Trend Detection Podcast - Special Edition - Discussion with CooperVision - Part Two

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.


For this special episode of the podcast, we were joined by Matt Walter, Principal Controls Engineer at CooperVision to discuss all things Industry 4.0 and their predictive maintenance journey.

In the second episode of this series, we discussed CooperVision's predictive maintenance journey, why a PDM champion is important, and the common mistakes to avoid when deploying a predictive maintenance tool.

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)

  1. CooperVision & their digital journey
  2. The integration process
  3. The Predictive Maintenance journey
  4. What types of assets do you monitor using Senseye?
  5. Senseye PDM Champion
  6. Key results achieved

Niall Sullivan, Senseye: Welcome to the Trend Detection Podcast, powered by Senseye, an industry leader in using AI to drive scalable and sustainable asset performance and reliability. For this episode, I'm joined by Matt Walter, Principle Controls Engineer at CooperVision, a soft contact lens manufacturer. In the second episode of this series, we discuss CooperVision's predictive maintenance journey, why a PDM champion is important, and the common mistakes to avoid when deploying a predictive maintenance tool.

And so I wanted to move on slightly and broaden this discussion out a bit and focus on digital transformation. So when you join CooperVision, it'd be interesting to see where CooperVision's digital transformation journey, where they are when you started and where they are today. Maybe you could even jump ahead to where you look what the vision for as a future. So that's quite a long detail question, but yeah. So maybe we start where were CooperVision when you joined and how has they developed their digital transformation journey?

Matt Walter, Coopervision: Sure. So when I started, we were already in part on that journey. Now what I mean by that was all of our manufacturing equipment was very heavily networked, which meant the availability of data was already enabled by the nature of our process and the regulatory requirements that we are strictly governed by. We also had huge amounts of data that was being generated and stored. So again, we had the silo data available. What was lacking was the ability to look at that data holistically and apply some level of intelligence or perhaps modeling to start to unlock the correlations and the relationships between those various data sets.

As the business has evolved and the adoption of industry has become more and more of a priority, we very much look to bridge those gaps. So things like the creation of data warehouses or data lakes to start providing a common platform whereby we can ingress a lot of these siloed data sources into a common location, applying some good data engineering practices where we've correlated, we've cleansed, we've aggregated where needed so that the data contained within our warehouse is validated and is now ready to apply analytics or modeling to. Part of our standardization approach for our machine design as standard, we now incorporate additional analytics modules, be it physical modules in terms of energy monitoring of both air and power, but also from a software perspective.

We pre cleanse and prepare all of the data required, developed largely in conjunction with Senseye so that every machine coming into the business now has already got the primary metrics pre-prepared, ready to go, essentially plug and play. So that as and when we are ready to onboard the next line, the next equipment, the next assets, most of the work is already done. We simply need to point it at the equipment, draw out the relevant information and it onboards very, very quickly.

Niall Sullivan, Senseye: Yeah, no, that's interesting. So it's almost a standard approaching for onboarding new machines and collecting the data from that. So is that process developed over the years, bit by bit I guess, to where you are today or is there further... I guess there's always further optimizations you're looking at. Is there any optimizations in that area to take it further than that?

Matt Walter, Coopervision: Absolutely. So it is certainly been an evolution in part due to the asset types. So for each asset type, we need to create a slightly different version of the process or the application that extracts and aggregates the data that we use. So they started off in almost a development phase where we created Mark 1 of the code, working very closely again with some of the experts within Senseye. But as our understanding and as our collaboration with Senseye has continued, we've actually revised and in many cases we've simplified our original approach. We recognized that actually we were almost being overly complicated for no real value. So that enabled us to simplify and strip back some of the metrics that we were monitoring. We've then incorporated that into the next generation of onboarding platform, which we've then taken to the next level, again, in terms of our standards whereby it's no longer a bolt on, it's simply part of our machine control where the data is readily available, straight out the box if you like.

Niall Sullivan, Senseye: Yeah, and that's it. I mean we've talked about the data silos and the different systems. Obviously you've deployed a lot of different systems for a lot of different reasons, Industry 4.0. But what I wanted to know is how important is the integration of those systems to deliver what you are talking about there? And I guess you could maybe start by what systems do you draw data from? So the asset digital machine is one, but what other data sources do you have? And then how do you go about combining or integrating those data sources because it sounds complicated for a non-technical person like me.

Matt Walter, CoopervisionSo integration is key. Ultimately, if you cannot connect the various data sources together, you cannot apply Industry 4.0 Philosophies. So in many cases, we're very fortunate that a lot of that infrastructure has been in place from the beginning. The integration for us or the asset selection tends to be based on historic maintenance information. So what areas of a process or a machine typically cause the most problems? Now it's not necessarily just related to component failure as you would expect from a predictive maintenance platform, but what we've actually identified along our journey is that the data that we can get from our individual assets provide huge insights to our process analytics platform. Now, to give you an example of that, we had a sudden increase in a particular defect that we suffer with in contact lens manufacture. Historically, it's generally been assumed it's a raw material problem.

The raw materials team carried out an investigation and drew the conclusion that we cannot find any correlation. There's no real obvious reason why a material change has caused this defect. However, the process that we were particularly looking at contained eight assets that we'd already onboarded with Senseye. So at that point we thought, well, that's an additional data source that we previously never had. We pulled the data source back, we applied that into the analysis that the team had done and immediately identified a correlation. And it turned out that we had got some damage tooling that the Senseye app was able to detect and identify. We rectified the tooling problem, the defect went away, and we regained our quality and yield. So again, that integration of the raw material data, the process data from the machine, the analytics data from the individual assets as well as our quality information to identify the defect, ultimately unlocked the ability to identify a real correlation that we could actually act on.

Niall Sullivan, Senseye: So I think what we talk about Senseye in terms of data is adding layers of context. So you add another layer, gives you a bit more of a picture, add another layer, a bit more of a picture. And it's a bit like that really is how you'd say, it's adding more layers to add more and more context until you've got, let's say a complete picture of the health of your asset or assets.

Matt Walter, Coopervision: Exactly that. We tried to look at it as almost a process flow of data. Now the raw materials, obviously they go in at the beginning of the process and quality information is typically measured at the end. We verify has the product we've made met specification, but of course from the raw materials into the quality product out, there's a huge number of process steps. Now that fundamentally are the raw simple assets that we are able to monitor with the use of a predictive maintenance platform which is exactly what we're doing with Senseye.

So we look for which interaction, so which part of the process do our process experts know has the maximum or the most influence on a particular defect or a particular shift that we've noticed. And we can then apply the various technologies that we've adopted to help us unlock and almost bridge the gap between the raw material in the quality data route. The flexibility of the technology that we're adopting means that in some cases we can actually apply that to a gap that previously had no data available whatsoever. So we're talking about edge devices and retro fittable sensors where we can create brand new data that's never previously existed.

Niall Sullivan, Senseye: Yeah, I was just going to touch on that actually. That's a good... I mean, we're going to move on to predictive maintenance, and yours and CooperVision's journey as set. But in terms of identifying sensors to capture, to fill in those gaps, as you say, how do you go about identifying the right sensor or the right edge technology to introduce to fill those gaps?

Matt Walter, Coopervision: Largely it's with engagement and collaboration with our process experts. So we are able to identify through the data and through some of our quality systems where a process or perhaps a problem has started and the point at which we detect it, we then take a fairly common sense approach. Well, if it started at station one and we've detected it at station three, we probably need to have a look at station two. We then work with our controls and automation team, so members of the innovation group that I'm in, to identify what equipment is currently on the machine that we could get data from. And if the answer is nothing, then we look to other solutions. So again, during our Industry 4.0 Journey, we have been working in close partnership with several other edge based companies where not only do they have an impressive suite of various sensors, but they've also got the ability to provide edge based technology where we can plum the sensor data directly into it, do whatever data manipulation we need, and share that directly with Senseye.

Now, the simple thing to do would be to simply add more sensors to the machine. The challenge of that is that all of our equipment, and therefore all the software, is very heavily validated and regulated. Now, if we start making fundamental changes to the process by adding in new hardware, new equipment, we then have to go through a big round of re-validation, something that the business is keen to avoid, because it can take a lot of time and therefore costs us a lot of downtime in terms of production. By going to an edge based solution, it's essentially invisible to the machine and to the existing process, which means we can extract all of the additional data points, we can integrate it into a location that's going to complement data that we can already extract from the machine, but bridge the gaps that we've identified through our analytics platform.

Niall Sullivan, Senseye: And I think that raises a good point as well. So I think when we talk to prospects or customers, I think the natural assumption from their side is, "Oh, what sensor do we need? What do we need to install"? And it's like, well, if possible we try and see what they've got already and try and limit that, procuring new sensors or installing new sensors unnecessarily. So I think that really resonates with me. I actually wanted to move on to predictive maintenance. We've started hinting towards it with some of this asset machine discussion. So I think overall, and I've heard this story before at the user group, but just a very brief overview of CooperVision's predictive maintenance journey, I guess would be a really good starting point.

Matt Walter, Coopervision: Sure. So when we embarked on the adoption of Industry 4.0, the first thing we did is we went out to industry. We needed to understand and identify what is Industry 4.0, what does it mean, and more importantly, what does it mean to CooperVision? After many, many discussions, attending many conferences and trade shows, we were able to pull together our Industry 4.0 strategy, which contained a number of elements, number of deliverables that we wanted to first of all pilot. So we ran a number of proof of concepts, very much an opportunity to test the technology, as with a lot of Industry 4.0, it was brand new. It was a concept, it was an idea. So we couldn't provide any guarantees to the business, but we could demonstrate the potential and the sound philosophy that the idea was based on. So one of those elements of our strategy was around predictive maintenance.

Quite simply, how do I predict when an asset or a piece of equipment is going to fail so I can make far more intelligent decisions around my planning and start to move the maintenance strategy from a reactive to more of a predictive and ultimately prescriptive model, allowing us to minimize downtime, avoid anything, or limit certainly the amount of unplanned downtime and manage things like our stock, our spares, and our resources in a far more intelligent way. So once we'd identified the strategy, we then went out to industry of, okay, who do we want to collaborate with to partner with? After investigations, we identified Senseye. Not only were they based just around the corner from us here on the south coast, but also at that time they were recognized as one of the market leaders for PDM solutions. So we agreed and started up a proof of concept, and very quickly we recognized that actually the integration mechanism was fairly straightforward.

Now, what I mean by that is, as part of our Industry 4.0 journey, we've implemented a brand new IOT platform, which essentially contained all of the key components to enable the extraction and connection of our manufacturing processes to a cloud-based solution such as Senseye. So we then followed the basic process of, okay, this is the asset that we'd like to monitor, that we'd like to onboard. We would typically take a time series, 10 to 20 minutes of the variable metrics that we have available to us. We share that with Senseye and explain or contextualize the data, who then explain to us, okay, well we want to know the various points in our trends. We would then manipulate our software to create aggregations of those specific data points. That would be then shared with Senseye on typically a 10 minute basis to fully onboard our solution.

Once the asset is onboarded, we go through a learning period, which essentially Senseye adapts to our process, our data, until essentially they've created a model that represents the asset, at which point we move into a full prediction model. We then work with the asset or the notifications that we receive from Senseye. Typically, it's an email notification, which we then use to identify which of our assets do we need to react to? Do we need to perhaps perform maintenance or at least an investigation?

But more importantly, which assets or notifications can we ignore? As with any new system, when you first start up, you get quite a lot of noise, but we recognized early that if we can explain why a sudden shift has occurred, for example, we've been changing machine settings or playing around with parameters to try and adjust the process, well then obviously the data's going to change, which sensor are going to pick up as a shift, but we can ignore it. We can acknowledge that shift. We don't need to disturb maintenance. That then allows us to refine and only notify our maintenance teams with shifts or potential issues that we believe they actually need to react to.

Niall Sullivan, Senseye: So I just wanted to go back a bit to what you were saying towards the start and what we've mentioned before about thinking about what's the business problem that needs to be solved here before the technology? And you mentioned downtime, which in most industries I guess is a big driver, but were there, outside of downtime, were there other drivers from the business for introducing predictive maintenance technology?

Matt Walter, Coopervision: So the other piece was around stores consolidation. So for example, if we know that asset A is likely to fail in the next two to three weeks, we can make sure that not only have we got the right spares available in stores, but also that we can coordinate and plan our maintenance resources to ensure that at the next planned shut down, it might be a batch change. Not only have we got the right spares, but we can also ensure we've got the right resources available so that we can make the component repair or replacement as quick and as efficient as possible. As I mentioned earlier, the accidental discovery, if you like, was that actually by onboarding our predictive maintenance platform, what we actually did is created a brand new data stream that we never previously had that has actually unlocked great correlations with our process, which when we combine with our raw materials and our quality data has unlocked greater insights than we'd ever previously seen.

Niall Sullivan, Senseye: And does some of that, obviously the spares part does, it influences other metrics like sustainability for example, because if you're not using as much spares, not consuming spares unnecessary, not doing stuff on a schedule, you can actually be more specific about that. Is that something you measure of in CooperVision, obviously you look at it as a company, but in the context of predictive maintenance, is that influencing any sustainability metrics on that side specifically?

Matt Walter, Coopervision: To be honest, not to a great degree yet. We recognize that absolutely monitoring and management of our spares consumption is key. There's a huge amount of expense and resource available or are required to support some of those activities. But at the moment, for us, the bulk of our value purely comes from the avoidance of the downtime. So they're more the metrics that we're focused on in order to demonstrate value to the business.

Niall Sullivan, Senseye: I guess just your opinion on this, maybe it's an Industry 4.0 wider thing, but is it a challenge for other businesses, other peers that you speak to about measuring these, the impact on sustainability? Is it something that is still progressing like it is at CooperVision?

Matt Walter, Coopervision: Absolutely, and I think with the world is shifting, everybody's become very, very green conscious, obviously with a recent dramatic increases in energy bills, sustainability and the consumption of energy is becoming more to the forefront of manufacturers and suppliers. So certainly it's something that we are looking to adopt and react to very, very quickly. But we are noticing a trend with some of our suppliers and some of our vendors that they're very much all going in the same direction, be it with the development of new sustainability based technology with the sole focus of looking to reduce energy consumption or it's looking at more efficient machine designs so that it requires less raw material to make as well as less energy to run as a final product.

Niall Sullivan, Senseye: Yeah, no, of course. And I think that's a good target because you've talked about it before and a really interesting process for onboarding new assets, but be I think, interesting for the audience to know what these assets are. What types of assets do you monitor using Senseye?

Matt Walter, Coopervision: So a lot of the assets we monitor Servo Motors now. We use a huge number of Servo Motors to manage and control the various processes throughout contact lens manufacturing. So that certainly represents a large portion. We've got an awful lot of motors that we are monitoring that are typically controlled via variable speed drives. They might be controlling things like pumps and compressors, vacuum pumps, for example, which uses a very different type of data stream to onboard. We are now moving more towards our facilities assets where although the basic compressor is typically a pump and a motor, we're actually starting to monitor far more metrics such as vibration and temperature and ambient environmental factors as well, because we recognize that it's not just the core data coming from the device itself. Often the ambient environment has an impact on the potential behavior and future predictions of the reliability of that equipment. So we're starting to broaden what data we're monitoring and extracting as part of our onboarding process.

Niall Sullivan, Senseye: And I guess the other question is, so you've selected assets, but what was that selection process? I guess it was in collaboration with Senseye, but what sort of things did you take into consideration about, let's say, which assets to onboard first?

Matt Walter, Coopervision: So we tried to make it as smart and as data driven as possible. So we largely relied on historic maintenance records. So rather than reliant on gut feeling and what a particular individual may feel, we actually went back through our maintenance records and identified which type of an asset, which particular machine or station has historically caused us the most amount of problems. And then tried to contextualize the cost of those problems so that we can then almost prioritize which ones we wanted to onboard first. So yeah, it's either based on purely the number of failures and therefore the stores and the consumption of spares, or it's the impact of a failure. So if something like a compressor plant, it may not happen as often, but when it fails, the impact is much, much greater than say an individual Servo Motor or a vacuum pump within a manufacturing line.

Niall Sullivan, Senseye: And in terms of onboarding assets, again, referring back to your process. I think that's really fascinating, but how easy was it actually to deploy predict and maintenance technology? What were the sort, I guess challenges you faced, but how smooth was that deployment?

Matt Walter, Coopervision: So there was very much two elements to our deployment. One was the technology side, one was the change management side, and I think we can confidently say one went a lot smoother than the other. So from a technology point of view, very, very easy. All of our equipment is very well networked, so we could get out the data very quickly. We have a number of platforms in place supporting our IOT projects that enabled us to get the data from the manufacturing lines and out to Senseye in a secure and reliable way. So technologically wise, very, very straightforward. We had some challenges around our maintenance records. So one of the requirements that Senseye has is that when a maintenance intervention occurs that there's some form of feedback or record of that maintenance intervention against the particular notification that they sent so that they can then potentially retune or increase or improve their modeling.

At the time, a lot of our maintenance records were stored in an Excel spreadsheet, not particularly great for automating. So we identified an opportunity and we actually designed and built a bespoke maintenance logging tool, which was fundamentally database hosted, so that as soon as a maintenance ticket was raised specifically against a Senseye asset, we could then automatically go and extract the relevant pieces of information that Senseye required and post it straight to Senseye. So there was no duplication of work, there was no manual entry required. So we found that that was really, really positive.

The change management side of things was far more challenging. Initially, there was almost a perception that predictive maintenance was trying to take over the maintenance engineer's role, which of course it absolutely is not. And we worked very long, very hard to try and explain that this is a tool to make your role more efficient and more effective so that you only focus on the important things, which should in theory, free up your resource and make your role more efficient. The introduction of the maintenance logging tool certainly went a long way because when we first started we didn't have that. So we were essentially saying, well, every time you create a ticket in Excel, we also need you to go and create it on Senseye via their online app. Well, the guys don't like writing out tickets twice. So safe to say it didn't happen very often. So although we were getting the notifications from Senseye, we weren't gaining the benefits of that feedback loop by populating the work events that Senseye require.

Niall Sullivan, Senseye: And actually on your first point, that system you built, so that's an interesting IT conundrum. It's like build versus buy. So I'm just wondering why you went down the build route rather than the buy route, let's say, for that particular piece of functionality or technology.

Matt Walter, Coopervision: Largely because we'd already built it, if I'm honest.

Niall Sullivan, Senseye: Oh, Okay. 

Matt Walter, Coopervision: So many, many years ago we've got a controls support team within CooperVision who are responsible for any controls and automation breakdowns or challenges that occur within our manufacturing lines. Now, it was a very, very complicated role with a huge amount of knowledge, but it was almost stuck in the engineer's minds and we wanted to have a very flexible dynamic way that we could record and create this huge knowledge base. So it was actually another innovation groups deliverable, was to create this tool that allowed our controls call out team to log and record the resolution to many of our controls breakdowns. Now, it proved to be really, really successful, particularly with new members that perhaps joined our team because immediately they get called to a machine.

They've probably never seen it before, but you could refer back to the maintenance tool, which would have a record of every time that machine has gone wrong, you could then filter through and find, oh, has this exact problem occurred before? Yes, it has. Okay, what was the resolution? Great. I'll just do the same again. Manufacturing is back up and running much, much quicker. The engineer is happy because they've been able to do their job. So it was very, very successful. The basic philosophy of ticket logging was then easy to adapt and modify to support our maintenance requirements. So yeah, the reason we went with build rather than buy is because we'd essentially already built it. We just needed to tweak the front end a bit to suit maintenance rather than our controls [inaudible 00:27:21].

Niall Sullivan, Senseye: And because you built it, did that make things easier to integrate with Senseye or did it make things easier or more of a challenge? Just interest-

Matt Walter, Coopervision: Definitely easier because the basic mechanism was already there. We worked with Senseye and identified well specifically what does the work event requirements, what are they for a Senseye ticket, then we could very quickly adapt and modify the application so that if a particular asset or station that the engineer is worked on contains a Senseye monitored device, it will pre-populate and automatically notify this is being monitored by Senseye. Can you please also add the extra information we need? We can fill that out very, very efficiently. Most of it was already pre-populated anyway, so when they closed the ticket, not only have they created a new record for their interest, we've automatically sent that ticket straight to Senseye. So because we had total control over the front end application as well as the database that was hosting it, it made that integration into Senseye very, very easy.

Niall Sullivan, Senseye: What I wanted to touch on is this importance, and it's something that we've identified internally at Senseye, how important having a Senseye champion, what we call PDM champion, to drive this project forward and be that go between, I guess, between the different departments. So could you talk about how that role has evolved or how it came about at CooperVision, how it's evolved to where it is today?

Matt Walter, Coopervision: Sure. So in short, it's absolutely vital, as far as almost to say essential. So for us, when we first implemented the Senseye solution, the notifications were being shared directly with maintenance. As I mentioned, when you first onboard an asset, you get quite a lot of notifications whilst the system tunes and refines and learns what is good and what is bad. The problem is that each of those notifications is an email which was landing into the maintenance inbox. Maintenance CooperVision, as with most companies, very, very busy. So we found that actually it was far easier to ignore those emails and not really do anything about it, but ultimately that then meant we weren't demonstrating any key value from the application. So it was something we identified during our monthly review process where we were looking at, well, how many cases have been raised, how many have we responded to or reacted to, and what value have we realized from it?

And it become very clear that the value and the benefits' metrics were very, very low. The business questioning, well, why is it not working? So we basically went back to our maintenance teams, we revisited our strategy and identified that actually the problem was that we were bombarding them with too many, therefore they were overwhelmed, so they were ignoring them. We therefore recognized the need for the PDM champion, which was a role that it was a particular member of production who was already heavily involved in our CI work. He was very, very well placed because he's got a great relationship between maintenance and manufacturing and had also been very much part of the predictive maintenance journey right from the very beginning. So incredibly well placed to support this. The role is essentially now such that all notifications go directly to them and them only. Their role is to now filter, identify which one of these can we explain and therefore we just close the case.

No need to worry about with maintenance. Which ones can we perform small internal investigations using some of our support technicians, and ultimately, which ones does the data suggest and the history of that asset suggest that we really have got a problem here, in which case those are the ones that we need to notify. So what would happen in that instance is that they would raise a new maintenance request using the maintenance application that we've discussed, which would essentially then become another job for the maintenance team.

But the fact that he'd already pre-populated much of the details, part of the details within the ticket would be justification or the explanation based on the Senseye data. It then reassured the maintenance engineer that received that ticket that, okay, well there's a valid case of why I need to go and have a look at this. The real value for us comes from when maintenance intervened and they find something because then you got the reassurance and the really positive impact of, well we've pre-filtered, we've asked you to go and take a look at something and ultimately you have found something that you were otherwise oblivious to, because all of our other monitoring systems hadn't yet detected it.

So it gave really good reassurance that these tickets that we are raising after our PDM champion has evaluated it, generally have a very positive output. So therefore the confidence in it, the engagement and the support that we get has grown significantly. From a metrics point of view, there's a real obvious step, you can look at all of our historical trends and charts and there's an obvious step. From the point at which our PDM champion started. Every value add metric that we're interested in has gone up significantly.

Niall Sullivan, Senseye: And actually that's quite timely. You mentioned metrics as well, and that's great to hear that they're becoming more and more confident in the system and what it delivers, and it's there to help not take over. It's actually there to really help them with their role. But can you summarize what's the types of positive results you've seen from deploying Senseye?

Matt Walter, Coopervision: So a lot of it is around the early identification of known problems. So some of our applications, it's often not the asset itself. So typical example being a Servo drive, but it's what the Servo is connected to. So it might be linear drives, it might be some kind of tooling. And often the problem is actually in the mechanical setup or alignment perhaps of the tooling or the end effect, not the Servo itself, but of course it's the Servo data that we are monitoring and we're aggregating, so it becomes a sensor. So we are no longer just saying that it's the asset itself that's failing. What we can say is that look, I've seen an increase in this metric, which suggests that something is going wrong. The maintenance then use their experience that actually if I see an increase in this variable, it doesn't necessarily mean that the drive is failing.

The tooling might be bent or misaligned or dirty. So we've seen a really good number of cases where that's exactly what's happened. They've gone in, they've identified a mechanical problem that they can either adjust or repair or replace, which has then reset and all of the values, all the metrics that we monitor through Senseye, reset back to nominal, confirming that the adjustments and the modifications they've done have absolutely worked. Often we've also had the double benefit where the slight damaged or misaligned tooling is also causing an increase in some quality defects. So not only have they now satisfied that the Senseye monitoring and our asset monitoring is now healthy, but we've also inadvertently increased our quality as well because we've essentially reset and realigned all of the setup.

Niall Sullivan, Senseye: And as we come towards the end, I have a couple more questions. I guess first of all, it's great to hear that you're on successful path with Senseye. I guess my next question would be, in all those different sites all around the world that you mentioned before, is there a plan to expand to different sites? Or I guess first of all is, are you sharing knowledge I guess, regularly with your colleagues in the different sites about the benefits predictive maintenance can bring to them as well? Are they as engaged as you or open minded to that as well?

Matt Walter, Coopervision: Absolutely. So again, it's that change management piece that we learned initially with our local teams, our local maintenance teams that were supporting the application. But we also recognize the value of our Industry 4.0 journey and the fact that we are a global company. Therefore, any value that we can realize, any new technologies that we'd like to adopt, we want to share that on a global platform. So we've been hosting regular updates with the senior management teams for most of our manufacturing facilities all over the world to keep them engaged, but also keep them updated with the latest success stories, with the latest tweaks or modifications to the process, the latest changes in our strategy and the improvements that we've been able to demonstrate as well as the core metrics, the actual value add measureables that we've achieved on our lines here in the UK. But the values and the way in which we're measuring those successes are directly relatable to all of our sites globally.

Niall Sullivan, Senseye: That's fantastic. And as a final thought, it'd be great if you'd summarize, I think in the area I'd like you to summarize actually, because you've been through this journey and had the bumps and the successes, but if another manufacturer was starting their journey looking at predictive maintenance tools, what are the common mistakes to avoid? What are the lessons you've learned, I guess from the deployment of Senseye?

Matt Walter, Coopervision: So I guess for us, it's probably the two biggest lessons learned. One was start simple. So in hindsight, looking back at how we used to onboard and identify the data sets for a particular asset, we went way too complicated. We were very much in our engineering mindset rather than our business mindset where because we could, we wanted to, but we recognized that in hindsight that it added so much complexity, become very difficult to manage, and ultimately we've subsequently learned it didn't offer any greater value. In fact, quite the opposite. We found that since we've changed our approach to how we measure the various metrics for a particular asset, and we've simplified it greatly, we are actually getting far more robust and reliable predictions from the PDM solution. So that was certainly a lesson learned that we've most definitely adopted.

The other one is the change management piece. Include and engage with all of your supporting teams right from the very, very beginning. Make sure that they're aware of what it is, what the impact it's going to have, what the potential benefits are going to be, but also what the expectations are because we did reach out to maintenance and ask, "Look, we kind of want you to change the way you work slightly to support us". Now we very much took the approach of, well look, we'll barter. If we give you a new tool that will make your day to day job much, much easier and more efficient than your old Excel spreadsheet, in return can you please help us out and support us with our PDM deployment? Now that actually worked very, very well. I'm not saying that all other manufacturers should do the same, but it is in our experience that did work. But certainly engagement and inclusion from the very beginning, not necessarily on the technical deployment because that was very much under the scope of myself and my team, but ultimately they're the end user, they're the customer.

So what do they want? What do they need? How do they find not only our integration, but the Senseye app itself? Senseye offers huge amounts of training, but that training program has evolved over the years and in many cases has got a lot simpler, which was one of the comments we had at the early stages was that it was quite difficult to follow and to understand. Now the training has been broken down into much smaller modules that the team they can fulfill. It takes them now 10 minutes, which a maintenance engineer can find rather than an hour, which is quite a big chunk of time to find. So again, it's that what works good for you? What do you need? How can we incorporate this new approach to maintenance in a way that benefits you in the best possible way? And then how can we keep you notified and up to date with any developments and improvements that we'd like to make, but make sure that you agree, because ultimately it's you that's going to be affected by it.

Niall Sullivan, Senseye: Yeah, no. I don't know, that's a really nice way to finish. So yeah. So thank you for joining us again, Matt. That long journey around the corner. Hope you have a safe journey back, but it's been a real pleasure to speak to you again and some really interesting insights as well. So thank you again. Thank you for our audience for joining us. Hope you got a lot out of it and look forward to seeing you on the next episode.

So that was the second and final part of our conversation with Matt Walter at CooperVision. Predictive maintenance deployments comes with challenges, but by getting stakeholder buy in early, you give yourself the best chance of success.

Please subscribe by your favorite podcast provider if you'd like to be notified about future episodes. And it would mean a lot if you could let us know your feedback by leaving us a review. You can find out more about how Senseye can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants by visiting

Subscribe to our Trend Detection podcast

We post weekly episodes on a wide range of different subjects including predictive maintenance, industry 4.0, digital transformation and much more.

Here are all the places you can access/discover the latest episodes of the Trend Detection podcast: