Shu Powders: Plant Monitoring & Traceability with Ignition, Canary, and Sepasoft

00:10
Speaker 1
All right, so we’re excited about the next part. We will do just as a recap, we’re going to talk through two amazing case studies, and I must say it is very intimidating for people to tell their story. There is so much value in hearing the story for the rest of the community, but it is very intimidating, and it does take a lot of time to put it together. So, just as a precursor and as a disclaimer. So we thank everybody who will be on stage with us today, and we will kick it off. First of all, am I right? Sorry, I didn’t introduce you. I’m so used to Gary always being here. So, those of you familiar with Gary Lowenstein, he is our grey wizard, Gandalf the Grey, at Element8, Sales Engineer.

00:53
Speaker 1
Fantastic person to have a conversation with about building an application. So Gary will actually join me for these sessions. He was quite close to the process of looking at the applications. And we will kick it off with, first of all, with Afrilek, I’m going to invite Damian from Afrilek and Rajiv Singh from Shu Powders to the stage. And please give them a warm welcome to help them with their work. Thank you. There’s Rajiv. Oh, there’s Rajiv. I thought you’d run Rajiv. All right, so I’m going to move up a bit so I can see you guys. So before we get into the, I mean, the application itself is, it’s not recent as you’ll see from the slide there. This is a project that was initiated in 2020. Can you believe it? It’s been, it’s been that long already. So it’s not recent.

01:55
Speaker 1
It’s been a long-term in the making Agile Approach project, but maybe before we get into the details, Rajiv, we’ll start with you and then Damian, if you can give us your kind of, your title, your role and a little bit of background behind Chupa and what the business is and what you look after.

02:15
Speaker 2
Okay. Yeah. So thank you for inviting us. I’m Rajiv Singh, Operations Manager of Shu Powders. Shu Powders is a cobalt manufacturing company. We’re the only one in Africa and we basically 98% of our customers are brought in Europe, America and in Japan. A few years ago, we embarked on a journey together with PwC, the auditing firm, on a lean manufacturing journey, whereby we needed to remain competitive in this industry. The cobalt markets were very unstable, as such, and so we embarked on this lean manufacturing journey. And we got introduced to Afrilek via PwC, who then partnered up with us and took us on a digital journey whereby we had a phased approach and we worked from one end of our plant to the other end, not yet complete.

03:19
Speaker 2
So yeah, we also, with the pressure of the ESG requirements, we had to take this approach. Most of our European customers also want us to meet certain requirements, and we’ve seen how Ignition can help us support those requirements as well. Yeah, so we do have some big customers that many might know. Some like Bosch, Mitsubishi, Sumitomo and. Yeah, so yeah, it was a great journey, and I think. Yeah, maybe Damian can explain a bit more.

03:56
Speaker 1
Perfect, thank you.

03:59
Speaker 3
So my name is Damian. I’m the BU manager for Smart Manufacturing at Afrilek Automation. As Rajiv mentioned, we are a technical implementation partner for this project. We weren’t necessarily responsible for scoping and managing the project and so on, but perhaps just a bit of history regarding. This was actually our very first Perspective project. So, looking at the date 2020, it’s basically when Perspective launched. That’s right. And went right in there. Boom.

04:35
Speaker 1
Yeah, I think one of the first Perspective applications certainly is in our market. That was brave, Damian.

04:42
Speaker 3
Hey, I was ready. Yeah. There were many diverse requirements that sort of demanded the need for Ignition, perhaps just as a general platform. There were competing offers on the table for more custom-developed solutions. And yeah, I think we made a good decision by going with Ignition for this particular project because of the capability that it’s enabled and the journey that we’ve been able to take with Shu Powders.

05:17
Speaker 1
Yeah, fantastic. So the requirements were initially complete, end-to-end, so kind of raw materials all the way through to the finished product. Right through. It wasn’t a specific application case of a functional part of a plant. It was essentially the entire process.

05:36
Speaker 3
I would say yes, that was sort of the end goal. Okay. The end goal was more like a consolidated system that did more than just plant visualisation.

05:45
Speaker 1
Unified view.

05:46
Speaker 3
Yeah. So it started off with one part of the process, but once we started realising the capabilities of it, it quickly expanded to this and that and that and that. And ultimately, the scope sort of grew under, grew like almost exponentially, because the capabilities were, you know, it was unlocked, innovation unleashed.

06:11
Speaker 1
Okay, all right, so maybe talk us through the initial problems that you were looking to solve and the requirements that. Let’s maybe kick off with that. So we can talk here about siloed information. Lots of paper processes. Manual. Manual paper processes, I would imagine—clipboards, whiteboards, those kinds of things.

06:32
Speaker 3
Yeah, like Rajiv alluded to, the environmental requirements of the European customers sort of demanded very high levels of traceability, quality control, you know, that kind of thing. So the auditing process is particularly important for them. In fact, they do environmental audits. I don’t know how often. You guys are always doing audits. And every time there’s an audit, it’s always like, okay, where did this product come from? What happened? Where was it through this process? Why was it the way it is? So a lot of this was very paper-driven, like we were saying. And obviously, being able to consolidate that information from a very paper-driven process is just an immense exercise. I mean, having to dig back years and years. How do you do that? Yeah, it’s simply, it’s simply a big challenge.

07:38
Speaker 1
Yeah. Okay.

07:39
Speaker 3
Yeah.

07:40
Speaker 1
So, so genealogy, traceability, rolling all of that up with the assurance that you need that the data is available, if you need the data, and then providing a unified view.

07:51
Speaker 3
Yep. And not only that, the process itself is obviously quite dangerous. It’s an explosive environment. So being able to remotely monitor process parameters, to be able to receive alerts of those things, those are critical to save human lives. You know, the truth of the matter is that if something goes wrong by accident, perhaps there’s an instrument failure, people will die. And you know that’s a significant risk that must be mitigated. So having systems in place to be able to do that, you know, significantly reduces their liability. It also increases worker safety and that kind of thing. Cool.

08:34
Speaker 1
So, challenges, I see there’s a whole list of challenges. I would imagine you kind of, on inception, rolling out, you encountered your first kind of challenges that usually come in the form of connectivity and devices. I’m very happy to see that there’s only one bullet point for the solution.

08:51
Speaker 3
What is the solution? We put down ignition.

08:53
Speaker 1
But many challenges as we typically do. Can you maybe give us the highlights and the lowlights of the main challenges that you had as part of the initial project?

09:03
Speaker 3
Yeah, so the, as Rajiv mentioned, there was a drive for like lean processes and a lot of that kind of thing driven with the consulting partner. So essentially, the idea was to present all of this information to the very specific people. There are lab personnel, operators, maintenance Teams, senior management, middle management, directors, shareholders, and the whole team. So there’s a lot of information that needs to be collected. There’s a lot of information that needs to be made available globally, not necessarily only locally. And there are various systems that we need to collect information from, and so on. And yeah, obviously, our solution in this case was primarily Ignition at the start. However, as we encountered new challenges, it started to expand. We started running into performance limitations of the SQL-based historians.

09:58
Speaker 3
So we put down Canary to resolve a lot of those challenges. We then started involving other things like standard operating procedures, all of that. And then we started incorporating Sepasoft into the solution.

10:12
Speaker 1
Okay. And then the result is currently, and I know it’s ongoing, and in many ways it will always be ongoing, otherwise it’s perfect. So it’ll always be a work in progress. There’s always something to enhance and to do better. But right now, you refer to it as unified operations. And Rajiv, from your point of view, does it feel like a unified system with unified operations, does it feel like that objective is at least closer or at least it’s closer to that objective of a unified view?

10:39
Speaker 2
Yeah, from my side as well, I.

10:41
Speaker 3
Think.

10:43
Speaker 2
Having unified operations also from a business point of view, you save a lot of cost. So I can give you some stats. Prior to installing Ignition, we used to have a recycle rate that was above 6% with a target of 5%; our targets are now 1%. So we are able to manage and control the process. So we have a very complicated process. At the onset, we have a furnace with 13 different zones, and you have to control each zone in order to get a proper finished product on the other end. And in the past, as Damian mentioned earlier, there was a lot of paperwork. We literally had operators dedicated to running around with a logbook, looking up information. And now we can just go into the system and pull back information since it was installed. So what comes with the cost savings apart from the savings and recycling?

11:48
Speaker 2
Our process is kind of interlinked. So when these are recycled, then we get scrap product generated, which you can’t recycle, and we have to sell it at a loss. And so that takes care of scrap. A big portion that Damon mentioned as well is safety. So our process is very explosive. We manufacture hydrogen from decomposing ammonia, and we don’t store it. But if we cannot get the process right, you not only get recycled, but you can have an explosion, which can basically be fatal. So from that perspective, yeah, we have full control of the alerts that we get through to make changes. Great. It allows us to basically be proactive. In the past, we had to wait for six hours to see what came out of the finished product and then react. We’ve already got a few tons of recycling and so on.

12:48
Speaker 2
Now when we see the, we can see. We can see the problem before it happens, basically, and we can react immediately. So, yeah, I think AI Ignition has basically helped us since it was installed. And there are great benefits. It’s all. It all has a ripple effect as well.

13:13
Speaker 1
Perfect. So, the technology is one aspect. I think the way that it was deployed was the fantastic bit. So there’s visibility now, and there’s the ability to make a decision faster, which is the value of information as opposed to just raw data.

13:30
Speaker 3
Cool.

13:30
Speaker 1
Damian, do you want to talk through any of the other results before we look at the architecture? Some things that stand out.

13:36
Speaker 3
No, I think we’ve largely covered.

13:38
Speaker 1
Yeah, I think probably the only one was the mobile aspect, which I think is probably important.

13:42
Speaker 3
Correct? Yes.

13:44
Speaker 1
Because there is a massive design consideration there.

13:46
Speaker 3
Yes, 100. That, that was one of the big things. Mobile first means that we, working with devices that aren’t the same, never have the same screen size. We need to be able to flexibly adjust to these screen sizes. And how do you design around that? Because you don’t have infinite real estate to work with. So it’s a particular problem to do that. Like a sort of mobile-first design. Yeah.

14:14
Speaker 1
So with Mobile first. Sorry, Gary, just Mobile first. So there are different types of clients, naturally. There’s a workstation, kind of a client that you would typically expect on an HMI. There’s a web client that is accessible via any browser on any device. And then you have the app, the Perspective app. So, in terms of mobile first, what were some of the discussions around accessibility? Whether it’s the app or whether it’s simply a web page? What. What did that look like?

14:43
Speaker 3
There wasn’t really a big discussion around that. The important part was that there was flexibility.

14:49
Speaker 1
Okay.

14:50
Speaker 3
And that was ultimately the real thing. That was essentially the requirement, and whatever. Whichever. Whether we use the browser version or the Perspective app, or somewhat immaterial, if there was a specific requirement, like we need a GPS, for example, then we would use the app, obviously. But predominantly, it wasn’t a big concern.

15:11
Speaker 1
Okay, just a quick question, Damian. So, mobile security is obviously a big concern. How are you securing the system?

15:19
Speaker 3
Yeah, so this is now like a joint thing with the IT team. So all of our systems are accessible remotely, which means that we’ve got to do secure tunnels back to the site. We then have to have the correct user permissions. We’ve got security zones, we’ve got dedicated logins, we’ve got that kind of thing. So yeah, it’s a joint venture. It’s not just Ignition doing everything here; it’s being fairly deeply involved. Yeah, but that’s pretty much it.

15:50
Speaker 1
Okay, so this is the architecture, maybe fairly self-explanatory, but it gives us an idea of what it is. It looks like a typical site server architecture.

16:03
Speaker 3
Yes, that’s correct. So we’ve got redundancy, obviously, because of the safety aspect. I’ve simplified it down to one PLC, but that’s quite frankly not a true representation of the site. It’s in fact many, many PLCs on site. Okay. But we do have four separate databases that these ignition gateways are managing.

16:26
Speaker 1
Sorry, the PLC is multiple brands and models.

16:29
Speaker 3
Yes, correct. Yeah. All was managed by a separate contractor who was doing the existing automation. So yeah, we do have one cloud-based database, and this is now for the more enterprise-level reporting. Yes, but essentially, this is roughly what it looks like. Obviously, physically it’s not looking like this, but yeah, it’s fairly simple like this.

16:56
Speaker 1
Okay.

16:57
Speaker 3
Essentially, the important part of the journey is about the traceability.

17:02
Speaker 1
Yep.

17:03
Speaker 3
So, basically every process that we’re working with has got raw materials that are put into a process that then becomes work in progress material, and we are allocating lab results to it, and you know, ultimately it becomes finished goods.

17:18
Speaker 1
Okay. And you have a lot of IDs and tracking.

17:21
Speaker 3
Yeah. So essentially, what we’re looking at here is just this is where. Typically, they’re scheduling work orders. So they’re scheduling it for a specific client. It’s being allocated a work order number or lot number that they refer to it as. We’re determining a product, we’re determining various aspects, how much and so on. And then we’re allocating it to the furnace, which is then ultimately going to produce this material. So yeah, it’s just a management screen.

17:50
Speaker 1
Okay.

17:51
Speaker 3
This is the desktop version. There are equivalent mobile versions, obviously. Yeah. But just for simplicity’s sake. So essentially, this is just about the management of work orders.

18:01
Speaker 1
Perfect. And the scheduling, is that from the ERP or so at this point?

18:06
Speaker 3
No, this is because they weren’t working with that back then. But yeah, it will eventually get there.

18:15
Speaker 1
Okay.

18:17
Speaker 3
So now talking about our raw material, in this case, we have got raw material that is received from a supplier overseas. Perhaps maybe we. We can actually go into the next slide, if you don’t mind, which just shows a different thing. So, in fact, what we’ve actually done, they. They receive a. They receive an Excel sheet from their supplier to say, This is the stock that’s currently on its way to you. So what we had to do was we developed an interface where we could just take that Excel sheet, drag and drop it into Ignition, and suddenly Ignition is aware of all of the raw material that’s inbound. So, yeah, I mean, this demo just shows that we can simply drag and drop it in. That red indicates that we haven’t found this material in our database before.

19:04
Speaker 3
So we are aware whether we’ve dragged in a previous Excel sheet by mistake so that we don’t over allocate all these things.

19:13
Speaker 1
So there is validation.

19:15
Speaker 3
Yes, correct.

19:15
Speaker 1
Fantastic.

19:16
Speaker 3
Yeah, there is. Fantastic. Then, over and beyond this, we also receive a COA, which is a certificate of analysis from the supplier to say this is the current quality of the material that we’re receiving. So very similar interface. We then drag that in, it immediately allocates it to the correct material, and then the lab teams have this result available to them, and they can then, when they’re doing their own internal validation of these results, they can see.

19:43
Speaker 1
This is what the supplier.

19:44
Speaker 3
Supplier reported it as. This is what we reported it as. And yeah, then this information is actually available to operators who then choose which raw material bags they want to make available as well.

19:56
Speaker 1
Seemingly such a. Such a simple thing. But prior to having this ability, having an Excel sheet, what did that look like for you and all for the team?

20:06
Speaker 2
Yeah, it was manual capturing.

20:09
Speaker 1
Exactly. Yeah. Then. Then you’ve got the. You’ve got the possibility of human error. Okay, all right.

20:19
Speaker 3
So the next portion that we’re looking at here is just a mobile view of the furnace. In fact, there is a video associated with this. I don’t think it’s playing, but essentially, we’ve got this interface where we’re representing the part of the process. We’ve intentionally designed it somewhat like a standard, an interface that people are sort of used to. So we’ve done it like a scrolling format, like Facebook or Instagram, where you can scroll through the process. But the important part of this was that we needed to try to replicate the function that’s available within the desktop environment. So the pop-ups and so on that are here on mobile, there are no pop-ups. It is all about navigation and so on. But functionally, it operates exactly the same.

21:08
Speaker 3
And in this particular process, we’re just showing here about these forms that we then had to build in because obviously not everything is automated.

21:18
Speaker 1
You didn’t use Iggy to build the forms?

21:20
Speaker 3
Say again?

21:21
Speaker 1
You didn’t use Iggy to build the forms like Kevin did?

21:24
Speaker 3
No, this was manually built. But essentially, the idea here is to try to reduce the amount of human error. So, instead of manually typing in which raw material bag we’re using, there is a drop-down, and we get that drop-down from the currently in stock material and so on, and then we can choose whether we’re feeding in recycled material or whether we’re feeding in raw material or whatever it might be. So there are various aspects of that. Fantastic. Then, as far as our process goes, this is now a bit more advanced than what we added. But essentially, when the furnace is on, we’ve got a state machine that’s monitoring the parameters of the furnace.

22:09
Speaker 3
When we detect that the furnace goes into a certain state, we then trigger an event in the Sepasoft modules that creates this form that we need to check that the operators now need to fill in to be able to start the furnace.

22:25
Speaker 1
Okay.

22:28
Speaker 3
Yeah, so it’s again sort of incorporating the SOPs and that type of thing instead of it being manual on paper.

22:35
Speaker 1
Yeah.

22:35
Speaker 3
Now we’ve got it stored digitally. It’s available for everyone to look at the startup shutdown.

22:40
Speaker 1
And is that electronically signed off, or use a specific sign off?

22:44
Speaker 3
Yeah, I mean, at this point, it’s a bit, it’s, I want to say it’s a bit immature. It’s not quite fully expanded to everything we’ve started with the furnace. Okay. Yeah. But it will grow. And then obviously the requirement was that it is mobile-capable. So we’ve got exactly the same thing on mobile.

23:02
Speaker 1
Fantastic.

23:03
Speaker 3
Then obviously now we’ve got work in progress material. Now we are allocating lab results to this. So this is just a representation of the mobile-like dashboard that we’ve got to show. Our lab data is allocated to a specific sample, and we can then relate this back to a specific work order and so on. But this is obviously a very generic view that can be used by an operator to see, okay, the process is trending out of control, so we need to make a Correction, or we don’t need to make a correction or whatever.

23:37
Speaker 1
It might be the very visual understanding of it.

23:40
Speaker 3
Yes, correct. However, the need is that we would perhaps. Yeah, we can actually go to the next one. Need is that we might want to look in more granular detail, specifically the parameters of the process that caused the sample.

23:54
Speaker 1
Okay.

23:55
Speaker 3
So there’s a lot of math that went into this thing. And the reporting module was really like a game-changer in this. The ability to have that scripting data source means that we could incorporate a fairly specialised report into Ignition directly. We didn’t need to do. Do this anywhere else. It looks simple, but, man, this was a pain.

24:17
Speaker 1
We’ll take your word for it.

24:18
Speaker 3
Yeah. Goodness me. Yeah, so, yeah, in terms of understanding the parameters, this is essentially our. Where can we do our golden batch comparison?

24:26
Speaker 1
Okay. So the complexity was in the calculations and the associated data.

24:30
Speaker 3
Yeah, correct. Because there isn’t, the data isn’t immediately available within the controller. So we’ve got to do a whole bunch of. Yeah, just a whole bunch of stuff to get it going.

24:41
Speaker 1
And there is. These reports are they. Are they available as a PDF or downloaded via email?

24:45
Speaker 3
Yeah. So all of the reports that are on site are available ad hoc. There are no scheduled reports. Okay, we can do them.

24:54
Speaker 1
Obviously, Rajiv can claim the midnight reports like Mediclinic did.

25:00
Speaker 3
Yeah, Rajiv doesn’t go and generate the midnight reports, I believe. But the nice thing is that we can just go and generate the midnight report whenever we want. Yeah, yeah. So that. That’s exactly the flexibility we wanted. And then. Yeah, now we’ve gotten to packing. So in some cases, there’s operator error. So we’ve got two tables here where we’ve done the best effort guess on which work order these materials have been allocated to. And then also the ones that we miss, or perhaps there’s a user input error. We can then take this and create like allocate them to the correct work orders.

25:39
Speaker 1
See?

25:40
Speaker 3
And here’s an example. Now the MD overseas, he wants to see, he’s getting a complaint from a customer saying, Where’s my stuff? And now he can log on and look at this production tracker, go to their specific order and see. Okay, it went through this step, at this step, at this step and this step. And now we’re at this part of the process. This is actually the full sequence where we’ve actually gone from start to finish. But essentially, we’ve got this full traceability for that work order to say okay. It’s stuck over here.

26:11
Speaker 1
Very visual, granular view of exactly where in the process the order is.

26:15
Speaker 3
And, obviously, this is like a full order trail, so we can see who started this process, when, who finished that process, all of that kind of thing. So that does have different visualisations.

26:26
Speaker 1
Very nice.

26:27
Speaker 3
And then relating back to that, now if you want to develop additional insights to see. Okay, this thing is stuck in a blending state. Why are we stuck in a blending state? And this is just an example of their maintenance. So again, we’re looking at downtime. We then categorised the downtime per machine. You know, there are various contexts that we analyse or that we allocate into this. But yeah, just for that visual thing to see. Oh okay. We’ve had downtime in this area, or this machine, or whatever it might be.

26:56
Speaker 1
You also have the drill-down capability in this case?

26:59
Speaker 3
Not yet.

26:59
Speaker 1
Okay.

27:00
Speaker 3
But on the enterprise reporting side. Yes. So on site. No.

27:05
Speaker 1
Got you.

27:06
Speaker 3
Yeah.

27:06
Speaker 1
Okay, that’s quite a lot over five years.

27:13
Speaker 3
Yeah, it’s a whole lot of.

27:15
Speaker 1
A lot of integrations have been.

27:17
Speaker 3
Been taken on here.

27:19
Speaker 1
Okay.

27:20
Speaker 3
So it’s a very comprehensive system. Yeah, it’s intimidating to maintain, but thankfully, you know, we’ve got systems in place that are modular, and we’re able to, you know, maintain all of it. Yeah.

27:36
Speaker 1
So that was a brief overview, as you can imagine, of what the system is in the background. Maybe as a. As future expansion, future plan, any. Anything specific planned, Rajiv, or what else?

27:48
Speaker 2
Yeah, I think in our next phase we’re going to probably move into our warehousing. We are currently busy with the implementation of an ERP system, where we’ll need Ignition to feed into as well. So I think yeah, warehousing and probably Damian have some ideas for us as well.

28:06
Speaker 3
You know I’ve got plans, but I have to make the budgets available, but yeah, pretty much there’s the ability to now expand. We can basically solve any production-related problem.

28:21
Speaker 1
Yeah.

28:22
Speaker 3
With the platform that’s in place.

28:23
Speaker 1
Yeah.

28:23
Speaker 3
Once the.

28:24
Speaker 1
And once the data is consolidated and mapped and associated correctly where it should be, it makes everything else a lot easier as well. Fantastic. Thank you so much for sharing. Any. Any questions for Damian or Rajiv? So the question was that there is a lot of MES ERP-type functionality in the application. If that is embedded in the system part of another system, and how that will evolve. So that very good question probably should have clarified that right at the beginning.

28:51
Speaker 3
Yeah, so it’s actually, that’s because of that, it’s actually very difficult to make like manage the scope of these types of projects because technically Ignition. Ignition will be able to do all of this. But just because we can doesn’t mean we should. So there were, I mean, at a certain point there was a request to even do invoicing and things from Ignition and I had to say, whoa, we are not an ERP system. You know, like, okay, so you know, there needs to be sort of a distinguishing line to say this is where we hand over now to a purpose-built system that is passed, perhaps managed by a separate vendor or whatever it might be. But as far as all of the functions that we saw, everything is.

29:35
Speaker 3
Everything is currently being done by Ignition, and then there are certain handovers to other systems. But as and when the system grows, some of this function will be taken away, and then we will simply be integrating with that new endpoint, which will achieve the same outcome, but in a more automated way, maybe to add to the.

29:56
Speaker 1
The. The MES-specific functionality is. So you would have seen Sepasoft, which we haven’t really had time to spend too much time on today. I think we’ve alluded to and mentioned it. So they are being modular Ignition. Everything is available as a module that’s the same. The same applies to the Sepasoft MES module. So there’s essentially a batch module, an OB module, and a track and trace module. So, in a similar fashion that you would deploy a reporting module, you would have a module for very specific MES functions. That’s adding to that question. Does that answer your question? Any other questions? Okay, appreciate it, guys, thank you so much.