Navigating Predictive Maintenance in the Heavy Industries - with Johnathan Bonner
Show notes
Welcome to the Trend Detection podcast, brought to you by Senseye Predictive Maintenance – which gives you visibility and insights into all your assets, from single machines to full plants to help you reduce downtime, increase knowledge sharing and accelerate digital transformation across your organization.
• Why predictive maintenance is crucial in heavy industry environments, and how it differs from traditional maintenance approaches - from preventing catastrophic failures to optimizing machine performance
• The real value of existing data in heavy industrial settings, and why organizations often underestimate what they already have while overestimating what they need
• How predictive maintenance directly impacts workplace safety in heavy industry, with real examples of how monitoring and early detection can prevent dangerous situations
• The rapid return on investment (ROI) potential in heavy industry applications, including examples of companies achieving ROI within weeks and saving millions through single preventive actions
• The role of AI and copilot technology in democratizing maintenance knowledge, breaking down language barriers, and preserving crucial expert knowledge as experienced workers retire.
You can find out more about how Senseye Predictive Maintenance can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants, by visiting: www.siemens.com/senseye-predictive-maintenance
Transcript
But it's also monitoring machines, equipment 24/7. So continuously gathering data, information about The condition of the machine. But what we're finding or or what has been a problem within heavy industry is that without visibility of these machines' condition. They are running to failure. Now, a human being wouldn't run a race in a desert without water, without support, without medical advice or check-ins. And this is actually, if we use that analogy, what maintenance teams and heavy industries have been doing, running to failure. But with predictive maintenance, it gives them insight into the condition, into the signs of degradation so that A specialist like a doctor can tell you when to slow down, when to hydrate, when to rest and recover, or for maintenance in this case. Welcome to the trend detection podcast powered by Siemens, and now your host, Nel Sullivan. Hello and welcome to the latest episode of the trend detection podcast. Um, we're in a new location today. I'm really excited in the studio today. Joined by my colleague, Jonathan Bonner, who's been on many times onto the podcast. Welcome back, John. Thank you for having me. Awesome. Um, today we're talking about heavy industry, um, and, and real world winds with predicting maintenance in the heavy industry. But I guess before we start, heavy industry is quite a broad term. Could mean a lot of things to, to lots of different people, so when, when I'm saying heavy industry, John, what, what does that mean to you? What, what sort of, um, let's say subverticals are we're talking about here? When you think of heavy industry. You can think of it in two ways. Either the machinery used in this process, manufacturing are heavy, big, extremely large, the size of warehouses, um, tall buildings, or what they're producing is heavy, big and large. So typically, heavy industry contains heavy stuff, makes heavy stuff. However, when you look at the subverticals, you're thinking of metals, mining, cement, those type of uh manufacturing environments. And in terms of, I mean, I think we know across all industries actually predictive maintenance is becoming a strategic priority, but why specifically in, um, heavy industry is it becoming so important? Well, if you think about predictive maintenance. It's a decision support tool designed for maintenance personnel. But it automates the function of a condition monitoring engineer on site. So the data collection, analysis and reporting are now automated. But like any good condition monitoring engineer, it continues to learn about failure modes, failure patterns, and can be quite precise in what it's recommending for maintenance actions and corrective actions. But it's also monitoring machines, equipment 24/7. So continuously gathering data, information about The condition of a machine. But what we're finding or or what has been a problem within heavy industry is that without visibility of these machines' condition. They are running to failure. Now, a human being wouldn't run a race in a desert without water, without support, without medical advice or check-ins. And this is actually, if we use that analogy, what maintenance teams and heavy industries have been doing, running to failure. But with predictive maintenance, it gives them insight into the condition, into the signs of degradation, so that A specialist like a doctor can tell you when to slow down, when to hydrate, when to rest and recover or perform maintenance in this case. So really extremely important in all industries, but specifically in heavy industry. So why is running into failure so common in, in heavy industry? What's, you know, sort of brought that mindset on, if it is a mindset thing or a cultural thing. I, I wouldn't say it's a mindset or a cultural thing. I think it's got to do with availability, availability of time, resources, skills, knowledge about what's happening or visibility of what's happening. Quite often. Environments such as heavy industry, your metals and mining have been set up with visibility on the process. What product is coming out, at what rate and at what quality. But very little thought has gone into The maintenance or understanding the condition of these machines. So, you'll find that often data is there, the ability to have visibility is there. However, it's just not catered for in the initial designs or the initial operations, and now after 2030 years of running, they've done it in a reactive way, day in, day out for decades, and they haven't had a reason to change, whereas technology nowadays enables this change. And you mentioned about data before, and there's obviously, you know, again across all industries, but in heavy industry as well, there's, you know, data is being collected, but are you saying maybe it's not being utilized in, in the right way at the moment, it's not being explored, the insights aren't being gathered from that data. Absolutely, so. At the moment, what we see is it's a lot of the data that exists on a site is geared for process control or understanding of production. Perfect, that has huge amounts of benefits, and that's great. But what you need, especially for predictive maintenance, is condition data. So condition monitoring data such as pressure, currents, torque, vibration, temperature, speed, flow, cycle time, whatever is relevant to identify the normal condition of an asset. Now, some of those are available within PLCs, drives, inverters from a customer's environment existing already. But a lot of them haven't been mapped in, so there is no record or historian of this data, and there's no analytics being employed on this data to gain further insight. But that's where predictive maintenance comes in. It's about using your existing data along with additional data that can either be mapped through your OT environment or supplemented with additional sensors so that you can get a centralized visibility of an entire plant's production, the signs of degradation with then recommendations on what maintenance should be done, when it should be done, um, and even who should be doing it. No, exactly, exactly. And I, I guess to your point as well, because we do like to say, you know, use existing data where possible. When there is a requirement to, you know, add additional sensors and bring in other types of data. I mean, I can only imagine in a heavy industry environment that's quite challenging. Um, well, I mean, it's a question for you, Joel, actually, is about, you know, is it challenging to collect data in such a sort of a rough, you know, potentially rough environment. Actually, no. Actually, no, yeah. No, but although it hasn't been mapped through, the data exists. The systems are in place already that are gathering the data, which are controlling, uh, the machines and that have these forms of data in it. So the data we find exists. It's there. It's really about how to access it. And some of the challenges we're finding with that is really around how a company has grown or let's take a steel plant, for example. A steel a steel factory could have many different plants inside of it. They could be different ages. They could have different OEMs, uh, responsible, um, different controllers, different communication pathways. So really, getting access to the data can be a bit of a challenge, and has been a challenge. But we do live in an age where gateways and getting access to the relevant data is actually relatively easy. There are single devices that can speak to a range of OEMs, can gather the relevant data, can do the data processing that is required, and can push it to the cloud in a very secure way, um, practically risk-free, um, for a small adjustment, huge amounts of value. And in heavy industry, the data that is existing already. It's there. Um, And what we're finding is the addition of, uh, additional sensors and hardware is actually not as much as customers would, would think to get started. So we can drive a lot of value from existing data, existing infrastructure, um, that is already there. Yeah, I guess it, is it one of those things where some, sometimes there's a, you know, underestimating the value within existing data already, and maybe the other way around as well is like overestimating the value in that. Is that something that you often come across those are very contrasting scenarios. I think, uh, the scenario that we come across the most is overestimating the data that that that that is available. Um. Yeah, quite often in in our investigations, the customer will say, yeah, we've got plenty of data, loads of data, no problem there. And then when we come down to actually investigating what they have, as far as time series data, continuous measurements of a, a parameter, like currents, like torque, and so on. That that isn't necessarily in place. Again, they'll have a lot of process related data, not necessarily conditioned data. However There is an underestimation of how much value can be derived from what is existing on site. So when we really do start to investigate it and we look at the possibilities for additional data mapping, for adding in a data gateway that can speak to OEMs or controllers and make the data useful, really, we have a lot of value and a lot of benefit from that existing data already. So part of my role within Siemens and Sensei. is to assist with those investigations. To help identify what data exists, where does it exist, how do we get access to it, and how can we process it and get it pushed to the cloud, so that we can use it. Well, I guess another angle on this as well, because we talked a lot about data and collection of data, but the other side of it is the human element, the, the knowledge that's often locked in people's heads or in manuals and things like that. And so how, you know, in the heavy industry, how do, how can they extract that sort of information out and utilize it in a sort of a predictive maintenance deployment. I guess that expert knowledge comes in a range of forms. The first format which is actually being extremely useful for us now is expert documentation. OEM manuals, maintenance manuals, decision trees based off of experience over many years. Written documentation on uh machines, what to do next, Pro, uh, problem shooting and so on. Troublesshooting rather. So we can load that into the application and that can be used and referenced directly using our Sensi maintenance copilot. But the other form of knowledge and experience that we want to capture or at least feed into the application is the human knowledge, the stuff that is inside of their brains. The stuff that has taken decades and decades to, to learn. And that comes in the form, or at least for us, in maintenance, uh, events information. Tracking what was done, what the problem was, what their, their process of fault finding showed, um, who was involved, you know, tracking. And making markers of their knowledge inside of the application. So using documented knowledge and the person's actual actions and feedback of their actions to unify that so that it can be made ref or so that it can be referenced ongoing in the future. And I wonder, in just the technology context, we know how valuable, you know, knowledge of people who know the machines like the back of their hands have worked in the industry for a long time, but when it comes to a technology deployment, is it sometimes forgotten that actually your knowledge is just as important to this as, you know, the data we collect from machines or how well the technology works, if that makes sense. Yeah, so the specialists on site who can help, uh, give us access to the data, can help us process the data. Having that. Specialist knowledge and input from the very early stages means that when we run into the initial deployment, the, the start of the project, that everything has been considered well, that we've had a smooth implementation, that the onboarding process has gone well, and the results that we're getting are really good because we've taken everything which is important, which can only be found out from the specialists on site, the, the people who have the knowledge and know-how and experience. They bring into a project is just phenomenal. And I guess that, that leads to a, you know, a more general point about, you know, workforce engagement in the projects and the change management required, um, for a successful rollout because we know you were just saying then about, you know, you need people engaged in the project in order for it to work, or, you know, people engaging with the app and, and looking at cases and things like that, but maybe you could talk a little bit, Jon, about the, um, the change management piece. What, what mindset needs to sort of change within an organization. Absolutely, without people. Adopting the solution without them using it, it's useless. Doesn't matter how good the functionality is, doesn't matter how impressive the features are. Without people using it, it is useless. So, we have a number of ways of addressing this. The first is to get the right people in the room at the right time. Early on. So when we start a project, there's typically 5 key characters that we look to involve. Project sponsor, someone who at a high level can enforce change. They can get you the budget and the, the PO that you need. They can get replacements for anyone who goes off sick in the project, high level change, project sponsor. Then we're looking for maintenance representation, because ideally, maintenance personnel are our desired end user. So we want them involved as early as possible, so that it's their solution, not something that management has done and now expects them to use. We're looking for a PDM champion. Someone who is gonna be responsible for day to day progress and moving the project along. We look for controls or controls and drives engineers, um, OT staff, people who can help us with the OT environment, and then IT essentially because we need the data to move from the OT environment and come into the IT space in the cloud. So having the right people in the room at the right time is absolutely key. But you also need to get them on board. You need to make sure that their concerns are addressed, that they are completely satisfied and happy, and that they become, every one of those people actually become your PDM champion. Because they can progress it along during the initial stages, but also act as your champion throughout their business. They can take the results and distribute it internally, um, and get other people excited so that it's more widely used and more widely adopted. And really what we want by that is to have other departments, other plants, other factories and other locations, begging to have the solution implemented because of the good stuff that they're hearing from SIITA. And you mentioned that, you know, about scale as well, because, of course, predicting maintenance across a line, across a few machines is one thing, but across the real, uh, value and return on investment, as we know, comes from scale across multiple plants. But what are sort of the challenges, again, thinking about heavy industry, you know, probably very remote environments. How, how, how tricky is it to, you know, scale predictive maintenance across those environments? Scale is not actually that much of a problem. The machines are there, typically the infrastructure is there. There's some small adjustments, even for remote plants, you can get connectivity, so we can still get access to the data. So, the ability to scale is not the problem. It's about targeting the use case, the value, the business outcomes that drive scale. Because you're working with the project team, they know the ins and outs of the solution you're proposing, they're all happy with it. But they all need to go and present these results or, you know, the possibilities to a higher team. Um, and once you're running the project, they need to advocate for it and, you know, push those results again to the higher team to get further buy-in, to get a further financial support to scale. So I would say one of the biggest challenges is really Starting a project that is going to drive business value, that's going to drive return on investment, and then making sure that you are Targeting everyone's concerns, their desired outcomes, so that you're running the project successfully, that you're constantly checking on milestones, making sure you're hitting the success criteria at the right milestones, so that when you do come towards the end of the project, as I said, people are dying for a scaled operation, not just, that was nice, but uh we'll try again in a few years. And so, and it, it's all got me thinking about um who leads these projects, and maybe it's not. You know, probably differs organization to organization, but from your experience in, in heavy industries, is it, you know, do you find these are sort of IT led initiatives or more on the sort of the maintenance engineering side, or it, is it really, you know, both coming together and meeting in the middle somehow? Absolute mixed bag. Um, we, within heavy industries, we've engaged with maintenance teams, uh, digitization teams, uh. IT teams, a full range of. Teams, I guess, but a full range of them who are interested in predictive maintenance. But all of them have different reasons for investigating predictive maintenance. So maintenance team, looking for a maintenance efficiency. Fantastic. That's what predictive maintenance is about. Um, digitization teams. Yeah, sometimes digitalization teams can come about because Uh, they were at a, a show, an event. They saw something really cool. Now they wanna investigate and, see if we can make change. Other digitization teams have a very clear use case. Fantastic. We like the ones with a very clear use case. Um, IT teams are typically more cautious of the digitization in a sense that If you look at the, the risks in integrating OT and IT or the perceived risks rather. Um, uh, that is quite high. And what happens if it fails is also quite bad. The, the risk of failure is, is extremely high. No one wants to start something with that amount of risk. So, IT teams can be a little bit tricky. However, when we start to really engage with them, speak their language, identify what is important to them, um, and what risks they feel, we can then start to mitigate those and make them a lot more comfortable with, with the project. But typically we're wanting to, to work with. Maintenance, digitization, those type of characters because typically speaking, their objectives align with what the deliverables are with predictive maintenance. Yeah, it's interesting cos you mentioned objectives and outcomes and that, that's of course, why we're here is to deliver the, the outcomes and overcome the challenges customers are facing, um, and I know we could talk about unplanned downtime and reducing maintenance costs would be more, more efficient. Well, I guess one, which I think's probably particularly interesting, well, particularly important, important in any industry, but in every industry particularly is, is safety and workforce safety. And I remember speaking to your, actually, a guy on your team, Tom, a colleague of ours, he was saying he, he spoke to a customer who, um, who actually brought that to his attention, you know, when they're talking about, oh, you know, what, you know, what challenges are you facing? Is it downtime? Is it, and then he was like, no, safety, which we all know is the number one priority. But again maybe you could talk about, you know, predictive maintenance, you know, role in that, which can play such an important role in that side of things. Absolutely. So, if I just run through some of the benefits of predictive maintenance, reduction of unplanned downtime, right, that's the, the obvious one. But also precise and correct maintenance. …