The key maintenance challenges manufacturers are facing - Part Two - With Andy Gailey

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 three part series, we're joined by Andy Gailey, founder UPTIME Consultant Ltd who provide a holistic approach to asset care and maintenance management.

In the second episode of this series, we discussed the key benefits of implementing predictive maintenance technology and how to fit this within your broader maintenance strategy.

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. Key benefits of implementing predictive maintenance technology
  2. Defining Industry 4.0 
  3. Artificial Intelligence
  4. Subscribe to our podcast

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 three-part series, I'm joined by Andy Gailey, founder of UPTIME Consultant Limited, who provide a holistic approach to asset care and maintenance management. In the second episode of this series, we discussed the key benefits of implementing predictive maintenance technology and how to fit this within your broader maintenance strategy. I hope you enjoy it.

Yeah. And I just wondered about the bathtub effect because another thing... We talk a lot about unplanned downtime, but obviously there's so many other benefit, just from a predictive maintenance perspective, but it's actually incremental improvements that you can make. So it's actually not just about preventing failure, but it's actually about making incremental, so those early signs where you can just make tweaks to keep machines, for example, running efficiently, because if a machine's not running as efficient, it's wasting more energy, et cetera, but if you make those little tweaks, those early warning signs we said earlier, then you can keep machines running at a more efficient pace. So, are those benefits starting to be recognized, I guess, in particular, from technology or condition monitoring standpoint?

Andy Gailey: Yeah, I do see some predictive efforts, condition-based maintenance, whereby people are carrying out something they have been employed to do, and they end up with a lot of information but no action, and the thing is, if the people I train to start out as predictive engineers or proactive engineers, you've got to concentrate on the output. All of the tools that we use are just tools, they're just tools that accentuate one of our natural senses. So thermography allows us to see energy emitted in a color map so we can understand where the energy "hotspots" are.

Ultrasound allows us to hear wave forms that take place outside of our natural hearing through heterodyning. So we can then listen to, for instance, a bare impact with the grease in through headphones, and monitor what that sounds like, and if you're trained at this, you can actually identify whether there's too much or too little grease, whether there's impacts occurring, and if you understand rotational speed and the sub-synchronous rotational speed that you'll find damage at, then you can do things with your natural senses with the tool, that's what I did for nine years.

I used all of these tools and brought them together, but the idea of me using those tools wasn't just to use the tools, it was to make an impact on the business, it was to make the availability flat line, and I talked to this about to somebody just this morning that, when we talked about large ovens, when I first started, there was repeated failures of oven bearings that were very expensive to repair and interact with, because they were very hot, you had to wait actually a couple of hours for them to cool down, before you could go near them, and it was a lubrication problem, and we identified the lubrication problem through acoustic commission.

So we used a tool of acoustic commission to understand it was the lubricant that was the issue. So where we thought there was planned maintenance being carried out, it actually was being ticked off and wasn't being carried out. So the people either didn't put the grease in or they couldn't find a gun or they ran out of maintenance time, so they timed it out, or they actually found a grease gun with incompatible grease in and they just put that in, because if you haven't trained anybody in lubrication, people think that all grease is just grease and all oil is just oil, and if you get an incompatibility between greases, you end up like we did with the actual chalk inside the bearing pack, because they'd reacted together, they'd formed a compacted hard white chalk like substance, it was the aluminum complex that had dried out because it was interacting with a semi-synthetic grease.

So the thing is, you then put in systems to take that out, so it never happens again. So, even though my job was as the practical predictive engineer, I then went into the CMMS system, and took out the lubrication approach at a timed interval for what they were doing on the shop floor, and I owned it. So I owned that and would only lubricate with the right lubricant on condition. So if acoustic commission told me it needed lubrication, lubricate it, and actually monitor it in real time to see that impact level reduce as we did the greasing, and then do that just on a seven day cycle. Once we were confident that we'd got the flat line that we needed, the equipment was, we weren't actually going there seven days and not doing any greasing, we put that into flight, 14 days. So straight away you start, the savings go, we put the unplanned down time to zero over two years in the high teens.

We stopped using as much lubricant as being booked out. We did it on conditions so we only lubricated when it was required. So therefore all the planned aspect that cost hours, that was a saving, so we took that as a saving, and the thing is what the prior to predictor engineer should gather all that information together because you will be asked, well, you should be asked by your operations manager and plant manager, how much you've saved within a period or a quarter or an annum, and that's where people find it hard to get benefit out of lubrication practices, but if you time with predictive technology, there's masses of gain to be made to put in the right lubrication, at the right time and in the right quantity, and most lubrication, rotating failures come from lubrication, it's either over or under lubrication or no lubrication.

Niall Sullivan, Senseye: And actually that... Well, what you're explained is really interesting, Andy, because we've interviewed one of our, well, I'm sorry, I've interviewed one of our customers in the US and they refer to predictive maintenance technology as another tool in the toolbox. So every day a theoretical toolbox, and then they'll assess the situation and decide, "oh, I need to apply this technology or this approach or this approach", which I always like that analogy, I repeat that quite a lot, which I thinks reflective in what you are saying.

Andy Gailey: The thing is, people should look at a maintenance strategy as an overarching piece of work, and it includes people, it includes resources, things like inventory, [inaudible 00:08:07] to be in control. So one of the things that I wrote about a couple of years ago, I've just revisited it, is if you've got a great proactive predictive strategy in place and you can pickle where a bearing is in production today, that there's a window in seven days time where there's a planned outage to do, say, a flavor change or a product change and you can put somebody against that with a part to change [inaudible 00:08:34] within an hour, then that's a massive saving, because you've avoided an unplanned down time in the future, that person being employed from a proactive point of view has the spare in hand and understands what I've got to do, they're prepared, they're waiting for the line to go down, they're literally like a cold spring late of it.

The problem is a lot of people do is, they do all the great work ahead, but they don't look at the inventory, and if you do that great work ahead and then go, "oh, well we haven't got a bear in stock, because we haven't covered our angle and it's on 14 days leads on", that's where you get, even though you do all the good work, you can be tripped up and still end up with the unplanned downtime because you haven't covered the inventory level, you haven't covered the full picture. So it's important that people look at predictive and proactive things that we do as part of the bigger picture, and that includes obviously people's knowledge and their training, their understanding of the failure modes of the equipment, their understand say and proper mounting of bearings, it's amazing how many engineers in maintenance that haven't had a basic cause on how to fit a bearing properly without damaging it.

Niall Sullivan, Senseye: No. 

Andy Gailey: So yeah, your client's right, it should be seen as a tool that is used in a range of tools to get to a place in the future that you wanted, that you planned for.

Niall Sullivan, Senseye: Exactly. And speaking of the future and technology, I know we had a conversation just before we came on the podcast about industry 4.0 and maybe that's not exactly your specialist area, but of course you've talked to customers who either investing, discussed in that area. So it'd be interesting, first of all to get your definition of industry 4.0x. I think there's a lot of differing, and I've certainly heard it at different events and stuff about different approaches and views on it. So interesting to get your thoughts on that.


Andy Gailey: Yeah. So if you look at the phrase industry 4.0, it was actually coined 2011. Again, I've googled this in various places. It came about from the German government, so it was a thing that was banded around the WEF, which is an unusual organization where you go and look at the WEF, an Electric bureaucrats. So they came up with this idea of industry 4.0, and then they came up with story of, "well, we started at industry one where we discovered oil and we made steam trains and Brunel", and all of the very things that were pre, probably, the Victorian era, and then we came into the Victorian era, that was industry 2.0 whereby, say, Ford came along with automation, things were still very early days in engineering and manufacturing things, and then they could probably then spin forward to probably post second World War, a lot of discoveries were made during the second World War, purely through carrying out mass killing of people, where they came up with things like transistors. So you went into the second World War with valves and came out with transistors.

So they had this timeframe of this is what happened, and then 2011, somebody said, "we're heading to or in industry 4.0", and they saw it as connecting technologies through... Originally, the easiest way to do it is through consumer technologies, so that's where all the hits have been made really, so the iPhone and apps on your iPhone and things like that.

And things start there and then they tend to migrate to the industrial space. So it's been for, I would say, probably since 2015-ish, it's been that kind of push to, "let's get this into the industrial sector because we could see that there's benefits of having people using IoT devices in the real world, in the everyday, then surely there's benefits from it being IIoT", so industrial internet of things. And there's obviously lots of people that make their money through quoting industry 4.0 or IIoT, and a lot of them, that I've seen, are promised things that will be beneficial in the future, and when you go... Me, as a practical engineer, when I go and look at some of the benefits, I can't see that some of them approach reality.

And I was looking, when I started UPTIME Consulting up, I had a business plan, the business plan was to, it's probably in thirds really or quarters, I was going to spend a third or quarter of my time working for clients, so actually going on side client and training their people and trying to get them to think the same way as I think, because that's what I'm turned on by, and then I was going to use another piece, I wanted to go online and train people for free online, so give things away, because that interests me and I can afford to do it, which was good. And then the other part, I wanted to always look to the future and see how the proactive predictive things I was doing in 2005 could be extrapolated and automated, and when I spoke to the co-founders of Senseye back in probably 2015, 16, I was interested in Senseye because, A, they were in the predictive space and it was industrial.

So these are two things that I'm interested in, and they were talking of automating what I did as a manual subject. So it was something that I dreamt about in 2005, when you're on a 12 hour shift and you're going out and taking 300 acoustic commission readings and two oil samples and a whole lot of other readings have come out of assets and then tried to extrapolate and find out what 3% of that makes any sense to action, that becomes a long and arduous task, fortunately for me, it interested me that much that I was just turned on by it, and I understood over time as well that if you use some perito, you could drive towards 20% of assets that were causing you 80% of your headache and maybe let some of the low line stuff go into reactive, maybe just check it every couple of months.

But what I had in my head round was, if I could take 2000 readings every 12 hours but not have to physically leave my office and walk around every asset in the plant and then bring it back and download it into a database and then look at either trigger points or things that I'd found of interest when I [inaudible 00:16:12] around, that would be a really great benefit because then you could look at hundreds or thousands of assets over different locations in the world as well. So you could compare a robot in North America with a robot in Spain of the same type, and look at the conditions that they're running and gain a lot of information out of that, obviously using things that have become prevalent over the last 10 years, some artificial intelligence, really machine learning, to use those tools.

And again, they're just tools, those tools that our brains aren't very good at. So our brains are an amazing tool because I can think of 10 different things at the same time when I'm visiting an asset, and I can also at the same time speak to a human that's running that asset for 12 hours and ask them their opinion of what's happening. I can put all that information together in my computer and come up with, do I need to take action or do I not? And really that was my job with the industry I worked in.

Is this important enough to take action or can I move on? So going back, industry 4.0, go back to 2011, it was a construct, it was similar to AI being overused, AI, everything, I've actually collected some screenshots of electric toothbrush that is AI, artificial intelligence, enabled and things like that.

Niall Sullivan, Senseye: Yeah.

Andy Gailey: AI has been overused, really it's machine learning, it's a bit of maths, and the same way a lot of people jumped on the industry 4.0 name, and when you look at it, you go, "well, does this apply to a connected technology in any way?" And when you look at it, it doesn't, people have very little understanding and have used it as a tool to PR their product.

Niall Sullivan, Senseye: Yeah, no, I couldn't agree more. And actually I was at an event last year in Germany and there was actually a talk or a discussion about Industry five, and I thought, well, if you haven't even fully covered the landscape of Industry 4.0, if that's still developing, which, I guess, it probably is, that we shouldn't be talking about things like that yet, not for another 10, 15 years.

Andy Gailey: I think I mentioned it before we started that PDM, predictive maintenance, again, is a little bit of a misnomer, and is one that I used early on in my career when I was looking at maintenance and predictive and proactive, and it's something I realized when I thought about, I thought, there's nothing crystal ball about this, this isn't really predictive, this is the earliest sign of the incipient failure, it's actually the first reactive, this an early reactive event. So we look for a signature in a piece of plant and where it should be running and then we monitor that and then we try and look for a signal, maybe up, maybe down, but a signal that may trend over time that gives us more information about what path that asset or that component is on. It's very simple, people try to overcomplicate it, it's not, it's very simple, that if you can extrapolate that, spread it over many assets, you then have a more enlightened workforce that can be targeted at those 20% that I talked about, that take 80% of all your maintenance inputs and over time...

So over time, what I did, I looked at nearly everything in the process area, in the factory I worked in and over time, after a couple of years, I pared it back. I thought, actually this asset has proven to be very reliable over 18 months, I know we could be looking at something another six months, but there's redundancy built in, I don't really need to look at it every 30 days, let's go look at it every 180 days say, and that worked the same with oil sampling which is a predictive type of methodology, oil sampling and oil monitoring in gear boxes. So when I started, it was 12 months cycle, no debate, you were going to change our 40 liters of food grade oil, very expensive operation. I started doing monthly oil samples and I trended over time the health of that oil pack, and they also looked to...

Other things that were telling me what's going on inside the asset, and then we pushed oil changes out to 3, 4, 5 years based on condition. So, it's a hell of a lot of plan time. The oil samples turn out about 20 pounds each per month, we pushed that out then to two, three month samples, because again, it wasn't proven beneficial. So it's a learning phase, if you're within a plant and doing it in one plant, it's a learning phase, learn about your assets, getting some knowledge about how things fail, looking at the history, talking to your operators, understanding what operations require.

Niall Sullivan, Senseye: So that was the second part of our series, looking at the key maintenance challenges manufacturers are facing. I hope you enjoyed it. Please subscribe via 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, since I can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants by visiting Thanks a lot for listening.


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: