When Predictive Maintenance Is (and Isn’t) the Right Tool for Your Plant - with Natalie Kurgan

Trend Detection Podcast ·

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.

In this episode, we’re joined by Natalie Kurgan, Head of Delivery for Senseye across the Americas at Siemens, who shares a delivery‑side view of when predictive maintenance is (and isn’t) the right fit—and what plants need in place before they start:

Why predictive maintenance is a strategy, not a tool, and why success depends on people + process, not software alone.

The readiness checklist that’s often missing: leadership support, a clear workflow, and a technically minded champion who drives action.

Where projects go wrong in practice—from weak ownership to poor asset selection and low/limited data quality.

What PdM can realistically deliver (planning spares, reducing unnecessary planned work, avoiding risky unplanned failures) vs. what it can’t.

How AI copilots help and their limits: they need context and feedback; they don’t replace human judgement.

If you’re not ready yet, how to get there: define KPIs, audit maintenance logs, identify problem assets, then assess what data/sensing you actually have.

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

Oh, there's, there's so much value that it can add. For example, you can make sure you're prepared to have a spare parts inventory. If you know 3 months in advance that a, a part of your equipment is starting to degrade, then you can plan to have its replacement ready to go for when you perform maintenance. We have people that have eliminated planned maintenance on the assets that they're monitoring with Sensii, so they no longer send people on a round to put hands on a machine unnecessarily. 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 section podcast. Um, I'm really pleased to be joined by an old favorite, old in the sense that you've been on the podcast many times, I should say, um, Natalie Kergan. So welcome back, Natalie. Thank you for coming back and recording. That's a strange intro, isn't it? But we'll roll with it, we'll roll with it. It's fine. Now, uh, I mean, today, so we're welcoming you back, and it was actually you. Who approached me, which is always nice, so it's not just me, um, coming up with these topics, um, as I hear them, but it was you who came to me, so, um, I guess before we go into the topic, which, um, we're talking about when predictor maintenance is and isn't right for your plant, which is a very important question actually. I'm surprised we haven't covered it, um, in depth so far. But I guess before we go into that, Natty, it'd be great, um, maybe to remind people of your role at Siemens and with the Sensa, um. So I was actually part of Sensai pre-Siemens and I worked in delivery, so I was delivering Sensai directly to customers and post integration with Siemens, I am now the head of delivery for all of the Americas. So I, I train our regional colleagues to deliver Sensai and I coordinate efforts and I consult and I still work with some customers, so. I've got a lot of experience at this point into what makes a good and a bad PDM project. Exactly, perfect for this discussion today. Um, but I'll, I mean, I'll put the question back to you, like I said, because you, you came to me with the topic. Do you remember why, I'll say do you remember because it was a little while back, but do you remember why this topic sort of came into your mind, um, and it's all these projects you're involved in, right? Something sprung from there. I genuinely, I don't remember exactly why, but it's something that we encounter all the time, is that we end up with, with deployments in plants where it's just not necessarily the best case scenario for starting with PDM. Uh, and then, of course, my role in delivery and in guiding people to deliver whatever the customer thinks that they should get out of the PDM project, we have to deliver that regardless of whether or not it actually makes sense. So that's probably why I came to you. He came to me to talk it out on the podcast. That's, that's fantastic. But I guess, and I guess it's, it's a lot of that driven because, you know, I'm in the marketing side and I've seen sort of the hype rise in predictive maintenance. Like everyone's talking about it. Everyone says they do it. I guess that's driving a lot of this, right? Like we must do predictive maintenance no matter whether we're ready or not, kind of thing. We, we just need to do it. So, I think there's a lot of misconceptions. I think there's a lot of buzzwords, which we can't. Get away from because Sensai is a predictive maintenance decision support tool, right? However, I think some of the, the confusion comes from all these different places are saying that they do predictive maintenance and when I actually go and look at them, they're, they're actually not all doing predictive maintenance. They might just be doing condition monitoring, which is fine, that's needed too, but now customers. Might not have the right idea of what PDM actually is. So sometimes they still kind of think it's this magical solution where they're going to just give it their data and the system is going to tell them exactly what is happening with 100% certainty and what maintenance actions they need to go perform and where there's like no input from a human. That's just not reality. There's too many unknowns for a system like this. Just be able to take that amount of control and certainty over the situation. But I think there's also misconceptions around labeling tools themselves as predictive maintenance, because as you've heard me say before, this is a crusade that I am on, which is that no tool is predictive maintenance. Predictive maintenance is a strategy that encompasses many tools and many practices. So. Senseii is part of that, it's a very useful part of that, but if you don't embrace the other parts of predictive maintenance as a strategy, you aren't going to succeed. Yeah, I, and I guess we consider predictive maintenance isn't a, isn't a tool, it's more of a, a strategy. So what I guess are the core conditions that make predictor maintenance valuable to a plant, if you think of it through that lens instead? Oh, there's, there's so much value that it can add, uh, you can, for example, you can make sure you're prepared to have a spare parts inventory. Uh, if you know 3 months in advance that a part of your equipment is starting to degrade, then you can plan to have its replacement ready to go for when you perform maintenance, you know, just before that 3 month period. We have people that have eliminated planned maintenance on the assets that they're monitoring with Sensai, so they no longer send people on a round to put hands on a machine unnecessarily. Uh, no longer do people have to go waste their time going to some far reaches of the plans unless they actually have to, as directed by Sensai. So having that visibility allows people to just trust in the system to tell them when they need to perform the work, and they, they go and they know what needs to be done, so they go and do it, but. I think reducing planned maintenance. There's also, of course, the safety aspect. If, if we are preventing unplanned downtimes. I think unplanned downtime, like if something goes down in the middle of the night, maybe it's 2 o'clock in the morning, it's crucial, the whole line gets taken down, you want to get in and fix. It as fast as humanly possible, you're a lot more prone to shortcuts and making mistakes that could have safety implications. So those are just 3 off the top of my head, but there's so much benefit to just knowing about the health of your machine rather than just waiting for something to happen. Yeah, I, I don't know if this is a strange question, but are customers actually aware of all these different benefits? I mean, they might be aware of something like, you know, reducing downtime, for, for example, but are they aware sort of the broad benefits rather than just bucketing it in, oh, it's just another digital thing, another digital project or initiative we, we should get off the ground, but we've no real idea of what They're trying to solve with it. I don't know. I don't know if they're aware of that from the onset, but it's definitely a big part of our conversations as soon as, pretty much as soon as I start working with them, uh, so as soon as it gets sold, you know, we start these conversations not just around what data do we need and how do we get your data, but it's around what value are you going to get from this because everybody's different. And some of those, some of those value indicators are really hard to quantify. How do you quantify safety? I don't know, but it, it's something that should be taken into account. Exactly. And I guess you mentioned there as well, you know, like data as well, but, you know, it encompasses so many different things of predictive maintenance projects, and you, you know, whether it's technically, whether it's You know, the organization, whether it's a cultural thing. So, with those three key areas in mind, so what's, you know, what are the warning signs that something's not quite in alignment for a predictive maintenance project? Well, it, it always comes down to the organizational structure, I would say, of whoever's implementing Sensei. So, I can give you some examples of Times when it's been very difficult, and it, it usually comes down to lack of support. I have a customer I'm working with right now, who is a really good example because quite literally the CEO is in the calls during the case reviews with me all the way down to the maintainer level. Tell me about another company that has that level of support from the top and the bottom all at once. So anything that needs to get done, they get it done. So, um, I think what, what can happen is people get this tool, they feed their data in, and they don't realize that you have to do something. When Sensi tells you something, you have to make the decision. It cannot make it for you. So you need somebody who understands how to look at data, and you need somebody that understands and, and has a workflow in place to actually do something about it. So let's say we do spot something. Um, and, and, you know, part of the, the aspects of predictive maintenance is we are predicting. So it could be something that's 6 months down the line, it's gonna be a problem for you. What are you going to do about it right now? OK, so since that tells you what is your workflow. If you don't have a workflow, OK, let's make sure we have a spare part inventory, let's check our plant maintenance schedule, let's plan for a scheduled downtime. You know, during this event to make sure that we can address this way before it's an actual problem. Let's change operations to, you know, slow it down or something to baby the machine to make sure it can last to a different plan maintenance. So you need somebody like a maintenance planner involved that can, can help make these decisions. So there's a whole workflow involved that needs to be in place. And at this point I've been rambling and I don't remember the original question. No, that's OK. Rambling's fine on a podcast in the right place, Natalie, don't worry. Um, but what I was going to lead that to, I mean, that's an incredible example of the CEO. I've never heard of that. It's great. I mean, what, what sort of role do they play on the call, because I mean, some people might imagine the CEO's overbearing and like, you know, like wants to, you know, with other people on the call might feel like all the CEO's here, but what you're saying is that's, that's not the case at all, right? They're just genuinely interested in what's going on, I could absolutely see what the example you just gave. Being a risk because usually, you know, these, the people maybe in a steering committee aren't so technically minded and they don't understand quite so well, but I would say the CEO might be the person that understands the best. I think he was the main driver for getting Sensai. He's the one that was the main person excited about it. So there were other people in the call asking questions about things that were actually out of scope for this project, for this deployment. And he would step in and say, yeah, that's not on our, our path right now, uh, when we scale, you know, we'll we'll talk about that. But he's an engineer, so he understands what he's looking at, and he's, he's helping, you know, I would, I would raise something. I said, OK, this, this case is indicating something. Your machine is in trouble. What are you going to do about it? And for a moment, his team was silent and he stepped in and said, Come on guys, what are we going to do about it? And I'm like, this is, this is great. Wow. Yeah, I mean, because we're always saying, and we're always saying on this podcast, right, that, you know, you need multiple levels of stakeholders involved, but, you know, to have that sort of involvement from the CEO is just. Just something else, right, driving it. Um, but I mean, and who, I mean, if we're talking about, you know, because we're talking today about what isn't, uh, you know, whether PDM's right or isn't right for a plant. So who's involved in that decision on a, we just want a plant because we know it can roll out much further than that. But if we just talk about a plant itself, who, who'd be involved in those case reviews or some of the earlier discussions about it? So the, the critical thing when we get involved with the customer is we need what we call a champion. This champion needs to really want Sensai. They need to be interested in it. They need to be able to see the benefit that Sensai will bring to them in terms of efficiency and savings. So they need to care about it. They also need to be technically minded. They need to understand their machines. They need to be able to read the data in the charts because again, some level of competition. Competency is needed in our in in data analysis. You don't have to be an expert data scientist, but you need to be able to look at a time series and understand what the data shown might be indicating for your machine. So a champion has to kind of encompass all of those, and they need to be also proactive and have channels of communication open to, like I said, follow up after the case. How, how is Sensa embedded in your work? Workflow, your predictive maintenance strategy. So they need to be able to go talk to operations and have some amount of uh control there to be able to maybe change how the machine is operated or they need to talk to the maintenance planner, um, so they need to have some amount of authority. So I would say that champion is critical. If we don't have a champion, we have a really hard time because we need one person who is taking charge and saying, I'm owning this. Eventually we get end users in who are just regular users, but we need that initial admin, that champion from the, from the beginning. And, and again, yeah, we talked about champions a lot, but I think it's interesting because you brought up the situation with the CEO joining the call. So is the champion usually someone in a certain type of role, or, or can it vary? Like with the CEO maybe becomes the champion in a rare case. I would say CEOs are generally not our champions. They might be our, our leading sponsor, which I think, I think is this case. He's just more technically minded than most probably. Um, I would say a champion, my ideal champion is a reliability engineer. Uh, sometimes people don't have those, actually, the, the customer I'm talking about right now doesn't have a reliability engineer, but they found somebody who, if they decided to have that role, I think it would be this person. And that just proves it's, it's a per organization thing, right? You know, it could be the CEO, it could be reliability engineer, it could be. Someone in maintenance, but the idea is you need someone who's proactive, who's going to not just, you know, see the information, but actually connect with the relevant people so people take action on the information that's actually been provided. And to make, just to make the decisions, make, make decisions on what needs to happen, we actually have, it almost looks like a, a job posting for like the required characteristics of a person to be in the role of a champion and an end user. So we actually have those in our help center to help organizations find the right people. Within themselves to to fill those roles. I think that's interesting as well. So they might not even, even if they see it on paper, these are the roles, they might not even understand themselves within the organization who's who's right. Again, it's clearly it's not just a, you know, cookie cutter approach to every organization, right? Right. Well, sometimes they come to me and they say, OK, who is our champion? And I said, I can't answer that for you. But here's, here's kind of what this person needs to have the characteristics of, um, so we can start there. In terms of examples where another way to look at it, I guess, is when everything looks right on paper. So, you know, right organization, you've got a champion, then it just doesn't work for whatever reason. I mean, do you have any examples in that area where, you know, everything could look set up for real greatness, but it just doesn't quite work out that way? Hm, that's a good question. I think, I think it's. So I've said this already before in this call even, but it's so important to have support from the bottom and the top, because if you have, let's say you have a great champion, and they're doing their thing, and then they learn everything, the ins and outs of Sensai, and then they leave their job or they move, and then no one else cares enough to take it up. What do you do? So. I think that's where the support from a sponsor is so important because you can rely on your champion a lot, but you need support in case something changes organizationally. And I would, I would also say, so from the, the people point of view, that's always the struggle, but From the assets and the data point of view, of course, that could be our other area where, where things don't quite work out. Maybe the right assets were not chosen in the beginning. Uh, maybe the assets chosen are super complicated, so complicated that maybe the champion can't make sense of what Sensai is trying to tell them about the. Changes in their data, maybe they don't have the right data or low data quality or, you know, anything to do with data. You need to, since I will give you what you put into it, so if you put in terrible quality data or not enough of it, it's, it's not going to be able to tell you magical results. So, of course, that's always a factor. Yeah, let's go more technical side in, in terms of data and connectivity. So that, I guess there must be like a minimum amount that that's required to get something up and running. I'm gonna presume a little bit that it's underestimated what's actually required, maybe, or maybe you could answer that rather than me presuming. It's, I would say that everybody's least favorite answer is it depends. I think as long as expectations are manageable or reasonable, so for example, let's take something really simple, a motor. Let's say all you have is motor current, and that's the only thing you can get out of it. We can start with that. We can, but are we going to detect bearing problems if all we have is motor current? No, we can, we can detect electrical problems, maybe not all of them, but it's a place to start. But what we find is, you know, we onboard some data like that, and we might have a customer that says, well, the, the motor went down, why didn't you detect it? And we say, well, why did it go down? And they say, oh, because of This thing that happened downstream or, you know, this bearing was crushed or something, and it's like, how could we possibly detect that? I'm looking at the data with my eyes. I see nothing to indicate any sort of degradation. So if you want to detect that, we need more data, but that's like an expectation that has to be managed from the first conversation with a customer. And it, and it links back to what you were saying earlier, Natalie, about, you know, predicting maintenance being some magic box you, you could apply and it'll just spot anything and everything, even like you said in that, in that those examples, even without the right data to actually spot that, right? So you can only do so much, right, with the data and everything else that's that's put into it. Right, so one of my, one of the things I like to say is sensei predictive maintenance cannot predict unpredictable events. Which sounds so obvious when I put it that way, but sometimes something might happen instantaneously, and since I doesn't detect it, because it was an instantaneous failure, there was no degradation. It just maybe something, maybe a meteorite fell out of the sky and landed on your motor. …

Full transcript (WebVTT)