Rethinking Predictive Maintenance: How Do We Roll It Out Across All Our Sites? - with Peter O'Hara
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.
This is the fourth and final episode in Rethinking Predictive Maintenance, a Trend Detection podcast series exploring the people, data and decisions behind successful predictive maintenance.
Host Niall Sullivan is joined by Peter O’Hara to explore what it takes to expand predictive maintenance from one site to a wider, multi-site programme.
Peter explains why scaling is not simply about copying the same deployment across every factory. Different sites may have different reliability challenges, infrastructure, assets and maintenance priorities. The conversation explores how manufacturers can share lessons and standardise where appropriate, while still adapting the approach to deliver value at each location.
In this episode, you’ll learn:
Why predictive maintenance often demonstrates more value at scale
Why a successful deployment cannot always be copied directly to another site
How local engagement and global knowledge-sharing support adoption
Why each site needs clearly defined problems and measures of success
How documentation helps preserve knowledge as programmes expand
What successful multi-site adoption can look like over time
Listen to discover how manufacturers can build momentum, share successful practices and scale predictive maintenance without ignoring the needs of individual sites.
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
The communication from the management, why they're doing this is particularly these days with AI, you know, a lot of people are worried that AI is going to take over their jobs, but I've yet to see AI hold a spanner. Welcome to the trend detection podcast. Powered by Siemens. Hello and welcome to the latest episode of the Trend detection podcast. I'm really pleased to be joined again by Peter O'Hara from our customer success team at Sensai. How are you, Pete? I'm very good, thank you, buddy. How are you? Very good, very good, thank you. Been a busy period for recording, recording podcasts actually. This will go out. I'm on holiday in early September, so I'm really pushing through the topics at the moment, so looking forward to another great conversation though. We're, we're talking today about scale, that favorite word of ours that we've been using for a number of years, but more specifically about how we roll out PDF across all sites. Because this episode actually you came to me with some real life questions from a customer, which, you know, translates perfectly to our podcast format, and this, and this was one of them, and I thought it was a really interesting one to start with. So just to get your, your thoughts on it first of all, you know, how, how have you seen sort of the scaling of PDM sort of across your time and with Sensai, Pete? How has it changed? I think, I think one of the, the key changes from the, the world of Sensi was when we were acquired by Siemens, and then that allows us the conversations perhaps at a, a broader level with customers who want to scale. We still see some people wanting to, you know, prove the technology, as they say, and start off with a pilot, which. We're a little bit against because from the previous podcasts, if you want to try this out on 10 machines, then what's the likelihood of it being interesting to anybody. So it's nice when we have customers who say, yeah, I'm like the idea, we want to run with it, and then they commit and they provide enough human resources and the. ITOT infrastructure and all of that sort of stuff to actually do it at a reasonable pace, so yeah, it's exciting when customers are like that. Yeah, well, they're, they're almost more, more eager than us to scale, right? I guess maybe that situation happens from time to time. It's like, no, you know, may, you know, before some of the groundwork's put in place, they're, they're already looking further ahead, which, which is obviously a positive, right? Yeah, I think, yeah, having a customer who has a plan of, right, let's roll out across, you know, 10 number of factories this year, then. That is good because if, if it's a case of well we'll roll out this factory and then we will wait a year and try and get the budget to do the next one and that sort of thing, you're always. How can I say this politely, it's the same amount of work from my point of view, but it's, you're not sure what the outcome's going to be, whereas you can get a bit of momentum if you roll out across site in, say, 3 months and then. You can move to the next site and the. You know, there's a similar team of people with the similar ambition and there's a little bit of enthusiasm to get things moving, that is much better than being the other way where it's like, right, OK, we've got these 10 machines and then people are less engaged because the 10 machines aren't failing because out of, you know, in a space of 3 months, probably 0 of them will fail. And the PDM people who should be using SenseI are out fighting their usual fires, and so it's harder to show, I suppose, yeah, we use the term success, what does that mean? But you know, it's hard to show success when there's nothing actually going wrong to prove yourself on, whereas doing it at scale, as I always say, Sei's brilliant at scale because it's not about everything you look at, it's about everything you don't need to look at. Mhm it's being monitored. And uh, I guess the first question I'd ask, when you, when we have such enthusiastic, when we do have enthusiastic customers looking to, you know, at scale very early on, I mean, do they, do they realize exactly what's required to do that and what they're taking on internally as you mentioned, you know, quite rightly about some of the, you know, some of the factors in place and maybe some of the things that need to be in place, but do they sort of realize at this stage what it takes to, to scale? Hopefully we can tell them, but yeah, we do have customers who don't listen to our advice and our previous experiences, but usually someone who's enthusiastic to run with it. Then It's easier to. Maybe I say, slow them down a little bit more than someone who just wants to start slowly and then prove in every area, when like I say, they're probably not monitoring enough equipment to make it, you know, a good investment for, you know, to see return on investment when you're only monitoring 10 things and then they're not failing, then you've got the cost of the people and Sensai and whatnot looking at things that aren't failing, whereas if you do it at scale. You will usually capture things fairly quickly that are in the, in a failure mode, and then that's. When customers Can see like oh this has picked up X or Y, then they are more engaged to, you know, to roll it out elsewhere, is what I find. And, and in terms of, you know, so I guess I'm just thinking the differences between sort of rolling out PDM from one site to multiple sites. I mean, what, what sort of gets harder as, you know, thinking starts to go beyond that first, that first site. I think. One of the areas, and it's one of the things I've said in numerous other podcasts about cookie cuttering, the solution is that if the. Further sites don't have the same issues, then sometimes a cookie cutter approach will not have the same effects, and I have been even like around different shops within a facility and people say, well, they're. Telling us we've got to roll out sensor, but we don't have those problems, and you're like, OK, so what problems do you have? And in one of the previous ones I was saying about someone who, saying we don't have a downtime problem, and then we're looking at his equipment and we're like, well, yeah, we, we replace these motors every 6 months because we once had a failure after 7 months, and it's like, OK, well maybe we should look at monitoring those and pushing that window, and then if we can see, you know, degradation in the data. Prior to when you're changing them out and you change them out and the data returns to normal, then perfect. If, however, there's no degradation in the data, well maybe we can push the, you know, the life expectancy of these motors and increase that. So again, I'm not saying. Don't do all your usual maintenance strategies. Just because you're now monitoring the sense side, but it's a way of trying to, you know, hopefully adapt and improve the maintenance strategies to make them more cost effective and by not replacing parts that don't need to be replaced, you can save an awful lot of money. And it's really interesting because I do remember we, probably on the last podcast we did together, you, you talked about the cookie cutter approach and, and you know, you know, in marketing we talk about, you know, scale, but of course different sites can have different approaches, different system architectures, different. Processes you can't, you can't sort of never presume. So how, you know, if we're talking about, and we talk about champions a lot, I mean we've talked about that, and the importance of like one, let's say global, or maybe more than one global champion, but how do they, how do they sort of navigate, you know, that trying to roll it out when, like I said, different types of different, completely different situations, different KPIs, different success criteria and that kind of thing. Yeah, so I think. So we've got some customers and they have completely different IT infrastructures in every site, and so, when you say roll out site to site, then they may even have the same sort of machines, but if they have different. There are equipment providers around the world in different IT then we may not be able to use exactly the same approach to get data from their site, whereas we have other customers and they've got the same IT infrastructure all over the place and then it is almost as simple as just hooking up the next section. So, but I think with. Both situations. You know, stakeholder engagement is going to be critical. The communication from the management, why they're doing this is particularly these days with AI, you know, a lot of people are worried that AI is going to take over their jobs, but I've yet to see AI hold a spanner, so I think it can improve the maintainer's efficiency because we're removing some of the the routine checks and the things that they don't need to do, whereas AI can do that at scale, but then. They're just, we're making their job more efficient and effective, but this isn't going to be replacing people. I think, like I say. Capturing lessons learned from one site to the next is important, but I would also ask customers when. They engage with us to listen to some of our advice about certain things, and we have customers who have completely different. Like sensing setups, um, way of users engaging with the system in one country compared to the other, and then they're asking why the, the results are varying. Well, I've got one customer who in one part of the world, they have people, they pay Siemens to actively monitor Sensi and to tell their. Maintainers, what is going wrong and what to do about it, and then in another part of the world, they don't have the sensing technology to allow that. They bought their, you know, they didn't go through the sensing technology that we had recommended, but they've got good sensors for PDM but not that, not to be able to provide that kind of remote service level of support, because we can say. This vibration is increasing, but we don't have the level of data coming in to say which part of the machine and, you know, what they need to do about it. So you can't compare the results from one site to the results from another site, but you can look at operating in slightly different ways and still getting good success from both sites, you just, it's harder to compare them. And, and I think, I mean, I, I've got a great example from the other day. I can actually say it's because the podcast's just been approved, but our customer Highland Peelets I spoke to, uh, I interviewed for this very podcast, and that one thing that sort of stood out to me is how Sensi's enabled sort of increased collaboration. I know we're not talking about big rollout, but even through the various teams reliability operations and things like that, and it, and it makes me. You know, wonder both on a site view, but also a global view. I mean, how do you balance that sort of, you know, giving people sort of global access, because that's a lot of knowledge and a lot of sharing of knowledge, I think you mentioned as well, versus, you know, what should be contained within one site. I don't know if it, it's probably an approach different customer to customer, right, in terms of when they think about scale, but, you know, you know, how is, how is that generally balanced? Uh, yeah, I, I suppose the, in the Sensi world. We believe that sharing knowledge is a good thing and we are against, particularly knowledge silos. There are some customers out there who for. Reasons that I still don't understand, will not want one site being able to see what another site is doing, and. For me, surely you want to share best practices, and I would suggest our best customers are the ones who openly share best practice and if a, a site is monitoring something and they want to bring their similar version online, the best place to do is to look at the other site and what are they doing, what results are they getting, and what would we do in our situation. So I'm, I'm more. Open to the standard Sensii way is you have one customer instance per, well, customer I suppose, and then you break it down to the different installations. And then from that we have. Everybody who can access the application can see everywhere, but the, the, the right access is controlled to only their area of operation. So, you can have someone from one country looking at how another country's doing stuff, you, they can search for, but I wonder if they've got this asset type in there and then they can see, OK, this asset type, what are they measuring, and again, you can look and see what. Success KPIs, etc. that they've entered there, so I think transparency. Is a good thing from the customer's point of view because it shares knowledge and learning, and then from a sensei point of view, it's so much better if customer sites get a little bit competitive with each other because then they, they will, you know, try to be the site that. Adds the more machines or has recorded the most particular KPIs and that sort of thing, and that's, if you can gamify it a little bit like that, then it's good from a sensei point of view, and I would suggest if the customers are using the system to its fullest capacity, it's good for them as well. Yeah, that's, yeah, it's interesting, the internal competition, it's always, it's always gonna be there though, it's just human nature, isn't it? Of course, cos everyone wants to be successful. I look, I know you, I know you talked about it before in terms of, we talked about cookie cutter and it's very difficult to do that, but I guess the lesson here is, you know, it, you know, it's very difficult to standardize, but you can certainly learn. Yeah. Right, in terms of what, what was successful for you, what was maybe not, and then it's maybe adapting that thinking to that particular site. It's maybe the better approach rather than thinking, oh, they've done a great job, let's just copy it, the blueprint, exactly, and, and go again. Cos if you're trying to, if you don't have the same problem as the first site, for example, then you might have newer machines, whereas the, the first site has taken on PDM because they've got a reliability problem. Well if you've got a lot of brand new machines, then maybe you've got, you know, a, a spares capacity problem or something like that, so, just monitoring the same thing and expecting the same results, you know, you're not going to. Reduce downtime if you don't have a downtime problem. But if you've got an over maintenance problem, you can reduce your maintenance costs or your spares costs or you know that sort of thing. So that, that's the only thing I would say, like I say, about the cookie cutter is you just have to. Two things really, make sure you're trying to solve a problem that exists and. Make sure that the people who are engaged, well hopefully people are engaged on the site, but people at the site, you know, want to solve this problem with you, you know, the, the engagement of those people, if they, sometimes, depending on the company culture, if they see it as something being forced on them by head office, there's gonna be resistance. And then, you know, part of my job is to try and overcome that resistance, try and get them to see what a good thing it is, but if. The, the management. Don't, you know, the local management don't celebrate. The success of a sense I Win, if, if people are, you have a downtime avoided or production materials saved or whatever, and people don't want to celebrate that, then it's going to be harder for. Someone like, or, you know, people using Sensei to feel that, OK, there's a benefit of doing this because, especially if they're being pulled into lots of different directions, if you've got quite a small deployment and like I say, things are failing elsewhere and the, the bits that you've got Sensi running on are the newer bits because they have the, the readings available, then people's attention might not be. Put on, you know, the sense I world, they're going to be fighting the fires that they've, they always have done, so you need often, you know, it's a process management a change, I suppose that you're putting in a new system and you want it to complement their existing workflow. But also you've got to appreciate that if that existing workflow is already working at. 110, 20%, then fitting something on top of that is going to cause friction. Hm No. No, no, that makes, that, that makes a lot of sense as well. I mean, I'm also, I guess the other thing to consider because we've talked about, we've gone straight to sort of scale, or I've gone straight to scale, but actually what I was also thinking is what, where, I mean, which site do you start with as well? How's that sort of determined? Because I guess that's quite important, isn't it, to lay, I mean, maybe the, the answer is it's whichever. I mean, is it more of a customer led decision or is it more, you know, do we have input into that, you know, how does that work? Yeah, I imagine it's partly customer led and partly the region selling the solution led because they'll probably try and get. You know, they will be engaging with the, whatever local team that is. So. Yeah, I think when we, when we start to, right, the initial sales process. They'll be engaging, you know, the people in their region, and so. Again, I kind of believe that it will work better if the solution is rolled out in that area, in that region, because you have Bighton, and like I say, the thing you. I would like to avoid is sense I'm being forced on people because. Certain people buy our solution, but they don't actually use it. People who use it are the, the maintainers and. The, you know, maybe the project engineers or whomever at the different sites who are rolling this out. So The people who will make it a success are the people at the site, and if it's a somebody at head office thinking this is a great idea, I'm sold on it, right, you're gonna use it and those people don't want to, then the project will struggle for, you know, momentum, I think, so yeah, I, I think definitely customer led, but local buy-in, you really need. Somebody who Understands the vision of PDM wants to. Do things in a slightly different way to improve the. Customers their productivity and hopefully improve their. The, I can't think of a better word than, the work life is like less firefighting, less stress. You want to be planning your, your next few weeks ahead and not just reacting to something being broken and the management shouting at you, why did this happen? Yeah, and I, I guess when you, when we're thinking about a site level, choosing sort of assets and lines to focus on is another thing, but I guess. Choosing the site maybe becomes obvious because they, they'll, there'll be an obvious issue there, whether it's too much downtime, you know, too much over maintenance, too, you know, maybe that, that's what springs them into thinking, maybe we should look at predictive maintenance, use this as the starting point as we know this, this plant needs some attention, and then let's look for other use cases in other plants and it sort of naturally progresses from there. Yeah. I mean, one of our biggest customers. Years ago when we were first put in one of their plants and they were like, you know, you haven't saved any downtime here in the first like 3 months, I was like. Have we missed anything? Like, well, no, we don't have a problem on these machines. You're like, so, OK, so what machines do you have a problem with? Oh well, it's those ones over there, like, well, let's have a look at the data from those ones over there. But we needed to almost get to that point for the local team, I guess, to feel empowered enough to say. …