to bring this up because we. We invoiced like 85% of the order value, and there was nothing in the plot. And his management was not very happy with him. And I said, Well, trust me on this one. We need to do it the right way. After a month, we could start seeing panels and information being pulled through as a standalone system.

07:47
Speaker 1
Okay.

07:48
Speaker 3
Okay. May I speak for a moment?

07:50
Speaker 2
Yeah, that’s fine. Yeah. So when the first panel was installed, it went through some. A couple of iterations in terms of refactoring, let’s see how to sort this out and finalise. And when those other things were out of the way, the rest of the journey was relatively quick.

08:11
Speaker 1
Okay, so there was a reasonably unforeseen thing, according to your plan, that was not known at that time—so first hurdle to get over the pity factor.

08:22
Speaker 3
However, this project was divided into two phases. Phase one involved implementing automation for all the lines. The second was the implementation of Ignition Sepasoft Canary. We plan it in such a way that the integration into the current plant network, which is not currently in place, will be possible. That’s where my friends from IT had to get involved, and there’s an IT and OT issue. So. Ultimately, the integration between the IT team and us worked out quite well when they saw the implications of incorporating our Stratus redundant service into their system. What will happen? My question to them is, what will happen? And that’s where everybody got on board, considering the implications.

09:10
Speaker 1
Okay, cool. This is the solution. We discussed the EDGE devices, ERP Sepasoft MES modules, along with Ignition and the existing Canary and UNS, or at least the vision of an UNS that is enabled, in part, by the MQTT functionality.

09:36
Speaker 3
Cool.

09:38
Speaker 1
Some of the results before we look at the actual physical. Physical sites. Do you want to discuss the results you’re seeing and how you’re feeling? Ernest.

09:47
Speaker 2
Yeah. Kindly. Initially, we did not know what we were manufacturing on the floor, because most of our processes are paper-based and Excel-based. We want to ensure that we consume work orders from the LP into the MES and then run them on the machine.

10:06
Speaker 1
Sage. So.

10:07
Speaker 2
Yes. Sage X3. Yes. Yeah. So that is happening now. We consume the work orders, schedule them and then run them on the production line. And once we are done with the production, we can also submit the feedback to the ERP. In terms of how the production went, that also speaks to the raw materials consumed. And then we can use that to say, ‘Okay, based on the ERP, the ERP indicates how much is supposed to be consumed.’ Then we say, okay, this is what. What was actually consumed?

10:38
Speaker 1
Fantastic. You can track targets and actuals.

10:41
Speaker 2
Yes, targets and actuals.

10:43
Speaker 1
And of course, with that, you’ve got beautiful OEE screens, by the way.

10:49
Speaker 2
Okay. So, yeah, so that. Yeah. So, at least we now know what exactly is happening on the plant floor. What is the loading for each specific production line? Do we have capacity in terms of production? Can we meet the customers’ deadlines based on the. Regarding the loading on the machines?

11:10
Speaker 1
So, you can order the actual target, stall any downtime for whatever reason, and you’ve got OE over that entire view of the order?

11:20
Speaker 2
Yes.

11:21
Speaker 1
Yeah. Very, very nice.

11:22
Speaker 2
Okay. Then as. As we send the work order, we also consume the specifications. Of the product through to the operator so that they know exactly what is expected, and they actually set the machines accordingly.

11:38
Speaker 1
Fantastic.

11:39
Speaker 2
Yeah.

11:40
Speaker 1
So this is the architecture.

11:42
Speaker 2
Yes, this is more of the architecture that was put in place, and then we handed it over to CSI. This is what we perceive as division, but there may be some minor modifications to it, so that we don’t have the full budget yet. However, as we go through the process, we already receive some of the value. Mindful of the future. Those items in blue are not yet implemented, but they are part of the future. Our unified namespace is actually created within Ignition itself. And this is where we also perform data operations within Ignition.

12:22
Speaker 1
Okay. So, it’s been an agile development approach, but at the same time, you’ve got the platform and infrastructure in place.

12:29
Speaker 2
That’s true. Yes. Yeah. To highlight this, we are doing this for one specific Production area. However, the infrastructure is being laid out for the entire area. For the whole of the site. Okay, okay. And.

12:47
Speaker 3
Can I discuss that one, please?

12:50
Speaker 1
Okay, well, your name’s on it. So.

12:54
Speaker 3
Okay. All the good things. He’s talking about all the good stuff. That is the network architecture. All the panels at the bottom represent the actual lines. Firstly, we had to identify the IP ranges. Ultimately, it will need to tie into the Mentors’ network. So that was this architectural drawing. Looks like a few lines. However, it took a few months to compile because we also had to interface with it. From each line, we interface with four network panels that are installed inside the factory itself. We incorporate these into the new fibre network and connect them to the tops in the server room. Initially, we also allowed for a standard server, not a macro server, but a normal one.

13:44
Speaker 3
And looking at what Ernst wanted at the end, it was like, no, that is not going to work. Let’s get a decent interface with Ignition. And that’s why I went to my friend, Mr Albertus, and I said, ‘ Come, we need to make a plan here. ‘

13:58
Speaker 1
Okay.

13:59
Speaker 3
As Ernst mentioned regarding the budget, there was no budget. We started with one ZTC for phase one and then incorporated a second one for redundancy in phase two. Additionally, the interface between the server and the IT department is also something new for them. So on Ignition we. We actually incorporated the ztc, what do you call it, about this, into the Ignition. You can then see full access to your Stratus or server, including its status and other details.

14:36
Speaker 3
Okay, so that was all integrated based on the support from you and the starters.

14:47
Speaker 1
All right, these are some of the Ignition screens.

14:49
Speaker 3
Okay, that’s classy.

14:51
Speaker 4
Yeah. So this is the. To put it simply, that’s the best way to put it. Each line explicitly indicates the main page or operating page. From here, we give them access to pull the works order itself, along with all its information, including material types and everything allocated to that specific works order. We provide the operator access to be able to pull it down to the line level itself and then from there to make sure that all the information is correct, that they’re using the right material, that the setup of the machine is accurate, the stroke allocation of how long the sheet needs to be and everything else is in place for them to be able to do it.

15:33
Speaker 4
And then, from there, if they’re happy with it, they can also change orders based on having multiple orders lined up and then just filtering through each order to see what needs to be done and what’s a priority based on that. As soon as they accept the order, the machine will start processing it as if it were in production, and then they’ll have to make a decision. If they’re not running production immediately, they’ll need to make decisions based on that for downtime reasons. What’s going on physically on the machine?

15:59
Speaker 1
There is also a considerable amount of reporting that supports these reasons.

16:04
Speaker 4
Yeah. The majority of downtime reasons are grouped to facilitate reporting and provide an overall view of the time spent on downtime reasons. And then, behind that, we also have automated downtime reasons, such as E stops overloads, and everything else that is pulled through automatically is being pushed into the system.

16:30
Speaker 3
Just one step back. We started by implementing our system as a standalone system. Okay. So kind of you. Previously, everything was done manually. So, I got the order from the sales department and a piece of paper. A piece of paper and walk it through—piece of paper.

16:53
Speaker 1
Hand over between.

16:54
Speaker 3
Yes. And then every hour, a piece of paper to look at the production report on how many products have been produced. Okay. So, we had to start by explaining how the order actually goes into the system. So define that. We started with the production manager, Israel. What israel? Okay. To load it into the system, we obtained the format for how orders are pulled into the system. So it’s all Excel sheets. So within the customer cell, we had to load it into our SQL database. Okay. And we pull that into ignition and teach the guys how to use the system as a standalone, not interfacing with anything, as it stands. It was like a learning curve for everyone to get used to it. Leave the piece of paper.

17:46
Speaker 1
Because it wasn’t an application change, it was a complete process change.

17:52
Speaker 3
And even to get buy-in from operators, they’ve been working there for, say, 50 years or something like that. Of course, because meters are 75 years old. So to change that mindset from paper.

18:04
Speaker 1
To date, something that’ll improve their days quite significantly, though, so there is some incentive.

18:11
Speaker 3
The development of these screens was an ongoing process to make it as easy as possible for operators to understand. That looks very simple, but there’s a reason why it was built that way, because somebody needs to maintain it. Okay. Yeah.

18:27
Speaker 4
The entire concept is that the majority of the pages were actually template-based. Everything is running smoothly in the whole system. If it involves multiple machines, or multiple users or roles with different requirements, everything is templatized. So, if there are extra machines that come in, or even as shown now, the standard operator pages, each line has the majority of the specific parts in the machine itself, but everything is simplified. So if an operator needs to move from one machine to another, they’ll be able to understand what’s going on without struggling.

19:02
Speaker 3
That includes the PLCs, the structure of the PLCs, all the blocks, and function blocks. Yeah.

19:08
Speaker 1
These are the downtown pop-up pages we saw in the video earlier.

19:12
Speaker 4
Yeah. So it’s just. Yeah. Specific reasons based on the group. Specific groups, based on which the operator can use particular reasons. Obviously, tea times, lunchtime breaks, and everything else are automated. These are the specific inputs required by the operator.

19:32
Speaker 2
Okay. So, when we see this idea, everybody’s on the floor. Yeah. Because we can see there is a big problem. Okay. But in essence, what visual way of.

19:47
Speaker 1
Understanding where the problems are.

19:48
Speaker 2
Yes. In essence, it just shows us. Okay. What is actually happening on the plant floor on that specific production line?

19:55
Speaker 1
Very nice.

19:55
Speaker 2
But when it’s red like this, then there is war. Yeah. Everybody on the floor. No.

20:01
Speaker 3
Well, I’m laughing at the screen. When we, the CEO, I think we had to. We want to know what the system does. I mean, we spend so much money, but I can’t. We can’t see anything. So we’re sitting in this big boardroom, and I said to Kazi, Go online with the system. Go to the overview page. That’s the overview page. So anything that’s not green, he’s not in production.

20:25
Speaker 1
Yeah.

20:26
Speaker 3
And the guy was like, so you guys want to tell me that’s only those two lines that’s in production? And everybody in Metis looked at me and was like, What did he do? So that was not a very good first meeting. But at least you could see this live information that was previously delivered.

20:43
Speaker 1
You see, you have the visibility that you never had before. Yeah.

20:46
Speaker 3
Yes.

20:46
Speaker 1
And. And it’s a nice visual layout of on or off.

20:50
Speaker 4
I suppose.

20:51
Speaker 2
Okay, that’s true. And it speaks to the entire organisation. Is it maintenance, or is it production issues? Or we don’t have sales, or we don’t have orders. So Everybody needs to run. If there are no sales, they need to go and get orders.

21:07
Speaker 1
Yeah.

21:08
Speaker 2
If the orders are there, however, the manufacturing team is not fulfilling their part. They need to pull their socks up.

21:14
Speaker 1
Absolutely. Very nice overview. Some of the MES screens. There we go. There’s the OEE display.

21:23
Speaker 4
Yeah. So the enterprise is the homepage for the enterprise itself. It’s. It’s primarily based on the fact that it focuses on only one specific site. But as we progress and it continues, if there are expansions, it always allows for more areas or plots itself.

21:40
Speaker 1
Yeah. So, does that scalability get built into it?

21:42
Speaker 2
Yeah.

21:43
Speaker 4
The majority of the entire system is built for scalability and adaptability, allowing it to import any other plants into the system without needing to redevelop anything behind it.

21:55
Speaker 1
Very nice. Eventually, we’ll see many more over there, right?

22:01
Speaker 4
Oh yeah, the order tracking page.

22:05
Speaker 2
Okay. Yeah. This one was actually a request from the production manager. They want a single page where they can view all the production lines. They can see exactly what is happening there, hour by hour, and probably also the total for each production line. And then this is more of an overview of the actual production.

22:24
Speaker 3
So that was the replacement after walking around every hour.

22:27
Speaker 1
Yep.

22:27
Speaker 3
Writing it down.

22:28
Speaker 1
Yep. Do you have these as specific displays on screens, or are they on? On overhead displays?

22:36
Speaker 2
Possibly, or that is the plan. However, if all the managers or supervisors currently have access to that screen.

22:45
Speaker 1
Very nice line summary.

22:49
Speaker 4
Yeah. So the line summary page would. It’s more of an indication for a supervisor, manager, or whoever needs a clearer view of what’s currently happening on the machine. The running states, the outfit count, and how it’s being produced. Like the top right part showing for that specific machine. What could the production rate be? Or what’s the average production rate compared to what we’re actually doing right now? As indicated in the bottom part. Okay. Green will be running the reds’ unplanned downtime. And then just giving them an indication. As soon as you hover on that physical graph or the bottom part itself, you’ll be able to see. Okay. During this duration, these types of downtimes occurred. So just giving them more information based on what’s actually happening on the machines.

23:40
Speaker 1
Very nice.

23:44
Speaker 2
Okay, you can go ahead.

23:46
Speaker 3
Okay.

23:47
Speaker 4
Okay. The same applies to the run control page. Just giving them an indication at the top left of who the active operator is, and who’s currently active, based on the line status. Top downtime reasons, just giving them an indication of what’s going on or what they are presently struggling with throughout the shift. So, an indication of the outfit is whether they are continuing with production? And then the bottom part, the downtime table, where we’ve seen that there are specific scenarios where operators selected the wrong downtime reason, or something occurred, and they had to change it. This gives a supervisor or manager the ability to change it later if there is incorrect input or any other issue.

24:31
Speaker 1
So it’s manager level.

24:33
Speaker 2
Yeah, yeah, makes sense. Yes. This also speaks to the strategy. Currently, you can see that there is one downtime visit, which can easily be managed. But now, once you have the information in real-time, you go down. Guys, we need to organise and reorganise ourselves, instead of stopping the machine so that you can move the product. The product needs someone to be responsible for it. Make the. Let the machine continue to run production so it’s actually getting results. That’s very nice.

25:04
Speaker 1
Some reports.

25:05
Speaker 4
Yeah. One of the necessities was daily reports. We’ve got three. Okay. We don’t only have the two here, but they also needed a production report, a downtime report, and an area report. So, each morning at 7, these reports are sent out to give them an indication of the previous day’s results. Okay. What occurred? The weighted average of the material that was produced throughout the day from that line. Downtime reasons themselves, grouped by reason. So, each morning they have an overall view to go into a meeting and discuss what happened on the floor.

25:40
Speaker 1
At 7:00 a.m., you clock in automatically, and you arrive at the production meeting with the report. Lovely. Very nice.

25:48
Speaker 4
Okay. The axiom part of it, we’ve started with the dashboards, so we are currently busy with it. The majority of our idea was to understand the ground floor of the information, specifically how the data itself and the tags and everything were structured correctly, before pushing it back to Canary. We’ve already set up the connection between Canary and Power BI to push that information out. It’s just that we’re now able to send back the correct information. Over the last three or four months, since we’ve installed Historian, we’ve been trying to create a pool of data that we can start using and utilise.

26:28
Speaker 1
But yeah, that’s why it’s the ODBC connector.

26:32
Speaker 4
Yeah, via the OBC connector.

26:34
Speaker 2
So, we thought that by now, the MCP servers and agents would be ready. Then we can start talking. Yes, it will be.

26:43
Speaker 1
Kevin said quarter one. That was.

26:46
Speaker 2
Then you can tell us. Okay. What exactly is happening in terms of things that we can’t see with our eyes?

26:52
Speaker 1
Fantastic. Lovely. Ernest, You. You had the. You had the vision for the unified namespace. I think these guys did an incredibly fantastic job. To create that infrastructure using all the required elements, especially MQTT and the role that MQTT plays in it. Which, to most people, has never been heard of. It’s certainly not something as. As widely known as OPC UA. Most people don’t even know what mqt. You guys, that was your first experience with MQTT, project, and adoption. You did amazingly well. And visually, it’s fantastic. The data that you’re pulling through is now cleansed. It’s contextualised. You have visibility into OEE and the order process. You’re pretty close to your vision.

27:44
Speaker 2
Yeah, it’s very close. Yes. During the break with my executives, they were telling me they have some stories here now. We now need to ensure that we accelerate the implementation and provide those. Those. That’s why I’m discussing AI agents, because now it’s acceptable. We need to put a true. We need to put a true on the CEO’s desk so that they can clear it. Okay, tell me what is happening on my production line.

28:18
Speaker 3
Love me.

28:21
Speaker 1
Any. Any closing thoughts?

28:23
Speaker 3
Well, I just wanted to. On the separate side, we had a bit of. Not a bit of a challenge, but it was with. With all the requirements and what we actually wanted to do. Super soft.

28:35
Speaker 1
Yeah.

28:35
Speaker 3
I mean, the support we got from. Jesse from America was really. I mean, we sat every. Initially, every week, after hours, of course, with all of us. And it wasn’t a training session; it was a live scenario on what we had achieved, and we tried to replicate it, which was unique. It wasn’t like a normal thing. It’s something unique that we want to do in our system, and we’ve tried to implement it, but sometimes it breaks things. After hours of trying at home, we still can’t get it to work, and by tomorrow morning, it still isn’t working. However, we did it as a team to communicate with each other and ensure that one hand knows what the other hand is doing.

29:18
Speaker 1
So. Amazing. Yeah. The support. We are also grateful for the support we receive, especially from the US, where we have partners such as Ignition, Sepasoft, and Canary. The support is phenomenal. Makes a project a lot easier if you have that support on the same day. Fantastic. Any questions for these gentlemen? Very cool project. No questions. Best projects in the world. I agree. Okay, thank you so much, everyone. Appreciate your time.