As artificial intelligence continues to reshape manufacturing landscapes, the feed production industry stands on the cusp of a technological transformation. Matt Boland, sales territory manager at CPM Automation, joined the Chat to explore how AI is poised to revolutionize feed mill control systems, shifting from rigid, step-by-step programming to more dynamic, intelligent decision-making processes that could dramatically improve production efficiency, optimize scheduling and potentially pave the way for fully autonomous "lights-out" facilities.
Watch CPM's Matt presentation, "AI-powered feed mills: Paving the way to smart, efficient production," at the 2025 Feed Mill of the Future Conference.
Transcription of interview with Matt Boland, sales territory manager, CPM Automation
Jackie Roembke, editor-in-chief, WATT feed brands: Hi everyone. Welcome to Feed Strategy Chat. I'm your host Jackie Roembke, editor-in-chief of WATT feed brands.
This edition of Feed Strategy Chat is brought to you by the FMOTF Conference. The half-day event will bring together leading feed industry experts to examine emerging feed mill technologies poised to impact animal feed manufacturing. It will be held on January 28 at IPPE 2025. The conference, produced by Feed Strategy and Feed & Grain, is organized in partnership with the American Feed Industry Association (AFIA). To learn more about the 2025 edition of the FMOTF Conference, visit www.feedmillofthefuture.com.
Today, we're joined by Matt Boland, sales territory manager with CPM Automation. He's here to discuss how developments in feed mill control systems are helping to optimize production efficiency.
Hi, Matt, how are you today?
Matt Boland, sales territory manager with CPM Automation: Doing great, Jackie. How are you?
Roembke: I'm doing well. Thank you for taking the time to be here. Well, let's get right into it. Over the last 24 months or so, artificial intelligence, or AI, has gotten a lot of attention across manufacturing and pretty much in every industry. How do you see it transforming traditional manufacturing control systems?
Boland: Yeah, that is kind of the million-dollar question right now, right? Everybody's kind of talking about AI. You don't go to a conference now and not see somebody presenting on AI and automation in manufacturing. But I think if you look at history, there's a pretty clear determination of where it's going. You know, the first century BC, automation really kind of took off to take over some of the manual labor that was pretty intensive for the individual workers. Then we got to the industrial revolution. At that point, it was more about mass production and quality control. We got to the digital revolution, and that was about interconnecting all these disparate automation systems and sharing of data.
Now, with AI coming in, there are two very unique aspects to it. One is the generative aspect of AI, and then the machine learning aspect of AI. And so I think if you really look at history, and put AI into that picture, what you would probably find is that what we struggle with most today in automation is efficiency. A lot of times, you'll hear the expression OEE, or overall equipment efficiency. It's something that people have struggled with for many, many years. The nice part about AI is it does take away some of that decision-making responsibility from the individual people in the manufacturing facility.
I think where AI is really going to stand out is alleviating some of the decision-making that the humans actually have to do in the manufacturing environment. Right now, with the control system, we essentially tell the equipment how to do everything, step by step by step by step, with AI. I think the real improvement comes in that we are going to be able to tell the system, 'This is what we are hoping for,' and allow the system to make those decisions, to run the manufacturing process as optimally as possible.
Roembke: What are the challenges when it comes to feed manufacturers looking to successfully integrate AI into their production systems and production environments?
Boland: First of all, understanding AI. You know, AI is kind of a term like IoT (internet of things) that we've used for years, encompasses a lot of things, but we don't necessarily know yet where we're going to place that. And there are companies out there who have made investments in companies like Nvidia and other AI companies, and they are in the process of developing that. But I think right now, the biggest challenge for manufacturers going into the automation space is going to be computing power.
No. 1, most likely the AI applications that are coming up are going to be software based, and not necessarily rack-mount PLC based. So it isn't something that's going to reside inside of your control cabinet as much as it's going to be on a server somewhere. So I think manufacturers are really going to have to be cognizant about their plant-level networks, their servers, their PCs, the computing hardware that's out there, but then also the operating systems.
As these things do roll out, I imagine that, you know, everything's probably going to have to be up to date in order to run the AI system. First and foremost, I think that's something that manufacturers are going to need to look at. And second of all, something that most people don't think of is sensing. The sensors that we have out in the field are incredibly important to the manufacturing process. If you think of the PLC like a brain, the sensors are the eyes, the ears, the mouth, the nose, the skin. It's everything that we get from the outside world.
In order to process what we need to process, in order for AI to be effective, we need to be able to give the PLC as much information as possible about the outside world so that it can make intelligent decisions about what to do in the manufacturing space. I think that's one area that's probably overlooked a lot, is sensing technologies and making sure you have an adequate number of sensors, but also that they're placed in the right position.
Roembke: Please provide a practical example of how AI and sensors and the overall system would dynamically optimize an aspect of production or make something more efficient within the mill.
Boland: I think AI has some incredible capabilities. The possibilities are really endless. I think, you know, eventually we look at lights-out facilities and things like that. But I think where it's going to come in most effective is in collaboration of data. You know, right now we have data that's in our business system and our enterprise-level systems, and then on our plant-level systems, we have other data. That data is typically shared by databases or something. But I think that given the generative aspect and machine learning aspects of AI, when you start to combine those things, you really allow the control system itself to make a lot of the decisions on how to make the process efficient.
So think about production scheduling right now. You know, typically we have orders coming in, and then we need to fulfill those orders by a certain date. Those orders have multiple SKUs, so we've got to produce multiple products. We've got to produce them by a certain time so that we can make delivery and, right now, the way that we do that is, basically, it's ingrained knowledge. We know that to make Product X, it takes this much time, and Product Y takes this much time. So we, using our human minds, we try to figure out the most efficient way to get that done on time. And where I really see AI coming in is kind of taking that enterprise-level data, and presenting it to the control system and saying, 'Look, we need to have this much of these products by this date,' and then allowing the control system to say, 'Oh, OK, well, I know how to most efficiently produce those products by those dates.' So it can say, in between production of different products, you typically would have shutdowns or product changeover times, maybe some maintenance that needed to be done, that kind of thing. But the AI, can actually, it can kind of figure all that in, right? And it does it all instantaneously, whereas a human being, you know, we sit there and drink a cup of coffee, think about it for a while, usually just try to make our best guess.
But yeah, I really see that integration of data, that the decision-making in the automation process coming down to AI to say, 'This is the way that you guys should really do it, because this is the most efficient.'
Roembke: You mentioned lights-out operations, or the fully autonomous facility, earlier in the conversation. What does that look like? How far out is it before it becomes a reality?
Boland: It's a good question. There have been some plants overseas that have attempted it, but it’s essentially something that has to be built from the ground up. You have to go into the design of a plant to say, this is going to be an AI plan, that that's going to be lights out, because without any kind of human interaction inside of the plant, you're going to need to figure out how we're going to receive product and put it in the warehouse space, and how it moves from the warehouse space to the production space, how far out we are on that one. I don't know. I really can't say for sure.
I think, in the in the short term, where AI is going to come into plans is probably going to be data analytics, because it's sort of innocuous. It can take the data that you currently have and essentially create what you could call a digital twin of your system and apply some algorithms to that to say, 'Hey, I think if you did this, you might run a little bit more efficiently, and they can make those sort of suggestions. And you can implement those changes or not. But it's not a terribly intrusive application right now. Eventually, I do think that is coming, where we'll be able to just tell the system, 'Hey, have you run at it?' But I think starting off, people are going to be a little reticent about allowing the computer to do everything for you.
Roembke: Yeah, I can see that for sure. Now, what incremental steps could a feed manufacturer take to prepare their facility, their long-range planning, their future investments to start leveraging AI in their control systems moving forward?
Boland: I think you know, as AI is developed in the manufacturing space, there are going to be some things that need to be done, right? It's not going to be as simple as plugging in a black box and your AI just jumps in and says, ‘Hey, I know what to do with your plant.’ Wish it would be that easy, but I think you know to get prepared. It's really going to be about making sure that the system that you have right now is as up to date as possible. I run into a lot of applications where a customer says, ‘Hey, I just want to implement this new technology right over here, right?’ And that's all well and good until you start digging into it and finding out the network doesn't support it, the servers don't support it, the operating system doesn't support it. And what the customer thought was going to be a very small job ends up being a very big job.
If the customer does want to be prepared to start implementing AI I would, at this point, make sure that your servers and your OT network are up to speed and running optimally, and then everything that you want to automate, obviously, is on Ethernet.
Roembke: Excellent. Thank you so much, Matt. Now, if you are interested in finding out more information about some of the topics and applications that Matt discussed here today, consider joining us at the feed mill of the Future Conference on January 28 at IPPE, where Matt will be presenting on the topic.
For more information on the Feed Mill of the Future Conference. Visit www.feedmillofthefuture.com. Thanks again, Matt, and thanks to you for tuning in.
THE FEED MILL OF THE FUTURE CONFERENCE IS SPONSORED BY
