Understanding AI in Cement

In this episode of the World Cement Podcast, David Bizley is joined by Scott Ziegler, CEO of CemAI to explore the role of AI & ML in the cement industry. This episode covers everything from changes in the control room, to dispelling fears around personnel replacement, and exploring the importance of adequate cybersecurity.

 

Topics covered include:

• The impact on operations and cement quality
• Key factors to consider when adopting AI
• Countering the myth of job losses – how AI can enhance personnel rather than replace them
• The importance of cybersecurity in the modern cement industry
• And more...

As always, make sure to rate, review and subscribe. And, if you haven’t already done so, why not check out our back catalogue of episodes from other leading players across the cement industry?

This episode of the World Cement Podcast is sponsored by CemAI and their new process optimisation solution: CemAI Process Optimiser. Combining the power of AI & Machine Learning with cement-specific expertise, CemAI Process Optimiser enables a step-change in production volumes, energy efficiency and consistent quality output.

Find out more at cemai.com

Creators and Guests

Host

David Bizley

As well as the day-to-day editing of content and working with article authors and advertisers, he is actively involved in the commissioning of material for both the magazine and its expanding online presence.

Guest

Scott Ziegler

CEO at CemAI

What is World Cement Podcast?

The World Cement podcast: a podcast series for professionals in the cement industry.

David Bizley: 00:00

Hello everyone, and welcome back to the WorldCement podcast. With me, your host, David Busley, senior editor of WorldCement. The era of AI is well and truly upon us, with everything from large language models like ChatGPT helping people with everyday tasks, through to advanced machine learning systems capable of optimising complex industrial processes, like cement manufacture, for example. Indeed, it seems like there's some form of AI involved in almost every part of modern life, so it's a good thing that our guest for this episode is none other than Scott Zeigler, CEO of SemAI, who joins us to help explore the role of AI in cement. Now, before we get into today's episode, we have a brief message first from

Advert: 00:42

Optimising your cement plant. Empowering your team. SemAI presents the expansion of its AI and machine learning offerings by providing a new process optimization solution SEM AI Process Optimizer. Combining the power of AI and machine learning with cement specific expertise, CEM AI Process

David Bizley: 01:10

enables a step change in production volumes, energy efficiency, and consistent quality output. Predictive maintenance and real time optimization are now in your hands. Find

Advert: 01:25

out more at cemai.com.

David Bizley: 01:31

Scott, a very warm welcome to the World Cement Podcast. Glad to have you with us today.

Scott Ziegler: 01:35

Thanks, David. I appreciate you having me here.

David Bizley: 01:37

Now, before we get into some of the more technical questions we have for you lined up today, let's start off with a bit of an overview. Tell us about Who are you guys, and what are the origins behind the company?

Scott Ziegler: 01:48

So CemAI was created and developed back in 2022 when the Titan Cement Group decided to commercialize a number of their digital solutions that they had been developing for a number of years. Their main focus at the time was on plant maintenance, process optimization, And so the first solution that was commercialized was the plant maintenance side of things in the cement industry. And then further on from that was plant optimization. They looked at logistics, quality control. But the main focus of the commercialization at the beginning was plant maintenance and process optimization.

Scott Ziegler: 02:26

And so over the years we've been working to improve the algorithms, all of the facets of each of the solutions that we've been commercializing over the years.

David Bizley: 02:34

Excellent. Bill, one of the big selling points behind AI is that it can completely change the way a business operates. So when it comes to cement, how much does an AI or machine learning control system actually change the existing control room?

Scott Ziegler: 02:49

It's important to remember first and foremost before we dig into too much on AI and visualization is the cement industry is long known as a very slow moving, slow to change industry. High investment, a number of companies that have been around for a number of years. So it's a very slow to change industry. So when you start talking about things like digitalization or artificial intelligence, machine learning, in an industry like the cement manufacturing industry, it takes a lot. It takes a lot of time for acceptance, and it takes a lot of time for folks to get used to it and to bring that knowledge in there.

Scott Ziegler: 03:24

I think it's important though to remember that adding these tools are the same type of things that we've looked at over the history of the industry: automation, the use of PLCs and programmable logics, fuzzy logic. As everything has progressed, we've just gotten to the point now the way the industry has moved is we're moving into that age of digitalization. And so it's important also to understand that the AI and machine learning aspects of the business can touch every portion of it. It can improve everything, as I was mentioning earlier. Everything from maintenance to logistics, safety, quality all aspects of it can be enhanced by digitalization.

Scott Ziegler: 04:03

When we talk about digitalization, obviously the default position is we start talking about artificial intelligence and machine learning.

David Bizley: 04:09

Okay, so it's a whole kind of manifold array of benefits that can be derived across the process. On that note then, how long would it take to convert a traditional PLC or other system to an AI or machine learning system?

Scott Ziegler: 04:24

Well it depends obviously on a number of factors at the plant, and how, I guess, what we would call digital ready a facility may be. A lot of the installations and a lot of plants are actually already ready to start working towards digitalization. There is already a strong enough digital footprint to start making the changes now. When we think about how does someone get started, most plants right now have data historians and data collection capabilities, either through their current setup or in the basic control systems that they have in place. There's a lot of the data that's required.

Scott Ziegler: 05:01

And then we use the knowledge, and you have to use the knowledge of the folks that are already at the cement plant, the people and the employees at the cement plant as well. Right. So if you can't really combine the knowledge of the plant and the data that's already available, it really doesn't take a lot of time to physically set it up. I think the biggest hurdle you have on time is the acceptance of this is a change we want to make, and getting the buy in from both the plant and the industry that now is the time to do it.

David Bizley: 05:26

Okay. And speaking of time, Semi AI has made some claims of a fairly quick return on investment when it comes to adopting AI. What kind of returns can producers expect to see, and over what kind of timeframe?

Scott Ziegler: 05:38

It generally depends on what type of solution you're looking for on the digital side of things. Clearly, if you were making changes that impact productivity on tons generated or tons produced, Knowing the value of cement and getting a very small increase in productivity and production is a big financial impact of the business. When you look at logistics, saving on fuel and saving on wear parts on trucks and lorries is a huge number as well. When we think about things like plant maintenance, keeping repairs and replacing parts to a minimum is a big impact as well. I can tell you that generally speaking it should be accepted that something of payback that SemAI has is usually less than six months on the predictive maintenance side of things.

Scott Ziegler: 06:24

And so you're really looking at a very quick payback on a solution that can make a big impact very quickly as well.

David Bizley: 06:31

Okay, and speaking of sort of big impact things as well, does an AI production system impact the quality of the cement being produced by the pump?

Scott Ziegler: 06:39

So it can. Again, we mentioned about a number of different solutions. I think the one thing to take away right now is that we are still early on in the life of digitalization and AI in the cement industry. There's a number of producers that are faster to accept, that have focused on different areas of digitalization, artificial intelligence. And the other point to make here is that there's not one holistic solution that covers everything at one time.

Scott Ziegler: 07:05

There's not a single program that covers everything from maintenance to logistics to quality and everything else. There's a number of solutions that are off the shelf that do impact quality, that impact productivity, that impact the maintenance, and impact everything else. When you combine each of those together, you're really starting to get a full digital look at what a plant is. Generally speaking, the more stable the plant runs within the conditions that it needs to run without upsets, whether it's a mechanical breakdown or an operating condition, the better quality you're going to have at the end of the day. And so combining everything together is really the I think when you start thinking about the cement plant of the future and where it is as a fully digitalized plant is really the future of where we're going with this.

David Bizley: 07:48

Okay. Another topic we sort of briefly touched on a little bit earlier about mentioning the employees at the plant. Which kind of plant personnel should make up the team that needs to be in place when you're applying AI to a cement plant?

Scott Ziegler: 08:00

Yeah, it's interesting that the first thing to say is we're not staring at the future where machines are taking over people and employees' jobs. I'll start with that. Because it's important to understand that these are tools, right? Each of these things that we're implementing, each of the digital solutions that are implemented are essentially better tools that we've seen progress over time. And so they still need users.

Scott Ziegler: 08:24

We still need end users who are improving the information, improving the data, doing the work. The days of the process engineers and control room operators are not going away They're simply enhancing their abilities on the day to day side of things. On the implementation side, it's actually quite simple. You really need focus from the automation and IT department, so you'll need some help there obviously on the connectivity and the digital footprint of the plant. And then with anything that's newly introduced to one of the manufacturing facilities, you're looking for a plant champion, someone to drive this to really shoulder the work that's being done and champion it at the plant.

Scott Ziegler: 09:03

On a day to day side, the maintenance inspectors use the plant the predictive maintenance tool. The control room operators are using the process optimization tool. The quality control manager is using the QC tool. Your transportation manager is using the other ones. So it's really there to make their job easier, to give them more data at their fingertips, to give them some help in making decisions so that their jobs are much more efficient in the long term.

David Bizley: 09:28

Okay. And sort of following on from that, I mean, joked about it at the beginning of that, that one of the most common concerns you hear about AI in the public discourse is that, you know, it could take people's jobs. How do you manage that concern and keep a producer's personnel engaged and excited to adopt this technology that will actually make their lives better?

Scott Ziegler: 09:46

Sure. There's a big move afoot. Obviously the industry as a whole, I think you're seeing in areas across the globe, there's a significant amount of retirements and folks leaving the workforce, and that's just natural progression. I don't think there's anything fundamentally different in the industry. And so we also have a new generation of engineers, a new generation of folks who are much more in tune with things like AI, with the computer, have grown up with the mobile phone and mobile applications and all of the bells and whistles that come with that.

Scott Ziegler: 10:17

And so when you think about that, they use a phone as a tool, the same way that we would think of turning a wrench somewhere. So if it's another option for them, it just enhances what their job is there on the plan as well. The other side of the coin though is that it's important to know that we can detect a change in something like a bearing on the maintenance side of things. It still takes someone to go out there and physically make the change on the bearing, change out the bearing, do an inspection, verify what's going on. It's the same thing with the process.

Scott Ziegler: 10:50

The optimization can tell you to increase certain feed rates and different controls, but if those controls aren't operating properly at the plant and they're not working right, you still need that person to verify. It's that verification side of it that's still very important across the process of the plant and really won't it's not expected to change. The future may be things like inspections, drone inspections, some robot inspections in areas of the plant that we haven't been able to access before, right? Confined spaces, areas that are unsafe to enter. But we're not going to be replacing people with the artificial intelligence.

Scott Ziegler: 11:23

These are tools to enhance the day to day work and the day to day jobs.

David Bizley: 11:27

So it's more about empowering the personnel that are there rather than getting rid of them.

Scott Ziegler: 11:32

Exactly. Exactly.

David Bizley: 11:33

And so who are the typical decision makers when it comes to onboarding products like AI?

Scott Ziegler: 11:40

Well, need two there's two camps there, and we need them both to be aligned. One are the people who will agree to spend the money for the installations and for the digitalization change, and the other is the group of people who see the value in it on the day to day side of things. The purse holders will always make a part of that decision as well, but the plant personnel who can really find the value in enhancing and making their employees at the plant much more efficient in what they do on a day to day. Moving out of the world of firefighting with maintenance or constantly having five different set points on an operating because you've got five different control room operators. Looking for stability across the plant really makes everything work better.

Scott Ziegler: 12:25

It makes the plant more efficient, it makes productivity better, it gives people the opportunity to focus on other things in the plant that may need focus. So again, I think it's a good way for the plant to enhance what they've got and to make the most out of the employees that they have and make them much more efficient.

David Bizley: 12:41

Excellent. Now in The UK recently, and I'm sure this has happened elsewhere, but there have been several high profile cases of pretty major businesses like Jaguar Land Rover falling victim to cyber attacks. They've had to temporarily shut down production, costing potentially billions a pound. What kind of cybersecurity or data governance measures are required to ensure safe and reliable AI deployment in a cement plant environment?

Scott Ziegler: 13:07

Dave, it's one of the most important things that we start talking about with potential customers at plants on day one, which is really what is the IT infrastructure in place, both the IT and the OT infrastructure that's in place, and how can we enhance what may already be in place. Because we are talking about looking at terabytes of data on an annual basis per plant. It's a significant amount of data, and it's a significant amount of investment made with that data in order to improve it. I'll tell you there's still a number of ways for implementation. We all hear about the cloud and what the cloud may or may not be.

Scott Ziegler: 13:44

There's opportunities there with the cloud, but in that one, you've got some very strong built in encryptions. Cloud based servers are not less secure. They're actually quite secure with encryption and threat monitoring. On premises, it's important to know that we have things like additional firewalls. Any of the algorithm work or any of the AI machine learning algorithms and software are usually isolated from the plant's general network.

Scott Ziegler: 14:10

We also have, when it comes to being able to access the system with that remote access, we have everything from encrypted VPNs to IP whitelisting, so there's a way to do that. And again, the data, the algorithm, they live on a restricted environment. It's not as if it's sitting out on a general server where it can have access to it. Now, all that being said, there will always be threats, and so we do the best that we can in order to minimize that. I think the opportunity to understand that different types of AI do different things in the plant as well can certainly increase and decrease the risks when it comes to it.

Scott Ziegler: 14:48

But above and beyond, the basics are really the plant specific ITOT infrastructure as well.

David Bizley: 14:54

Okay, excellent. Well before we wrap things up for this episode then, do you have any final thoughts for our audience?

Scott Ziegler: 15:01

I started to say this about six months or so ago. We've been, like I said, we've commercialized SemAI. We're going on this will be our fourth year in 2026 And I think broadly speaking, some of the colleagues that I have in other areas of AI and machine learning in the industry, we always discuss the long lead time, the decision, the long decision it takes for someone to make the decision to go forward with it. But in every case, once the plan has taken that first step, it really starts to see the value of it and gets excited about the opportunity. We've got a pretty famous footwear company here in The US, and their tagline is just do it.

David Bizley: 15:40

They're pretty famous everywhere, to be fair.

Scott Ziegler: 15:44

I think it's, you know, that's it. Take the first step. The opportunity is there to learn. The opportunity is there to continue to enhance. I think there's two things that it leads to.

Scott Ziegler: 15:54

One, it's improving the day to day. It's improving the plant's opportunity, and it's enhancing the company's ability to improve their own, right? If you start early, you can help start to drive some of the decisions made in AI and the machine learning world for the industry. And it's better to be out front. So it's just get started.

Scott Ziegler: 16:12

Take the first step, just get started, and the risk is very low compared to the reward.

David Bizley: 16:18

Okay, excellent. Well, that is all the time we've got for this episode of the World Cement Podcast. Scott, thanks so much for your time today and for sharing your insights into this exciting new frontier.

Scott Ziegler: 16:28

David, thank you very much for the time.

David Bizley: 16:30

Thanks again, Scott. And of course, thank you to every one of you in the World Cement Podcast audience as well. I hope you enjoyed this episode, and if you did, please make sure to give it a good rating and subscribe if you haven't already done so. And while you're at it, why not go back and check out some of our previous episodes as well. Thanks again and goodbye for now.

Advert: 17:02

CemAI presents the expansion of its AI and machine learning offerings by providing a new process optimization solution SemAI Process Optimizer. Combining the power of AI and machine learning with cement specific expertise, CEM AI Process Optimizer enables a step change in production volumes, energy efficiency and consistent quality output. Predictive maintenance and real time optimization are now in your hands. Find out more at cemai.com.