Welcome to the Trend Detection podcast, powered by Senseye, an industry leader in using AI to drive scalable and sustainable asset performance and reliability. This is a new publication designed to help you go away with ideas on how to achieve maintenance efficiencies.
In the third and final part of our series discussing what manufacturers can learn from the metals and mining industry, I’m joined again by Joe Carr from Axora. You can listen to part one here and part two here.
During the episode discussing what manufacturers can learn from the metals and mining industry, we're joined again by Joe Carr from Axora to discuss the maintenance practices metals and mining companies follow, why machine uptime is key and how having too much machine data is proving challenging.
Please subscribe via your favorite podcast provider (Apple, Google and Spotify) if you'd like to be notified about future episodes and let us know your feedback by leaving us a review.
Key topics covered (click to jump to the section)
- The key maintenance practices in the metals and mining industry
- Utilizing machine data in metals and mining
- AI and Machine Learning use cases in metals and mining
- Sustainability in metals and mining
- Case study: Alcoa
Niall Sullivan, Senseye: Yeah, that's it, Joe. That's interesting, you mention maintenance because that's where I was heading next and actually a smarter way to do it. Again, just thinking as a starting point, and you're probably going to say it's a case-by-case basis, but in general, what the main maintenance practices that are in place at the minute?
What do these metal mining companies follow? Is it reactive maintenance? Is it like you suggested they're changing every 4,000 days or whatever it might be? In general terms, what sort of practice are they following at the minute, in terms of maintenance?
Joe Carr, Axora: In the mining metals world, uptime is key. A hundred percent, uptime is the most important thing, because a lot like an airline seat, if I don't produce in that time, I can't ever get it back. Right?
Once the plane is taken off and the seat isn't sold, I can never get the seat back to sell. It's the same if you're running a refinery, right? If you're not producing metal, a ton of copper for an hour, then that hour's gone. I'm never going to get that production back.
Uptime is key. The industry certainly doesn't want to run a reactive. There's always going to be some level of reactive maintenance. It's always going to happen. Certainly as an industry, we want it to be the exception, not the rule. Apart from the reactive maintenance, which we don't want, you have planned and preventative maintenance.
There is definitely a mixture there. Again, like I said, it's company-by-company bases. We find that certain things, especially... For example, trucks. That's a great example. We have a lot of sensors today in trucks, that can tell us how the engine is operating, the brakes, and all that kind of thing. In those places, for instance, we can be more predictive in understanding how long a component will take to fail, so that exists.
Then, there are areas where it's just straight planned. We change that every 5,000 hours. Why do we change it every 5,000 hours? That's what the manufacturer recommends. Whether that's correct or not, is up for argument.
It's certainly a mix today. I think, ideally, the whole industry would want to run unpredictive. We would want AI, sensor data, and machine learning to tell us when things are going to fail, a hundred hours before they fail, so we can change them just in the nick of time.
There comes an issue there around planning. We can't fix everything just before it breaks. We have to have a level of planning, but there's significant cost savings and benefits to uptime by maximizing component life. In the fact that if I can keep that truck running 10 more hours than I wanted. And I can plan to take it down when I have an available space, rather than parking it up and queuing it for a day while I do something else, because that's lost production.
I think today, when I was working on operations, we tried to aim for the high eighties in uptime. 88% generally, was the kind of thing. With better predictive maintenance if we can eek that into the eighties and nineties, there're massive benefits. There're massive benefits to the mining company. There're massive benefits to society, in that we can produce more metals with the same amount of assets that we have. We don't need to open another mine. We can continue to produce from the ones we have, and be more efficient with the equipment we have.
Niall Sullivan, Senseye: There's a mix of strategies going on from what you're saying. You obviously touched on predictive maintenance, which a very clear interest from my point of view.
From everything, you said, it's clear that predicts maintenance would add quite a lot of value. How well is that being embraced them in? It could be that predictive maintenance is still bit of a growing markets and might not have taken off yet.
We mentioned about digital transformation before and some of the issues there, but is predictive maintenance and the benefits' widely known? Or, is it widely known and they're not embracing it for the reasons you pointed out before, to do with security the concerns and that side of it?
Joe Carr, Axora: Yeah. It is really widely known, actually. Maintenance solutions are some of the biggest in the industry.
Again, it comes down to a little bit because it's so easy to calculate the value. It's so easy to say, "What does 1% more uptime deliver to me?" The spares that the sailing in spares, it's always great, but it's always dwarfed into insignificance by their production.
If it cost us more to run the mine than produce the material, we wouldn't have a mine. It's not how the mining industry works, right, like any business. There's always massive benefits. I think the primary issue is it just been a haphazard development over the years, and certainly within the vehicles and the plant, it's the most forward-thinking. Especially, if you've got very well known, big yellow truck, right?
If you've got Caterpillar, Komatsu, or those kind of people, they've been doing predictive maintenance on their vehicles a long time, they know. When they're selling vehicles today, there are a lot of mines which buy their vehicles. They don't buy them. They buy them on an uptime.
They're purchasing a vehicle, with a minimum uptime of 87% or something like that. If they don't get that uptime, they're not paying for it, because they're leasing it. A bit like you might lease a car. They're heavily motivated. The manufacturers are heavily motivated to keep those vehicles running as efficiently as possible, because it's a huge differentiator when you're selling a truck, saying, "My truck is efficient." They've got such massive pools of data.
If you think of Caterpillar, you've got tens of thousands of trucks with sensors, all providing data about how those trucks run. They're just getting better at delivering what they're doing.
If you're a mining company and you've got a hundred trucks, your pool of data is so much smaller to be able to deliver those benefits. When they do implement their own systems, they're just not as effective. That's an example where preventative maintenance and planned maintenance has been driven by the manufacturer.
Whereas if you go in somewhere like the concentrator, the refinery, or the smelter, no single company sells a smelter. You don't go to Hatch or FLSmidth and buy this smelter. They build them from components from lots of different manufacturers. Then, they might build a digital twin around that, if it's a brand new smelter, over the last five years.
That's really where AI, predictive maintenance, and machine learning really has an opportunity, because it's not controlled by a single incumbent OEM, right? It's not a drill, which is made by... It's not a car made by Ford, where Ford knows what's going on in all the pieces.
Every mine refinery is different. Every smelter is different. Every concentrator is different, made by lots of different pieces from lots of different manufacturers. Being able to pull all that data together, understand, and deliver the insights that, today, we have available for trucks, drills, or shovels, that's where there's a really interesting opportunity. Then, the same exists across the whole value mining chain. Those are just two really good examples.
Niall Sullivan, Senseye: Is it because data's always a... Well, is it important topic from our point of view, because there's either... It's all the feast or famine situation, where there's no data or there's so much data in so many different formats. It's quite hard to take some time, at least, to get to those real insights in there.
Are metals and mining companies quite good at gathering lots of data around their different types of machinery and are looking at ways to exploit that? Or, do they not gather a lot of data? Is it the other side of things?
Joe Carr, Axora: From personal experience, they gather tons of data. They don't have a clue what to do with it.
We found that from our forecast, our innovation forecast, when we did the work. One of the biggest problems, was one, we need to be able to collect the data. That's a challenge because you need network to do that.
Once mines had data, it was in lots of different formats, in lots of disparate locations. One, they didn't necessarily didn't know what to do with it or they couldn't do anything with it, because they weren't able to. They didn't have the skills, as we previously talked about. The data is a real challenge, because these machines today... One of the tech guys I was talking to a couple of years ago from tech, he was saying, "A whole truck produces two gigabytes of data a day." And, they had 83 on the mine running around. He said, "The storage is full. It doesn't take very long." Right? You've got 83 things producing two gigs of data every single day to fill up the server. What do we do with this data?
Or 99% of it, we're not interested in. Why aren't we interested? We don't have anything to do with it. It was probably incredibly interesting data. We are tracking speed, cycle time, and engine temperatures off the variables. We're probably tracking 20 off the 500 variables coming off these trucks, in real time.
It's a massive opportunity for mines, because they have all this data and then it's silo. It sits on the mine, because of all the reasons we talked about before, about cloud infrastructure and how do you protect the mine from cyber attacks? We have all this data that just sits on a server that nobody does anything with, because nobody on the mine itself, knows how to do it. Nobody's employed to do it.
What am I employed, as a mining engineer to be on a mine floor. I'm not employed to sit through server data. I'm employed to break rock, get things out the ground, and put it through the processing plan. It's a mix, right? Very rarely do you go to a mine and find 20 data scientists working on the data. Right? What you do, is find a lot of geologists, engineers, production people, and drillers getting on and working on the mine.
That seems such a shame because like I said, it's such an untapped opportunity. Like I was saying, they don't have the resources internally to sift through that data. They're also a bit more security conscious when working with cloud or the software platforms, who might be able to help gain some real insights into it. It seems like a missed opportunity.
Like I said, the industry will likely change. People change jobs and generate... Maybe it's a generational thing as well. Maybe that will change in time and people will be bringing up these ideas thinking, "Well, there are loads of data. Why aren't we doing anything?" We really need to go out there and find solutions so we can make use of this data, for our benefit.
It is also a job thing, right? When I was a production engineer, what I really cared about, was the production. If somebody came to me and said, "Oh, can we run an AI project on your trucks or something?" My view would've been like, well, that kind of gets in the way of my production and my job.
You also see a disconnect where the people on site, kind of just want to get on with doing their job. And the people who are in the innovation departments, are really interested in doing the innovative projects, but they need to get the buy-in from the people from site. There's this disconnect where if I'm a production engineer, I am paid my bonus and my salary on production. I am not paid it, on running AI projects.
There can be a disconnect where the people on site are just like, "Look, I don't have time for this." That's also an interesting dichotomy, as well, within a mine. The mine is siloed itself. The production people in the mine, don't talk to the production people in the mill, who don't necessarily talk to the production people in the refineries and the smelters, who are producing the end metal. They're all doing their own thing. Likewise, because they're all doing their own thing and it's really important to them, if you come along with a maintenance thing, you might not get buy-in. They might say, "Ah, that's really valuable. I really like that." The response tends to be, "Come back to me when I can just use it immediately, and I don't have to spend time getting it integrated." It's that integration step and the hassle that comes with that. There's not a person there whose job that is to do that.
Niall Sullivan, Senseye: Like I said, again, it's to do with the missing skills' element there, I guess. Or, it's not been a known issue, is it? Are they aware of the skills' gap?
Joe Carr, Axora: Yeah, they are. I think this comes down to what we talked about earlier, with the machine learning, AI, and remote operation centers.
When people are freed up from doing the things that are repeatable, part of the planning, and the steps that they still do... A lot of people do by hand, that could probably be automated. When the people are freed up from those things, they will be able to look at things like AI and be able to dedicate the time and effort to those projects, which will ultimately deliver incredible value.
It's the, "Well, I'd rather have 10 pounds now, than 20 pounds next week." What would you rather have, right? Would you rather have the benefit now or later? As humans with terrible long-term thinkers, planning 10 years in advance... Actually, the mining industry is fantastic at planning 20 years in advance.
Every mining department has short-term, medium, and long-term planners. The long-term planning people, their earliest forecast is 10 years away. They're looking at stuff so far in the future. The short-term person's planning window is this afternoon. I think, actually, the mining industry really good at thinking long-term. It's just about how we help the whole value chain think long-term. And how do we free people up to be able to do that?
Niall Sullivan, Senseye: I know we've mentioned AI machine learning a lot and obviously from my perspective, I know from predictor maintenance standpoint, how that can really offer benefit. Aside from that, in whatever areas are AI and machine learning, being applied? What examples can you provide in metals and mining?
Joe Carr, Axora: Well, there's really... the list is as long as your arm, as with many things.
Niall Sullivan, Senseye: Maybe talk about the top five. No, the top... I don't know..
Joe Carr, Axora: Yeah. We'll do the countdown, right? Number five.
I think there're several areas. We talked about maintenance. I think maintenance is a really good one, because it's the fruit's so low hanging, it's on the floor.
There are some areas where as long as we've got the sensors to be able to do stuff with, I think there's so much in maintenance that can be done that, today, is either reactive or just planned. We changed the air filter at 500 hours. Why do we change it? Because it says so. I definitely think there're benefits there.
There's certainly benefits in the understanding of the deposit. Not go off down the geology tangent, given the audience of your podcast, but just understanding where the metal is under the ground. There is a lot of benefit to be able to using not the human brain in just understanding that, because you can have millions of data points that today we find it, we click, and we circuit it's here and it's here, and we build a shape, right?
There's so much benefit for AI or machine learning, depending on which algorithms you want to specifically look at, to be able to apply to those models. Instead of making one model a week, a machine learning algorithm could make hundreds of models in an afternoon, and then the geologist can assess them, and say, "Okay, that one makes sense. This one that, I don't know what it's done here. It's gone off crazy a tangent, so we'll just ignore that."
There's certainly lots of benefits around basic AI. We are taking AI here to dumb things. It's smart. It's very good at one thing and it's terrible at everything else. The AI around driving a truck, that's great at driving a truck, but you can't do anything else when machine learning learns what's going on by itself.
The applications of AI, within a mine to do things over again, those repetitive tasks. There're tons of that, from drilling of holes, to digging material out, to marking up faces, to loading explosives. All those things are ripe for AI, because we do the exact same thing every single time.
I go to drill a face for horizontal development. I'm using the same drill pattern, every time. A lot of the drill now, the drillers don't even drill the holes. They set it off and it drills it itself. They make sure it's putting the holes in the right places, but is there really that much extra step involved in saying, "Well, why can't they just get on with it themselves? Just have someone in a remote operation center keeping an eye on it."
There are benefits there, and certainly in the metal space, when we look at things like machine learning, in terms of understanding the product we're producing. How we are producing it? Are there defects? If it's a refined metal at the end using something like machine vision, so cameras. Essentially, a fancy way of saying camera. Using machine vision to understand the outcomes of the material and then who we're selling it to.
Ultimately, the utopia would be, you want a specific metal with some property. You place your order, it goes to the mine, and the mine dispatches the equipment to where it knows that material is. It mines that specific equipment. It goes through the concentrator, the smelter, and the refinery because it's got very specific properties and it's delivered to you, at exactly the specification you want. Then, you process that however you want, or you just use it in your project. Today, we are nowhere near that integrated supply chain. That's ultimately what could... And there's no reason why we couldn't do it today, it's just doesn't exist today.
Niall Sullivan, Senseye: No, that's really interesting.
I know we're coming to the end of our time, but I wanted to touch on that subject of sustainability, which is a hot topic.
What I wanted to know is, in general terms again, how is sustainability impacting metals and mining at a minute? I know we talked about a resource issue. Also, what measures are being put in place by companies, in order to meet targets? I'm sure a lot of manufacturers, metal and mining companies also have targets that they want to meet, as well.
Joe Carr, Axora: Yeah. The mining industry's actually doing quite a lot. You look at it today, there's a lot of projects out there using battery vehicles on mines. Hydrogen is also a big topic of conversation.
I think just in the news yesterday, the guy who owns Fortescue Metals in Australia, he's just bought the battery arm of Williams F1, to look at batteries.
He's an interesting character, in that he owns one of the world's biggest iron ore mines. Or, mining groups, should I say. He's building a solar farm for, one of a better term, in the Outback. He's going to use that to create hydrogen from seawater, to create green steel in Australia. So, to sell steel without a carbon footprint.
When you look at the mining industry actually, steel, that metals arm, is one of the biggest CO2 producers. They take the iron ore and to produce steel, you have to burn coke and coal, which is a special type of coal.
You've got two types of coal. Thermal, which is what everyone thinks of when they think of down pit, throwing the coal in the fire, and burning it in a power plant. That's thermal coal. Coke coal is a very specific type used to create steel. That's the source of much of the mining industry, except in the coal mining industry, obviously. That's the source of a large chunk of the mining industry, is CO2 footprint.
There are big questions about how do we produce green steel? How do we reduce the need for coal in steel production? Can we do it with hydrogen or even natural gas?
Although, the prices of natural gas today, that's also not a great sense at the moment, with the way natural gas is.
Sustainability's massively high on the mining industries to do list and on its needs. It's that interesting question of, how do we produce all these really important metals that everybody wants? How do we produce them with this limited impact? Understanding that the mining industry is by its very nature, an industry, which is taking something from the earth. It's an extractive industry.
The question of how do we do that better, that's been on mining companies agendas for a long time. And it will continue to be, the more people are focused on the environment, which is only going to get more. We're going to see more of it.
For me, the really interesting bit... And the double think to quote Orwell's Nineteen Eighty-Four, that being able to hold two things in your head, diametrically opposed and not understand the benefit of them, is that a lot of people don't like the mining industry. You say you work in the mining industry and people go, "Oh, whatever." They don't understand the industry.
They have this view that the industry is a terrible industry, the metals' industry. Yet at the same time, they also say, "We're in a climate crisis and we need green energy more than we need anything." It's impossible to hold two diametrically opposed views that you don't like the mining industry, but you want to do something about climate change. You can't have one without the other.
We cannot do that. I remember I was at IMARK, the international mining resources conference a few years ago now, when I was actually allowed to get on a plane and go to places.
Down in Melbourne and the extinction rebellion people were there. I've actually got a video of an interview on Sky News Australia. They had the head of the extinction rebellion protest on there. More or less paraphrasing what he said. He said, "We are currently in a climate crisis and we need to do something about it. If don't wake up the world is going to be in massive trouble." Then, in the same breath, "They're building a copper mine in the Philippines that nobody wants."
It's impossible to hold both views, right? You can't be against copper mines and against climate change. You can't have the green energy revolution, without the buying industry.
Niall Sullivan, Senseye: That's a really good sentiment to end on, actually. Just before we finish Joe, how can people find out more about Axora and what you do for them? For the industry?
Joe Carr, Axora: Yes, you can go to Axora.com. You can view the marketplace and you can contact us through that. Or, you can find me on LinkedIn and drop me a message. I'll be happy to redirect you to whoever the appropriate person is within the company, to talk to about whatever it is that you want to talk about.
Case Study: Alcoa
Alcoa Corporation is a global leader in bauxite, alumina, and aluminum products, built on a foundation of strong values and operation excellence dating back more than 130 years to the world-changing discovery that made aluminum an affordable and vital part of modern life.
Alcoa operates production plants worldwide and have applied breakthrough innovations and implemented best practices that have led to increased efficiency, safety, sustainability, and stronger communities wherever they operate.
Discover why Alcoa partnered with Senseye to achieve best-in-class technology and operational practices for Predictive Maintenance.