WEBVTT
00:00:00.160 -->
00:00:05.839Vibration analysis, the different ways of acquisition, what we can detect with it, what's not very good at.
00:00:06.639 -->
00:00:08.160Hello, Reliability Gang.
00:00:08.240 -->
00:00:11.279Welcome back to another episode of the Reliability Gang podcast.
00:00:11.439 -->
00:00:13.039Hope everyone's keeping well.
00:00:13.199 -->
00:00:17.039And with the with the with the man, the man of the man of the moment, Will Crane.
00:00:17.199 -->
00:00:17.920How we keeping, buddy?
00:00:18.160 -->
00:00:18.800Good, thank you very much.
00:00:18.960 -->
00:00:20.239We've got the endorphins flowing.
00:00:20.320 -->
00:00:21.679We hit the gym this morning, don't we?
00:00:21.760 -->
00:00:25.280Before the office, had a bit of a workout, a bit of a pump, the sun's shining.
00:00:25.359 -->
00:00:28.480I feel like summer is on its way, and I feel like, yeah, I feel good.
00:00:28.640 -->
00:00:30.640I'm feeling better within myself in general.
00:00:30.800 -->
00:00:31.440Do you know what I mean?
00:00:31.679 -->
00:00:33.039Also, it's been a crazy week.
00:00:33.200 -->
00:00:33.840Just got back.
00:00:34.000 -->
00:00:36.159I'm just still recovering actually from Austin.
00:00:36.240 -->
00:00:40.640That was an absolute wild experience to be honest, Accelerate.
00:00:40.719 -->
00:00:51.039So big shout out to the guys at Fluke for inviting me and lots of fresh new ideas to bring back into the mix, which got me thinking, actually, which has kind of drove us into this new podcast.
00:00:51.119 -->
00:00:56.079So lots of really cool technology that we're seeing with the vibration analysis stuff that Fluke doing.
00:00:56.159 -->
00:01:04.159And it got me thinking, actually, because we've uh we've kind of really tried to cater the vibration model to be able to say how do we get and extract the biggest value?
00:01:04.400 -->
00:01:15.760And I think for us to be able to do that, today we're going to talk about vibration analysis and how can we utilize that technology for different companies and extract the value by different ways of doing it?
00:01:15.920 -->
00:01:26.480So whether that's us doing it handheld every single month and coming in as an actual expert and specialist, empowering people to do it themselves in terms of data collection, maybe we can analyse the data.
00:01:26.719 -->
00:01:33.040And also, how has technology been able for us now to unlock these ways of doing things?
00:01:33.280 -->
00:01:41.439Because I do believe that five, ten years ago, this was a lot more difficult to do because a price point was really difficult, but not even that.
00:01:41.680 -->
00:01:48.239B is mentorship and knowledge and teaching has changed within that whole perspective as well.
00:01:48.480 -->
00:01:53.760So, Will, give me a bit of uh introduction into the evolution, shall we say?
00:01:53.920 -->
00:01:56.480I mean, I think I mean even the evolution of us and Von Hash and the other.
00:01:57.359 -->
00:02:09.039I think all we have to do is explain our journey, but also explain to us how technology kind of maybe 10 years, even actually, maybe stay, maybe state it from when you first started at maintain to how it's evolved even to now.
00:02:09.120 -->
00:02:10.879Because I mean I think that's kind of a decade, isn't it?
00:02:10.960 -->
00:02:12.400It's roughly about seven years, roughly.
00:02:12.879 -->
00:02:20.159I mean, even when I was at British Sugar beforehand, I think that the general concept of vibration analysis for many years has been handheld data collection.
00:02:20.400 -->
00:02:22.479Yeah, yeah, it's kind of been the king of that, isn't it?
00:02:22.800 -->
00:02:25.520And it still is a very cost-effective uh way.
00:02:25.759 -->
00:02:39.759When we talk about vibration analysis, uh, obviously there's different ways we can acquire that data now, which has been evolved quite massively from wireless sensors to handheld data collection to continuous monitoring to protection systems.
00:02:39.840 -->
00:02:42.560There's lots of methods of acquisition.
00:02:42.719 -->
00:02:55.599Uh, but I think what we're kind of talking about a little bit with vibration analysis is understanding kind of what what is it good at, what is it not so good at, why would I need to acquire the data differently and what that means.
00:02:55.759 -->
00:03:06.719And I think definitely you say what it's not good at, where is it giving a false sense of security with certain things that it's not actually detecting some of the things that we're trying to do?
00:03:06.879 -->
00:03:14.560Like a lot of the time, I do see a lot of companies trying to over-condition monitor things, and it's like, mate, we can't, this ain't gonna find the problems here.
00:03:14.639 -->
00:03:21.680We need to do some some brute cause and defect elimination because what you're doing essentially is trying to find something that it can't find, you know.
00:03:21.759 -->
00:03:23.039That that happens as well, doesn't it?
00:03:23.439 -->
00:03:33.599We see it a lot of the time, especially from like it the problem is you have to get into the technicalities of vibration analysis to then be able to evaluate whether it's suitable or not.
00:03:33.759 -->
00:03:40.159And so, particularly, we see that with wireless sensors, they have more limitations than say handheld data collection has.
00:03:40.319 -->
00:03:47.919Continuous monitoring is very effective, but then you lose the engineer at the machine, but then you gain lots of trend data and lots of ability.
00:03:48.159 -->
00:04:14.719Yeah, and I think this what this is a podcast is going to be more about is educating people and giving awareness to what systems are available because again, this this system, there's something for everything out there, but the problem is what really actually fails is not the fact the vibration analysis is not working, it's the strategy to understand what equipment we need, what frequency responses we need to capture, which is more kind of driven off how that machine can fail and what we're trying to see, right?
00:04:14.800 -->
00:04:15.280And what we're gonna do is that we're gonna be able to do that.
00:04:15.360 -->
00:04:22.000And I think that's probably the first really good starting place for anyone listening, is is some basic vibration analysis one over.
00:04:22.720 -->
00:04:25.199Let's go back to let's go back to VA level when we said, you know.
00:04:25.680 -->
00:04:35.920The two things that massively affect, I think, vibration analysis across the product market space when it's wireless or wired or whatever, is how long can we acquire the data for.
00:04:36.000 -->
00:04:45.439So the time waveform, which is the the raw signal that we're collecting, we can change the parameters to determine how long we measure for in seconds in time.
00:04:45.759 -->
00:04:53.360And particularly for slow rotating assets, we want to be really aiming for six to ten rotations of the shaft, right?
00:04:53.920 -->
00:05:09.600And so, therefore, the system that we're choosing, whether it's portable, wireless, or whatever, depending on the rotation speed of the shaft, with the bearing, with the gears that we're trying to monitor, we need to ensure that we're able to measure long enough to capture that number of rotations.
00:05:09.920 -->
00:05:26.000And that's a great point as well, just to add to that, you know, in regards to knowing that if you are have got applications that you know are slower moving in nature, we have to make sure that we have the right resolution to be able to have that long enough time waveform to capture them events.
00:05:26.160 -->
00:05:36.560So, this is why sometimes wireless monitoring can be difficult and has its limitations because it's not necessarily designed to draw that battery life to be able to take them along the readings.
00:05:37.360 -->
00:05:37.680So, yeah.
00:05:37.920 -->
00:05:55.439And your other main parameter that we're looking at is what you may see on some of the technical data sheets is what is the maximum frequency that the sensor can resolve in terms of its spectrum, and that's going to determine how early we can identify the faults and what faults we can see within the data.
00:05:55.680 -->
00:06:10.399So the quicker it can sample, the higher frequency, because obviously if you if you imagine higher frequency events happening a lot quicker, yeah, well, then we have to sample quicker the signal to be able to capture them high frequency events, if that makes sense.
00:06:10.959 -->
00:06:11.360Exactly.
00:06:11.600 -->
00:06:34.800So, I mean, if you have a shaft that is rotating at, let's say, 25 Hertz or 1500 RPM, so it's like an input shaft on a gear, and you had 10 teeth on that gear, we would expect to see a gear meshing frequency of 250 hertz.
00:06:34.959 -->
00:06:39.199I'm not going to do the 1500 or is it 15,000 cycles per minute?
00:06:39.439 -->
00:06:39.600Yeah.
00:06:39.759 -->
00:06:51.759I think it is if your frequency range is less than 250, which is very low in real spec, but if it was lower than that, you wouldn't see it.
00:06:52.639 -->
00:06:53.279You wouldn't see it.
00:06:53.360 -->
00:07:05.680That's that's the principle there to understand is that we need to make sure is that uh the issues that we're trying to identify, and I think for bearings, we're looking for what is it, six times the the harmon uh fault frequency we want to be able to see.
00:07:06.000 -->
00:07:09.600I think for envelope, what you're trying to see is the fourth harmonic of that bearing, really.
00:07:09.680 -->
00:07:10.480That's what you want to see.
00:07:10.639 -->
00:07:15.120So for in a race that is more higher frequency, and this is another thing as well.
00:07:15.199 -->
00:07:25.920What we're doing, like we're saying, right, we're gonna set our system them up to be able to find or see the picture where the faults are gonna lay within that frequency range, and that's what we need to try to do.
00:07:26.000 -->
00:07:35.120It's so when when we're looking at strategy and setting the database up, what we don't want to be doing is measuring a stupidly high frequency, so all of our frequencies of interest are down the bottom.
00:07:35.439 -->
00:07:43.279We want to be able to really window between exactly where we're gonna see the last particular frequency of interest that we're interested in.
00:07:43.439 -->
00:07:51.519So this also is how band alarming does work, and certain different fault conditions will appear within the spectrum at different frequencies.
00:07:51.680 -->
00:07:56.720So ones that we are particularly interested in or we know are damaging to the machines that we've got.
00:07:56.879 -->
00:08:07.360So, say for example, you've got a straight couple machine, you know, we want to be able to potentially see the two times um, you know, running speed frequency for misalignment is one of the fault indicators that we're trying to look at.
00:08:07.519 -->
00:08:11.920Maybe you want to see that in the envelope as well because it's going to be able to see if there's any coupling stress.
00:08:12.160 -->
00:08:18.079We're able to signify where we really want them band alarms to tell us what problems we could have.
00:08:18.240 -->
00:08:22.879And this is, I think, a really important thing to do when within vibration analysis.
00:08:23.120 -->
00:08:26.959We need to lean it off the failure modes of the machines we're working with.
00:08:27.120 -->
00:08:32.159Whereas I do feel a lot of the time it's like, oh, vibration, we'll just put loads of sensors on everything.
00:08:32.320 -->
00:08:32.559Yep.
00:08:32.720 -->
00:08:52.000I'm not saying that we're not capturing the frequencies that we need to, but if you're not really pinpointing what you want to see, if that makes sense, this is where you're gonna not have the clarity about how to make the best decisions within your maintenance functions because your alarms might be going off, but you have to also remember sometimes a low-level alarm at a certain frequency is a problem.
00:08:52.159 -->
00:09:14.799And if you're not actually trending that and actually looking at them certain places, the vibration is just a vast spectrum of information coming in that doesn't actually mean anything until you put meaning to what you're trying to find, and this is also what we're trying to evolve into now as well as a company, because it's not a case now of just saying, Oh, you've got a hundred assets, we're just gonna test a hundred assets.
00:09:14.960 -->
00:09:27.600You know, a lot of the time we are, you know, even the data handheld data collection model is really outdated in the sense that a very high critical asset is getting the same attention as something that is could be run to fail.
00:09:28.080 -->
00:09:39.519So, how are we doing, like how are we actually now assigning, you know, in terms of criticality the equipment, and how are we actually prioritizing putting more effort or more kind of monitoring on the things we want to find?
00:09:39.679 -->
00:09:47.840And this is this is where it's evolving, I feel, because uh five, six, ten years ago, online systems inherently were expensive, you know what I mean?
00:09:47.919 -->
00:09:52.000It was the cost of entry was very difficult to be able to justify.
00:09:52.080 -->
00:09:55.440So, this is why hen handheld vibration now is so good, right?
00:09:55.519 -->
00:10:04.480Because it it was you was able to come in and you was able to get a very good sweep, and albeit you might only be taking a reading every single month, you can still find some great problems with that, right?
00:10:04.960 -->
00:10:10.639The one problem and limitation that has though, especially on a complex machinery, is a snapshot in time.
00:10:10.960 -->
00:10:16.639And to make a decision on such a critical piece of equipment, we sometimes struggle even on monthly day as well when we see it go up.
00:10:16.960 -->
00:10:21.840But especially for us as well as contractors, where we do have to make a decision within that snapshot.
00:10:22.159 -->
00:10:23.039That's what you get paid to do, right?
00:10:23.200 -->
00:10:25.279We we have that one opportunity we've got to make.
00:10:25.519 -->
00:10:27.759We don't have the ability to go, oh, we'll go test it again tomorrow.
00:10:28.000 -->
00:10:28.080Right.
00:10:28.159 -->
00:10:30.320Well, I'll go with my data collector and go tomorrow.
00:10:30.480 -->
00:10:42.879So this has also you know got us thinking as well because now we have a model with some of our customers where they, you know, they collect the information, is when we do find a problem or they find a problem with their alarm sets, is like, okay, we're gonna take it every day now for a whole week.
00:10:42.960 -->
00:10:49.519You know, this ties into there's a there's a piece I want to come to in a little bit around uh sensors and and AI, but I'll come back to that in a little bit.
00:10:49.600 -->
00:10:57.600But I think what would be really good is maybe we can bounce off each other a little bit between portable data collection, maybe wireless systems and continuous monitoring.
00:10:57.840 -->
00:10:58.480What are the pros?
00:10:58.799 -->
00:11:00.799What are the cons of each of those?
00:11:00.960 -->
00:11:04.720So portable data collection, we're gonna all bounce back and forth, right?
00:11:04.879 -->
00:11:09.039Obviously, portable data collection, we have the capabilities to have a very wide frequency range.
00:11:09.200 -->
00:11:12.320Yeah, good high quality data, high quality data, high resolution.
00:11:12.480 -->
00:11:18.799We have no limitation on the shaft speed whatsoever, and we can tailor those measurements whilst we're there.
00:11:18.960 -->
00:11:19.600Yeah, 100%.
00:11:20.000 -->
00:11:20.799Real important process.
00:11:21.759 -->
00:11:24.960Another pro of data collection, you can collect a lot of now.
00:11:25.039 -->
00:11:28.240Obviously, technology and data collectors has advanced as well.
00:11:28.399 -->
00:11:31.200It's a lot quicker now to collect the information and data.
00:11:31.440 -->
00:11:34.240You can cover a lot of assets within one day.
00:11:34.320 -->
00:11:38.480So, value-wise within a day, it's cost-effective, it's cost effective, 100%.
00:11:39.120 -->
00:11:41.519Yeah, uh portable data collection.
00:11:41.679 -->
00:11:53.679Okay, so a big thing with portable data collection that you're not really gonna get in either continuous or in uh wireless sensors, is you have got a technician at that machine.
00:11:54.000 -->
00:12:02.960So if they see anything in the data that is abnormal, they want to take an additional measurement, they want to change the measurement parameters, they want to have a look and a thing.
00:12:03.120 -->
00:12:07.600Yeah, they want to do some more troubleshooting, maybe they can do it there and then at the machine.
00:12:07.840 -->
00:12:08.240Exactly.
00:12:08.320 -->
00:12:16.000And you know, to add to that, if they want to do any phase measurements with you know a couple of single axis sensors, they can go there, troubleshoot a little bit, get a bit more clarity.
00:12:16.320 -->
00:12:20.080And again, observations, you know, they can see an oil leak.
00:12:20.159 -->
00:12:22.799Sometimes you don't hear what's happening with the machine.
00:12:23.039 -->
00:12:23.519Exactly.
00:12:23.600 -->
00:12:35.919So there's being there and being, you know, doing the analysis by the machine sometimes that enables an expert to be able to dig a little bit deeper, even with not having one a few readings in time.
00:12:36.080 -->
00:12:43.600You can use your senses around it that allow you to be able to make a better decision and close the loop on what the decision needs to be.
00:12:43.840 -->
00:12:48.480And I think that there's a lot of positives to data collection with handheld.
00:12:48.559 -->
00:13:02.399Um, there's not many cons to it because we have so much capability within the data, but I think the biggest con with portable data collection is the fact that the skill set that is required to collect the data is somewhat limited.
00:13:02.559 -->
00:13:02.879Yeah.
00:13:03.120 -->
00:13:12.399Like you do need to know what you're doing, it's not the easiest subject to learn, and it's not something I'm gonna go on a training course and suddenly I can go collect data and I know what I'm doing.
00:13:12.559 -->
00:13:13.200Exactly, yeah.
00:13:13.279 -->
00:13:15.120And I think obviously there's two parts to it.
00:13:15.200 -->
00:13:20.879I say I think with the training, it's becoming a very good observationist, I say, and good data collector of understanding.
00:13:21.039 -->
00:13:22.639Well, where do I collect the data from?
00:13:22.879 -->
00:13:25.279If I get a high reading, do I collect more data?
00:13:25.360 -->
00:13:27.360What do I look for through that process?
00:13:27.600 -->
00:13:32.000That that that can be taught, you know, to anyone, to any engineer, really, to be fair.
00:13:32.159 -->
00:13:35.759But again, it's so important that that's done correctly.
00:13:35.919 -->
00:13:38.639Uh, and we'll get on to the reasons why with when when we get to that.
00:13:39.039 -->
00:13:47.120And I think that you have to look at like data collection, whether or not you have a contractor coming to do it, or maybe you have built up some capabilities to do in-house.
00:13:47.360 -->
00:13:48.720There's two parts to that.
00:13:48.799 -->
00:13:54.720Like, are we treating going around with the vibration analyzer as just data collection?
00:13:55.440 -->
00:14:05.440So is your current service provider just collecting the data in the zone, or are they analysing as they go, doing additional tests?
00:14:05.600 -->
00:14:08.480Uh ideally for us, you know, we've done it for a number of years now.
00:14:08.559 -->
00:14:10.960Uh, I was with a customer the other day explaining the concept.
00:14:11.039 -->
00:14:16.159I said to her, I said, realistically, I'm not leaving the site not knowing what the issues are.
00:14:16.480 -->
00:14:16.720100%.
00:14:17.200 -->
00:14:18.879I know what all the problems are before I've left site.
00:14:19.120 -->
00:14:23.679Yeah, and you've got the cover, you've had that conversation with the you know, with the engineer manager on site as well.
00:14:23.759 -->
00:14:25.759They're aware of everything that you've found.
00:14:26.000 -->
00:14:33.679And you know, it's it you almost care for the site like it's your own because you're on a mission there to find the problems, and that's what you're paid to do.
00:14:33.759 -->
00:14:46.080So I do feel that there's a huge value in having that because what you've got now is that person is almost looking to find things and an eagle eye out there, they're not fault blind to the plant because then that's what they're there to do.
00:14:46.320 -->
00:14:49.200I suppose though, uh a really good uh this is a good con.
00:14:49.279 -->
00:15:04.000I say good con, there's no good cons, but it's uh a good point to make around portable data collection, is it in general, from an expertise perspective, especially like for us, we send our engineers, and generally we've taught them to be able to analyze the data.
00:15:04.080 -->
00:15:11.360We don't tend to send data collectors from sending high-level vibration analysis, typically all our guys are level two.
00:15:12.240 -->
00:15:14.000In general, it's quite inefficient.
00:15:14.240 -->
00:15:18.399Yeah, you've got maybe a hundred assets, ten problems.
00:15:18.639 -->
00:15:27.200Yeah, and that engineer is on site collecting data for those hundred machines where only maybe 20% of them require of analysis.
00:15:27.440 -->
00:15:27.679I agree.
00:15:27.759 -->
00:15:29.039That's I think that's a really good one.
00:15:29.200 -->
00:15:34.639I think that's a and as well, I think for me, is the limitation of when you're on site again.
00:15:34.799 -->
00:15:38.159That that I think that's the biggest operability as well.
00:15:38.320 -->
00:15:55.360So if you come to site and say, for example, a lot of the plant isn't running that day, or there's an issue, or there's a problem, or there's a batch process issue, so you you miss half of the critical assets, you'll really define on the randomness of the process on that day, whether you can collect data or not.
00:15:55.519 -->
00:15:58.399And I do think that is a bit of a con when it comes down to it.
00:15:58.799 -->
00:16:07.679And but I think you can also flip that into a pro, which is if you do have variability and you have an engineer on site, they can work with the operations team like, oh, I need to test this at the end of the day.
00:16:07.840 -->
00:16:08.480Can we get it running?
00:16:08.720 -->
00:16:09.519Can we get it running?
00:16:09.679 -->
00:16:10.559Can we do this?
00:16:10.720 -->
00:16:16.799Um, I mean, I was at a site in uh Germany and they have these machines that go through different processes.
00:16:16.960 -->
00:16:23.519Well, I was able to plan my day around the fact that I need to test it at that point in the sequence.
00:16:23.600 -->
00:16:24.639I need to test it at that point in the sea.
00:16:24.879 -->
00:16:28.000Yeah, I agree, and I think if it's planned, it can be highly effective.
00:16:28.080 -->
00:16:44.559Yeah, but also another drawback is when you do plan it and a customer hasn't maybe planned it in or whatever, you can come to site and uh no, my my my daily collection day is kind of ruined because of the so with planning, it can be highly effective.
00:16:44.799 -->
00:16:47.039And in general, this is I think why.
00:16:47.120 -->
00:16:48.559I mean, there's lots of pros there.
00:16:48.720 -->
00:16:51.440This is why generally I'm struggling to think of some cons.
00:16:51.759 -->
00:16:56.399I know, and we're trying here because this is generally why I'll tell you what a good con is actually.
00:16:56.639 -->
00:16:59.519For for us, uh you know, it's our core business, right?
00:16:59.679 -->
00:17:03.120But we're doing a lot more now in building capabilities for teams so they can understand it.
00:17:03.360 -->
00:17:06.480We're evolving, we're gonna cut we're gonna come onto that a little bit later.
00:17:06.799 -->
00:17:13.519But it's a generally speaking, for if we're talking about analysis at the machine, it's a high cost to entry.
00:17:13.680 -->
00:17:13.839Yeah.
00:17:14.000 -->
00:17:15.200Data collectors are expensive.
00:17:15.519 -->
00:17:16.319They are 100%.
00:17:16.720 -->
00:17:21.279So for service providers like us, this is why we have to charge the day rate to cover the cost.
00:17:21.599 -->
00:17:33.759But you're right, I mean, a high grade analyzer is still probably 10 grand plus depending on what manufacturer you use, and even more some depending on the higher ends of the manufacturers you use.
00:17:34.000 -->
00:17:43.519So, yeah, you know, doing that essentially, even yourself as a site, there's still an outlay of cost to be able to before you get any value back to be able to go into there straight away as well.
00:17:43.839 -->
00:17:45.119I say it's a high cost to entry.
00:17:45.200 -->
00:17:49.200I mean, we're gonna come onto wireless in a moment, and we're gonna come on to continuous monitoring.
00:17:49.440 -->
00:17:52.240It's actually compared to those two, very cost effective.
00:17:52.480 -->
00:17:53.519It is, yeah, 100%.
00:17:54.079 -->
00:18:02.799So this is why predominantly handheld data collection is still widely adopted now because the cost of doing it in terms of value of what you get back is very high.
00:18:03.039 -->
00:18:03.759Do you know what I mean?
00:18:04.000 -->
00:18:07.279A lot of sites that do run very often, do you know what I mean?
00:18:07.359 -->
00:18:09.519You know, you get the value out of monthly data.
00:18:09.759 -->
00:18:15.039Generally, wear-related issues within every single month, which is generally kind of what we do.
00:18:15.200 -->
00:18:18.799I'll be honest, 95% of the time you're gonna win.
00:18:18.880 -->
00:18:21.599Like we've been doing this a very long time now, do you know what I mean?
00:18:21.680 -->
00:18:28.960With lots of our customers, and you know, very rarely you get a failure that's wear-related that you don't see between a monthly window.
00:18:29.039 -->
00:18:34.960Um, a lot of the time when the failures do happen, they're generally random, they're not really uh a wear-related event, you know.
00:18:35.200 -->
00:18:41.759Um, but there is there is a risk there as well with them where related in between them did them surveys, which we'll get on to some of the points now.
00:18:41.839 -->
00:18:42.960But I think we've kind of wrapped up.
00:18:43.200 -->
00:18:44.319I think that's portable data collection.
00:18:44.480 -->
00:18:45.599Yeah, so wireless.
00:18:45.839 -->
00:18:47.359Wireless, okay, let's go into wireless.
00:18:47.519 -->
00:18:51.200So pros of wireless uh data collection, ease of install.
00:18:51.359 -->
00:18:53.759Yeah, ease of install it's gotta be number one.
00:18:53.839 -->
00:18:57.119Like, I think this is where everyone loves a wireless device as well.
00:18:57.200 -->
00:19:01.759Like, you know, when you get your wireless headphones, there's no lead, I don't have to plug it in, I'm free.
00:19:02.000 -->
00:19:14.240So I think I think we have been conditioned a little bit with a lot of the wireless kind of other technology features that we've kind of benefited from personally in the in the past, and I think that's why it's quite nostalgic.
00:19:14.319 -->
00:19:20.960Oh, wireless sensor, we put it on, we connect, there's no leads, there's no problems to have, and yeah, this is great.
00:19:21.119 -->
00:19:35.839But there's a huge con to that whole element of oh, it's really easy to install because with wireless, you're guaranteed to have some form of connectivity issues, just like your wireless devices of like, whoa, my wireless headphones connect for God's sake.
00:19:36.079 -->
00:19:43.519So, with the big pro of the ease of use of it, there is always a connectivity issues as well.
00:19:43.599 -->
00:19:50.079But again, I think we've found that some manufacturers you have less problems, some maybe, some you have a lot more problems with as well.
00:19:50.240 -->
00:19:54.319So it's also finding the right kind of supplier.
00:19:54.480 -->
00:19:59.440But also, what's really important is to understand how that wireless technology is being connected.
00:19:59.599 -->
00:20:02.559So is it is it a mesh network where the sensors talk to each other?
00:20:02.799 -->
00:20:17.200Is it a different network where a central gateway, every sensor's you know, independently talking to that gateway, depending on how your site is set up in terms of where you want to place these sensors, it's gonna be highly dependent on whether it's gonna be a feasible option for you or not.
00:20:17.359 -->
00:20:27.119Well, one thing I really dislike about the wireless industry is it's sold on the pretense that every single application, there's always a way around it, which in some cases, you know, there isn't.
00:20:27.599 -->
00:20:29.839We've had applications where that's just not gonna work.
00:20:30.000 -->
00:20:39.119That's not gonna be sustainable for the client to be able to get reliable data because we also have to remember if we're relying on this information coming in, right?
00:20:39.519 -->
00:20:40.880It needs to be reliable.
00:20:41.440 -->
00:20:42.079To be reliable.
00:20:42.160 -->
00:20:45.039We cannot be our sensors disconnected again.
00:20:45.119 -->
00:20:58.799Yeah, we've missed a huge portion of the area of time where that would have been so valuable to capture it, or something fails because it's not managed to be able to ping the data across, or it's poor at communicating, there's a problem with it.
00:20:58.960 -->
00:21:07.200We've had a lot of systems where there's no alarm to even identify when the system's not running, so it's like, oh, everything seems to be okay.
00:21:07.759 -->
00:21:09.599Suddenly it's all just dropped out, and nobody knows anything.
00:21:10.160 -->
00:21:14.880So we don't have any indication of exactly when it's dropped out, and it is little problems like that.
00:21:14.960 -->
00:21:21.359It's like it's such an important thing to have, and you know, this this is where the cons can really start to build up as well.
00:21:21.680 -->
00:21:24.960We'll try and not make it too too like we don't like wireless, but they're not gonna be able to do it.
00:21:25.200 -->
00:21:39.680No, I think there's a place for them, but again, I think the problem is when you're really looking at strategy and you're really looking at the failure modes, and you're looking at how do I protect this asset in the best possible way, then you look at effectiveness in terms of how we actually apply it.
00:21:39.759 -->
00:21:41.839And this is going to be very evident in this study.
00:21:42.000 -->
00:21:52.960So this is almost like it its own little case study to identify to everyone of where these fit within the industry's industry, and also how how does mine and will's thought process actually go through?
00:21:53.039 -->
00:22:00.640Because when we're speaking to customers, we're not just trying to sell them something, we're not just trying to look, we're service providers, we're we're We're different.
00:22:00.720 -->
00:22:05.119We're not here to solve the solution providers that I love that we are.
00:22:05.279 -->
00:22:08.319We're here to be asked to say, Well, what's the best solution for this customer?
00:22:08.480 -->
00:22:11.599We don't want the problems, we don't want the customer saying it's not working.
00:22:13.920 -->
00:22:22.160So we're we are also thinking about what is the best way that we can provide a solution to this with the least amount of impact and problem to our operations.
00:22:22.319 -->
00:22:22.960Does that make sense?
00:22:23.200 -->
00:22:28.480I mean, like let's think of as well with with wireless sensors, what are some that we you know triaxial?
00:22:28.720 -->
00:22:38.640Okay, I ri understand though that it's triaxial at the point where the sensor is, so it's not necessarily true vertical and true axial, but it does give you three axes of vibration.
00:22:38.799 -->
00:22:39.599And it can be handy, yeah.
00:22:39.759 -->
00:22:40.640It can be handy.
00:22:40.799 -->
00:22:42.799Most of them come with a temperature output.
00:22:43.039 -->
00:22:43.759That's really cool.
00:22:44.000 -->
00:22:44.400Really handy.
00:22:44.480 -->
00:22:46.400Yeah, cool little metric to have, you know.
00:22:46.799 -->
00:22:53.599Um, like you say, ease of install is really handy with them, particularly harsh environments or where it could be very expensive to run cabling.
00:22:54.000 -->
00:22:56.160Yep, 100% in in accessibility.
00:22:56.480 -->
00:23:05.200So if you've got an actual fan that is up and up in the voids that you can't access, and running a cable run to it is just like, well, where am I going to run a cable to?
00:23:05.279 -->
00:23:08.880Like I've got 10 miles of cable, you know, and it runs all the time.
00:23:09.119 -->
00:23:13.200That for me is a perfect use case because two sensors can solve a great solution.
00:23:13.519 -->
00:23:16.720And even if it takes a reading a day, you've got some good trend there, haven't you?
00:23:16.799 -->
00:23:17.279Do you know what I mean?
00:23:17.359 -->
00:23:24.960Let's be honest, it's a lot better than having one reading, shall we say, over say, for example, and here's another pro of a wireless data collection.
00:23:25.119 -->
00:23:29.279You do get more information, you do get a bigger picture of the trend, right?
00:23:29.519 -->
00:23:35.440And you might be able to see a little bit more of an earlier indication of a defect initiating, which is which is handy.
00:23:35.680 -->
00:23:39.759We know we have things happen in between our 30 days, a lot can change in that period of time.
00:23:40.400 -->
00:23:45.599And some of the info could really give it indication of a of a change in parameter that a customer would want to know about.
00:23:45.759 -->
00:23:46.400Do you know what I mean?
00:23:46.480 -->
00:23:48.720So that there's there's definitely a probe with that.
00:23:48.960 -->
00:23:59.039And you know, if you can take you know, 30 data sets within a month, you know, you've already taken more data essentially over 12-month period, if that makes sense as well.
00:23:59.200 -->
00:24:05.200So I think the trends, what you can pick out from it, and we've got a few wireless sensors at a couple of our customers as well.
00:24:05.359 -->
00:24:06.640It's really nice to see the trends.
00:24:06.880 -->
00:24:09.119It's nice to have a bit of a gas fill between definitely.
00:24:09.440 -->
00:24:10.799I totally agree with that, you know what I mean?
00:24:11.039 -->
00:24:12.000Right, cons then.
00:24:12.079 -->
00:24:12.640Here we go.
00:24:13.119 -->
00:24:13.839Strabbin.
00:24:14.559 -->
00:24:16.319Um frequency range.
00:24:16.559 -->
00:24:20.559Yes, frequency response on tends to be limited compared to continuous monitoring.
00:24:21.119 -->
00:24:27.359Don't get me wrong, there are sensors out there that have got a high frequency range, but then you're really looking into high costs.
00:24:27.680 -->
00:24:32.400But then there's another con because because you have a high frequency range, it drains the battery.
00:24:32.799 -->
00:24:37.359So now you're only limited to maybe a read-in per day, for example.
00:24:37.599 -->
00:24:37.839Yep.
00:24:38.079 -->
00:24:44.960And you know, if you want more of a continuous kind of monitoring system that looks at process parameters as well, you lose that by doing that.
00:24:45.039 -->
00:24:50.400That then now it's not really considered a process kind of monitoring point, if that makes sense as well.
00:24:50.720 -->
00:25:01.599I think the other main con, so for people to understand, is is that the majority of the wireless sensors, I'm gonna say, are dumb in their how they work.
00:25:02.000 -->
00:25:10.720So they are not smart enough to understand what the machine's doing, what it's doing at that point in time, what else is happening within the environment.
00:25:10.880 -->
00:25:21.039They simply sit dormant on the machine and wake up at a set predefined interval, and then they decide whether or not they're gonna gather that data or not.
00:25:21.200 -->
00:25:26.799Yeah, if the machine's not running when it does that, it's not good, it's collecting the data with the machine off.
00:25:27.119 -->
00:25:33.839If the machine is working harder at that moment and the trend is higher because it's under more load, yeah, it knows no different.
00:25:34.079 -->
00:25:36.079It is just collecting data, that's what it's doing.
00:25:36.160 -->
00:26:01.920It's not learning, shall we say, off the parameters or any process input parameters that we've got, which we can do with continuous, which we'll get onto in a bit, and try not to um overlap here with some of the things, but it's more comparisons, and and this is where they're very different systems, and this is where I think it's really important for us to explain that if you are got a wireless system and you've got something that's continuously powered, they are different systems in how they operate completely.
00:26:02.079 -->
00:26:19.759So, this is why when we do strategies and we are seeing you know assets that have different variants of load and variants of speeds and stuff, wireless sensors has always been a very difficult system to track that and understand exactly when the loads are the same and when they're not.
00:26:19.920 -->
00:26:20.319That makes sense.
00:26:20.559 -->
00:26:35.839Yeah, and even some sensors on the market do have the triggerability function, but most of these sensors are just communicating at a regular interval, and at the time of interval, if it matches the trigger definition, then it decides to store the data.
00:26:36.160 -->
00:26:39.920And that's what I think that's where the technology is trying to go with the wireless sensor options.
00:26:40.160 -->
00:26:43.759This is why we're not seeing PZ electrics within you know sensor setups anymore.
00:26:43.839 -->
00:26:45.680We're seeing MEMS-based sensors as a proof.
00:26:46.319 -->
00:26:47.119The battery life is too much.
00:26:47.519 -->
00:26:48.960It's battery, it's battery intensive.
00:26:49.119 -->
00:26:56.880So I think the idea is eventually, and I know there's sensors out on the market that are trying to, you know, be able to be mag flux and harvest and be able to take that.
00:26:57.039 -->
00:27:04.640I do think I see a place eventually when we can have that working very well, but we've right now, currently, they're not at the place.
00:27:05.200 -->
00:27:09.519Ties in perfectly the fact that you're saying, Oh, we can see that in the future.
00:27:09.920 -->
00:27:15.839This hardware and even software obsolescence happens incredibly quickly.
00:27:16.160 -->
00:27:19.039So that takes me to my next con obsolescence.
00:27:19.440 -->
00:27:28.240So if you have got and invested in a uh a system that you know maybe was designed two years ago, technology moves so rapidly fast.
00:27:28.400 -->
00:27:44.880And if you have installed thousands of sensors on your plant and spend that money, I could probably guarantee two, three years that there's gonna be a better alliteration or modification of that sensor, which happens within all technology, and you're probably gonna have to potentially buy the hardware again.
00:27:45.119 -->
00:27:52.400So potentially, especially for some sensors that are don't have battery replaceable functions, so they're throwaway.
00:27:52.960 -->
00:28:09.440Some sensors do need to have their battery replaced, and we know that we've got uh working with one provider who is asking our support because they've got well over a hundred sensors, and all the batteries need replacing, and the site themselves do not have the time to go round to all these sensors.
00:28:10.240 -->
00:28:17.519We rarely sell people just sensors, we will not do it because we sell them the sensors and then they'll go obsolete.
00:28:17.680 -->
00:28:25.759Like generally, if we're gonna do the model, it's an OPEX model, we'll put them on, we'll we'll be the experts, we'll analyze, and we'll sort the rest out.
00:28:26.079 -->
00:28:29.279Because the other option is not sustainable for the customer.
00:28:29.440 -->
00:28:30.079Do you know what I mean?
00:28:30.160 -->
00:28:36.640It makes no sense to be able to write off hundreds of thousands of pounds of hardware because you've had updates and upgrades.
00:28:36.960 -->
00:28:38.880Does that make that makes no sense financially?
00:28:39.119 -->
00:28:45.920The thing is, is for the companies developing the sensors, you know, they'll still support them, the data's still gonna come in, but it's then oh, I've got the old version now.
00:28:46.000 -->
00:28:50.559Yeah, but not even that, then you're restricted to new features that are gonna add you value in the future.
00:28:50.799 -->
00:28:51.599Does that make sense?
00:28:51.680 -->
00:28:53.839Yeah, and and that's that is the problem with this.
00:28:53.920 -->
00:28:54.640It's like an investment.
00:28:54.720 -->
00:28:58.079It's like if I buy a car, do you know I want to want the latest model, right?
00:28:58.559 -->
00:29:06.319How annoying is it when you buy a model or you buy something or you buy the latest Apple AirPods, and then next two months it's like Apple AirPods thrive with that.
00:29:06.400 -->
00:29:07.759And you're like, ah, damn it.
00:29:08.319 -->
00:29:16.079We had one where we had version one of the sensors that a customer had installed where they actually were sold by another provider and asked us to come in and support them.
00:29:16.240 -->
00:29:19.039Version one sensors, version one gateways.
00:29:19.119 -->
00:29:26.319They the company had bought out version two sensors, but version two sensors couldn't communicate with version one gateways, they could only communicate with version two.
00:29:26.799 -->
00:29:27.440It's a nightmare.
00:29:27.680 -->
00:29:29.599It was absolute nightmare.
00:29:29.759 -->
00:29:33.200And I can see what the obviously you send the manufacturers can innovate, aren't they?
00:29:33.359 -->
00:29:33.839Of course they are.
00:29:34.000 -->
00:29:37.200Everyone needs to, but again, what is the impact of innovation?
00:29:37.279 -->
00:29:38.319We've got to also have a look at that.
00:29:38.799 -->
00:29:41.680We're not at a stable point when it comes down to wireless sensors.
00:29:41.759 -->
00:29:50.319When we come on to continuous monitoring in a bit and the and even handheld data collection, standard is a hundred millivolt per G accelerometer, which has been the standard as well.
00:29:50.559 -->
00:29:52.79915, 20 years, even longer than that, probably.
00:29:52.960 -->
00:29:54.240And that's where we'll move to.
00:29:55.039 -->
00:29:56.000Continuous monitoring.
00:29:56.079 -->
00:29:58.799And I think we'll start at that pro that point there.
00:29:58.960 -->
00:30:10.000You know, piezoelectric 100 millivolt per G sensors have been the go-to for vibration analysis for years, decades, decades, and now you know, we can even make them small.
00:30:10.079 -->
00:30:16.720And you look at the small little piezoelectric sensors we can make now, there's tiny little capsules that that Chris does at you know, Hansford.
00:30:16.880 -->
00:30:21.519Even if we need to have a really small form of that particular sensor, we can do it.
00:30:21.680 -->
00:30:26.880But generally, it you know, in general, the sensor size is fairly small, it's very low profile.
00:30:27.039 -->
00:30:32.240So, this is another big pro with the sensors that you can get a very good frequency response from the sensor.
00:30:32.559 -->
00:30:36.880Some of these wireless sensors, when they first were made, were disgustingly big.
00:30:36.960 -->
00:30:40.559Yeah, I mean, you look at the resonance curves of some of these sensors that were first made.
00:30:40.720 -->
00:30:41.359Do you know what I mean?
00:30:41.519 -->
00:30:46.400You know, remember the big square block ones that that used to hang off machines and stuff like that.
00:30:46.559 -->
00:30:53.039I'm not naming any names, but you you look at that model, it's like, oh, because you need a huge wireless transmitter, right?
00:30:53.200 -->
00:30:59.279Whereas when you get a nice piezoelectric sensor, you can put that sensor in the in the loaded zone very nicely.
00:30:59.440 -->
00:31:06.880So mechanical transmission with the high frequency response of that sensor allows us to get the highest quality data.
00:31:07.279 -->
00:31:09.920We didn't mention with wireless sensors, but it ties in nicely.
00:31:10.079 -->
00:31:13.680Wireless sensors do have a big limitation when it comes down to slow speed machines.
00:31:13.839 -->
00:31:14.319Yes, 100%.
00:31:14.720 -->
00:31:15.359Huge limitation.
00:31:15.519 -->
00:31:26.400Now, when it comes down to continuous monitoring or even handheld data collection, we have the ability to, because the sensors are hardwired, we can gather very long-time waveforms if we need to.
00:31:26.720 -->
00:31:30.400But also we can say put a higher sensitivity sensor on it.
00:31:30.559 -->
00:31:31.039Yeah, of course.
00:31:31.119 -->
00:31:39.759We can go say if we want a different sensitivity depending on what we're monitoring, particularly if we're monitoring something very, very slow, two, three RPM, we might have amplification.
00:31:41.519 -->
00:31:45.680Or if we're measuring something with very high shock loading, a big crusher.
00:31:45.839 -->
00:31:47.119And you're like, right, I need to reduce the sensitivity.
00:31:47.279 -->
00:31:48.400I need to reduce the sensitivity.
00:31:48.640 -->
00:31:48.880100%.
00:31:49.119 -->
00:31:52.160So you have that option depending on what you're monitoring, which I love.
00:31:52.319 -->
00:32:01.839And another great thing about continuous monitoring, this is probably for me the number one reason why I would suggest them, it's the ability to be able to monitor practically anything.
00:32:02.079 -->
00:32:14.559Because you're constantly monitoring a parameter or something, or you're measuring, and you can have lots of different outputs and digital outputs, and you can measure different process parameters, vibration, temperature, current.
00:32:14.720 -->
00:32:24.240You can trend all these things, but even more importantly, you are able to stabilize the data or stabilize the machine running conditions at certain points.
00:32:24.640 -->
00:32:30.720That allows us to trend very tricky machinery like CNC machines that we're doing now in terms of going into CBM.
00:32:31.039 -->
00:32:32.720And even refrigeration compressors, right?
00:32:32.799 -->
00:32:35.599So we do a lot for a company, they've got a lot of ammonia compressors.
00:32:35.680 -->
00:32:37.680We're now coming towards the summer period.
00:32:37.839 -->
00:32:42.480We'll see all in our handheld data, we'll see all the alarms start to increase as we get to summer.
00:32:42.799 -->
00:32:45.920Because it's more low, harder, they're trying to cool the freezers a lot more.
00:32:46.000 -->
00:32:48.559Yeah, and then as soon as we get to winter, we start to see them drop off again.
00:32:48.720 -->
00:32:49.039Exactly.
00:32:49.119 -->
00:32:55.039And what we can do with that is do very smart alarm in to be able to say high and low alarms to understand exactly where we're at.
00:32:55.279 -->
00:33:01.440Because at the minute, for the engineers, from their perspective, they have that that we can't just like, oh, we're bumping them up because it's summertime, oh, we're dropping them down.
00:33:01.680 -->
00:33:04.960Whereas actually within some of the continuous monitoring systems, we can do that.
00:33:05.119 -->
00:33:05.759Yeah, we can see that.
00:33:06.079 -->
00:33:22.720We can set an outside ambient temperature, we can adjust the alarms with the outside ambient temperature because we know the machines are working harder or definitely, and we've got that comparative data now to be able to make their models, and that is another good thing about wireless, no, not wireless, wired data collector continuous.
00:33:22.880 -->
00:33:25.440There's no limitations on how much data you collect.
00:33:26.160 -->
00:33:32.160The only limitation is how much data you store, but if that's valuable, let's be honest, data storage now is cheap.
00:33:32.559 -->
00:33:35.759It is like back in the day, oh it wasn't so much cheap, was it?
00:33:35.920 -->
00:33:42.880Remember we used to buy 512 megabyte um bloody like Walkman, like little the little tiny MP3 pack.
00:33:43.039 -->
00:33:48.880I remember I remember being in school and it was like, oh, do I want the 512 or do I want to 128 megabytes?
00:33:50.240 -->
00:33:54.480A four gigabyte USB stick now was like 30 quid, now it's like two pounds.
00:33:54.720 -->
00:33:55.599It's ridiculously cheap.
00:33:55.839 -->
00:33:59.119Yeah, like 100 gig USB sticks and all sorts of stuff.
00:33:59.519 -->
00:34:01.200It's it's crazy now with storage.
00:34:01.279 -->
00:34:18.000Like again, technology revolution allows us, and it's not just one thing when it comes down to kind of like technology, you need wireless technology to be better, you need connectivity to be better, you need data storage to be able to be better, and and and now half of it's held in the cloud anyway.
00:34:18.079 -->
00:34:28.159Yeah, so like again, the the technology advancements have allowed us now to be able to actually have a price point where continuous monitoring is now not expensive to do.
00:34:28.480 -->
00:34:32.719The the biggest con with continuous monitoring is you have to cable it in.
00:34:32.880 -->
00:34:38.639It is, it is the install, but my argument of that is if it's critical enough, why are we not doing it?
00:34:38.880 -->
00:34:52.079But not also that like the I've said to nowadays as well, when we're looking at going forward the digital plants and trying to digitalise what we're doing, no one's going, hey, should we um put all our pressure transmitters on wireless?
00:34:52.639 -->
00:34:53.440No, because it's not really.
00:34:53.760 -->
00:34:54.639Do you think they would do that?
00:34:54.800 -->
00:34:58.960No, they wouldn't, and this is this is this is also a crazy thing, as well.
00:34:59.280 -->
00:35:07.280Now, the adoption of vibration has become better, but it's still poor because half the systems that we see now haven't got sensors on.
00:35:08.400 -->
00:35:09.840You know, oh can't access that.
00:35:10.000 -->
00:35:13.599Oh, we've got a pressure sensor, but we haven't got a vibration sensor, we're not going to measure that.
00:35:13.760 -->
00:35:17.119You know, I went to a brand new site the other day when we're doing some work in the Netherlands.
00:35:17.199 -->
00:35:23.360You know, all everything's got all of these other sensors on, no vibration sensors whatsoever on assets that high in the system.
00:35:23.599 -->
00:35:28.159Which we know as well at the design stage, biggest influence, lowest cost to install.
00:35:28.639 -->
00:35:32.960It's tiny, like we, you know, even piezoelectric sensors now, the cost has come down.
00:35:33.039 -->
00:35:34.880Yeah, it's not the same as it used to be.
00:35:35.280 -->
00:35:50.239We're fine, we're finding now we're doing some pricing at the moment for a customer as a comparison, and the price difference between a wireless sensor install versus a hardwired install is a lot closer than it ever used to be.
00:35:50.559 -->
00:35:51.280Very, very close.
00:35:51.599 -->
00:35:52.159Oh, 100%.
00:35:52.559 -->
00:36:10.480And I'd also say is like now with that high frequency sensor in the right loaded zone, you know it's in the money, you've got constant data collection, you can see the data live if you want to, you can look at it and actually have a look at it and see what the response of the machine is doing.
00:36:11.039 -->
00:36:24.079Once you've got a continuous monitoring system set up in that way, and you've got it set up in all the things you want to look at, all the band alarms and all, and you've got a real technical company really looking at the failing modes, there's nothing really you miss.
00:36:24.559 -->
00:36:47.039And I think that is if you can if your if your company for your most critical assets be in a position where you're constantly monitoring every parameter that could potentially cause you to fail, and you're in that position, you are above and ahead of 90% of the people in your field and your competitors within within what you're doing with maintenance.
00:36:47.280 -->
00:36:56.719Because not, I'll be honest, there's still a huge amount of percentage of factories not even doing handheld data collection, let alone monitoring that.
00:36:57.039 -->
00:37:05.199And this is like overall, you know, we've talked about the pros and cons of handheld, we talked about the pros and cons of wireless, we're talking about the pros and cons to continuous monitoring.
00:37:05.360 -->
00:37:07.920All of these just are methods of acquisition.
00:37:08.159 -->
00:37:13.840Typically, you're gonna have maybe us or a provider out there that is analysing that data.
00:37:14.000 -->
00:37:25.440Maybe they're collecting the data for you in handheld form, but what we're seeing as well, more and more, is and I'm a big believer, and I know you are as well, in that there's not enough of me and you and our guys.
00:37:25.679 -->
00:37:27.519100% and there's no way near enough.
00:37:27.760 -->
00:37:40.239And what we're trying to do, and I think is really more important than ever, is that companies have some level of internal knowledge and capability to have some basic interpretation of that data.
00:37:40.559 -->
00:37:41.119Yeah, 100%.
00:37:41.360 -->
00:37:43.440And I think that that is that is doable.
00:37:43.679 -->
00:37:53.840So, you know, going on to kind of the customer collecting the data on the handheld point of view, you don't need to be a vibration level three analyst expert to collect good, meaningful data.
00:37:54.079 -->
00:37:57.280And that that's where you differentiate between data collector and analyst.
00:37:57.599 -->
00:37:58.000And analyst.
00:37:58.239 -->
00:38:04.320So there is a difference, and we have to make sure that you know we understand that difference if that is the model that we're going for.
00:38:04.400 -->
00:38:12.800We we have customers that go collect their data and send it to us as analysts, but we also do awareness training as well about mechanical transmission paths.
00:38:12.960 -->
00:38:16.880Where repeatability is so important to be able to make sure we're taking the data.
00:38:17.119 -->
00:38:20.880Putting disks on the machines of things, here's where we're collecting that and here's why.
00:38:21.039 -->
00:38:23.760And also giving training to the guys that are taking the data.
00:38:23.840 -->
00:38:27.840Are you collecting that data when the machine's in the in a nice constant road of the data?
00:38:28.159 -->
00:38:34.719And that probably is a really good uh con to portable data collection is if it's not set up correctly, it can go wrong.
00:38:34.960 -->
00:38:35.519It can go wrong.
00:38:35.679 -->
00:38:37.039I'm testing in the wrong point.
00:38:37.119 -->
00:38:38.239I'm not testing in the load zone.
00:38:38.800 -->
00:38:46.239We've seen sensors that being installed completely in the wrong load of zone positions, and we're not quite on the money, and we're not actually seeing the issues and problems.
00:38:46.400 -->
00:38:51.199And this is also it goes back to vibration 101, the basics of mechanical transmission path.
00:38:51.280 -->
00:38:53.039The laws of physics do not change.
00:38:53.199 -->
00:38:55.039The laws of physics don't change.
00:38:55.440 -->
00:38:59.519We know that, and we have to make sure that that is always the basis of where we grow from.
00:38:59.840 -->
00:39:12.400I think that going back to like we're doing a lot of mentorship plans at the moment, we've got customers where we're going in, we're teaching them how to collect the data with the long-term plan to teach them how to analyse the data, and then have us there always for support and help, right?
00:39:12.639 -->
00:39:23.280But I do think it's really important, no matter whether or not you've got portable data collection and you're doing it yourself, or or we're doing it, or some companies do it, or you've got wireless sensors, or you've got continuous monitoring.
00:39:23.440 -->
00:39:50.239Any company that is looking at outsourcing condition monitoring or vibration analysis or utilising wireless sensors or continuous monitoring, someone within that organization needs to have some basic knowledge and understanding of vibration analysis because even if you're looking at a wireless system on the market that does AI and tells you what to do, it's a very dangerous place to be in to not understand how it's getting to that determination.
00:39:50.559 -->
00:39:51.360I totally agree.
00:39:51.519 -->
00:39:58.159It's very dangerous to rely on external information if you do not understand exactly how them decisions are come to it.
00:39:59.840 -->
00:40:09.119If an AI model is telling you to make a decision on a highly critical piece of equipment, and you don't understand how it's come to that conclusion, that is very dangerous.
00:40:09.440 -->
00:40:19.519There was a podcast I was listening to, really interesting, by a guy that had um developed a lot of the training material in AI, and he basically said, if you can imagine like Chat GPT, we'll use that as an example.
00:40:19.760 -->
00:40:23.920We put an input in, and at the end, the output comes out of it.
00:40:24.000 -->
00:40:31.840But in between that is like a chain link fence, and that AI model is working its way along that chain link fence to get to the antenna.
00:40:32.320 -->
00:40:35.599We don't actually know how or what route it took.
00:40:35.840 -->
00:40:39.679No, no, even within AI, we don't know the route it's taken to get to that.
00:40:39.840 -->
00:40:48.159So obviously, the models that are being built around AI, obviously, there's some guidance that's principles that are given to it, and it's taking that route and it's spitting something out at the end of it.
00:40:48.400 -->
00:41:02.079But if you don't understand the basic principles of what it's looking at, or even even if we're analyzing the data, the more our customers can understand what we're talking about when we're putting our reports, and obviously, we have to be very conscious not to make them too tested.
00:41:03.440 -->
00:41:04.239That's the value.
00:41:04.400 -->
00:41:10.719We're translating something that they don't understand, so they can understand what's going on with the machine, so they can make better decisions.
00:41:10.880 -->
00:41:20.320And you are right, it is our job to try to bring them on the journey, and even if we give them some basic knowledge, they understand why they're making the decisions.
00:41:20.800 -->
00:41:23.840And we can see the value even more and more now as well.
00:41:23.920 -->
00:41:30.400We we've got a lot of customers now that are transitioning and being a lot more strategic with their vibration analysis strategy.
00:41:30.639 -->
00:41:34.880So with the really highly critical assets, we're moving on to continuous monitoring.
00:41:35.039 -->
00:41:40.239We don't see a lot of wireless sensors because the price point now is like, why would I go wireless when I can go continuous?
00:41:40.480 -->
00:41:43.599That but that that is on also another huge thing.
00:41:43.760 -->
00:41:48.239The continuous model is becoming now, it's it's doing this.
00:41:48.320 -->
00:41:48.559Yep.
00:41:48.800 -->
00:41:51.039So now it's like, well, where do I spend my money?
00:41:51.119 -->
00:41:58.719Well, do I spend it on sustainability or do I spend it on something that could potentially be outsourced or potentially could be superseded?
00:41:59.440 -->
00:41:59.679Exactly.
00:41:59.920 -->
00:42:01.119Like we don't want that, do we?
00:42:01.280 -->
00:42:06.639Like if if we're buying a car or psychological, you don't want to go buy now.
00:42:06.719 -->
00:42:09.199It's also people are like, do I buy a combustion car anymore?
00:42:09.360 -->
00:42:11.920Because is there going to be, you know, people are want what the feature is.
00:42:12.480 -->
00:42:13.119We just looked at it now.
00:42:13.199 -->
00:42:17.840We've got this customer that they're looking at moving all of their critical compressors onto continuous monitoring.
00:42:18.000 -->
00:42:22.800They've got some assets that are a little bit far away, and we were doing the working out the cost going, do you know what?
00:42:22.960 -->
00:42:33.920You know, it's about 30 metres, and we're like, well, okay, we could put them on wireless sensors because of the distance, but we've got channels available within the the system that we're monitoring, figure out how much core cable it is.
00:42:35.039 -->
00:42:36.079We'll just run the cable.
00:42:36.480 -->
00:42:36.719Yeah.
00:42:37.679 -->
00:42:40.960To have the ability to have like live data, like, oh, well, I can lock in and I can.
00:42:41.199 -->
00:42:46.960And I get it, labour cost can be an issue, but because we're a company that has a mechanical team that can do that easily.
00:42:48.079 -->
00:42:48.559We just do it.
00:42:48.719 -->
00:42:52.880So, like, again, it obviously depends on obviously your resource and what you're doing.
00:42:52.960 -->
00:43:00.079But when you're working with a company like us, we're always gonna pick, it might be a little bit more expensive, but it's gonna give you 10 times more value.
00:43:00.239 -->
00:43:08.159We're gonna always go with that option because we prefer go the extra mile to get that that sustainability out of that that option opposed to going the other way.
00:43:08.480 -->
00:43:14.000And we're seeing now is more, like we said, we looked at that that data collector versus the analyst, right?
00:43:14.239 -->
00:43:23.199The data collector or the person that's able to do the observations for us at the moment, sending level two or higher analysts can be very inefficient.
00:43:23.519 -->
00:43:26.079But your engineers know your plant better than we do.
00:43:26.239 -->
00:43:26.719Yeah, 100%.
00:43:27.360 -->
00:43:50.559They also understand how the machines operate, and we've got a lot of customers now, and we're doing a lot of mentorship and training around empowering them, not necessarily to be full-blown analysts, because it's not their day job, but if they have the capability to go to a data collector that is not maybe as expensive as a full blown singing and dancing analyzer is, they can go out and gather the data, they can feed it into us to be able to do that.
00:43:50.639 -->
00:43:56.800And we can be the experts, and we can be the experts, and that that that mate, honestly, what a great conversation.
00:43:56.880 -->
00:43:59.760So, guys, if you're looking to weigh all these options up.
00:44:00.159 -->
00:44:03.440And um the thing is with us, we don't really suggest one way.
00:44:03.519 -->
00:44:07.599We we wanna we wanna really be able to give you guys the option of what's good for your plant.
00:44:07.760 -->
00:44:13.360Of course, we can make advice and give recommendations of what we want to do, but get in touch with us in terms of getting that ready.
00:44:13.440 -->
00:44:17.039But what an incredible podcast, great discussion, and I'm sure you're gonna see.
00:44:17.360 -->
00:44:18.719We had to cut it, otherwise we would have just kept it.
00:44:18.880 -->
00:44:22.719Yeah, we're we're we're gassing, we we never get to 45, like this is crazy on one subject.
00:44:23.039 -->
00:44:28.079But then I think it does really show as an endpoint that look, this is a complex subject.
00:44:28.239 -->
00:44:32.800And and if anyone, if anyone if anyone like we really enjoy talking about it.
00:44:32.880 -->
00:44:36.800So if anyone just wants to chat about it and and extract some expertise.
00:44:37.760 -->
00:44:39.039Yeah, let's do it.
00:44:39.119 -->
00:44:39.679Let's do it.
00:44:39.840 -->
00:44:42.880Anyway, guys, take care, have a great week.
