An Animated Introduction to Vibration Analysis Q&A - Mobius Institute

hello and welcome to this presentation well I could call it an animated introduction of vibration analysis but actually that's a webinar that I performed recently and we received so many questions afterwards I thought well let's create another video answering all those questions so I am Jason Tranter the founder and CEO of Mobius Institute and you can still see the original presentation if you'd like to watch that anyway let's go through some of the questions what is the best way to be trained and certified okay we ask that question ourselves but hey we are doing this kind of for a reason so the answer to that question is of course Mobius Institute we have training centers in over 50 countries all of our courses category 1 2 3 & 4 are taught in many languages you can go to a training class which trained over 4,000 people that way each year and then we have distance learning and we have eLearning one of our great strengths I guess what we're uniquely known for as we create a lots of animations and simulations which you would have seen in the original webinar and you'll see more of today and that just makes all this stuff so much easier to understand you know a certification follows the ISO standards and we are accredited our certification program is accredited and since we started running these sorts of courses since 2005 just in these category 1 through 4 courses with trained over 26,000 people it's just since 2005 anyway enough of that enough of that ok so the first question is what generally causes harmonics versus a single peak in a spectrum so if I have just a single pure source of vibration just at one frequency I just get one peak in the spectrum now if you can imagine a machine rotating if it was just smoothly operating like just purely unbalanced if we just have one frequency but a lot of the vibration we actually experience like the forces involved with the vibration and purely sinusoidal if you looked at it more closely it's actually distorted a little bit now I'm going to show you that in just a moment and that creates harmonics so instead of just seeing the peak in the spectrum related to the frequency you see Peaks at twice that frequency three times four times five times so that's one way we can see those Peaks but the more distorted it becomes the more harmonics we get and the higher the amplitude is of those harmonics and if we get to the point where we're actually experiencing impacts like for example once per revolution you know bang bang bang bang because of looseness let's say or within a bearing each time the rolling element hits the damaged area in the outer race we see an impact and we get harmonics as a result so here is a pure simple sine wave and because it's just one frequency and what's just wind that down a little bit because it's just one frequency it's a beautiful sine way there's just one frequency as a result but if I play with this signal and I actually distort it and you can kind of see I'm changing the shape of it a little bit I won't go into too much detail what I'm doing to it but you can see this is not a sine wave anymore and lo and behold we get harmonics okay there little harmonics but they are harmonics so if it was if there were stronger impacts we would see actual harmonics that are much higher in amplitude so what I thought I'd show well okay if it was the time waveform that looked like that so not purely sinusoidal we would see harmonics for sure now I'm going to use a little simulator that is there to demonstrate what happens when bearings are damaged now just bear with me for a second so what we're seeing over here oops better keep it on screen we're not seeing it all on screen anyway so there's the sine wave of just the shaft turning in the bearings are just fine but if I have a little bit of damage on the at erase here then each time the rolling element rolls over that area of damage we get this little spike now the way this simulator works it's just actually trying to say hey if it's just a tiny little spike you may not see anything in the spectrum but what I'm going to do is I'm gonna make the spike much worse and you see these harmonics come up in fact we can just isolate just the spikes and that's what's causing you know these Peaks now in this instance the frequency we have here is they're not generating harmonics of one X so this is my one expect this is my running speed peak and there are questions on that coming up so if you're not sure what I mean by One X you will in just a minute so there would be two x here and 3 X + 4 X and because it's a rolling on load bearing I'm actually seeing harmonics of 4.2 - I happen to know that so at four point two two and eight point four four and three times that and four times that and five times that we get peaks in the spectrum anyway that's probably enough said about harmonics but that's why so any time you see how monix in the spectrum you gotta think aha you know what is happening inside the machine is it just sort of slight distortion and there will be weak harmonics distortion because as it as a shaft rotates maybe it's more stiff vertically and horizontally so it gets a little bit of distortion like it's not a pure sine wave anyway if you see really strong harmonics you'll say to yourself okay what is impacting why does mechanical looseness generate multiple harmonics of one X vibration and really I kind of just answered that question mechanical looseness involves impacting either within the bearing itself due to excessive clearance that can be sort of within the clearance of the bearing or in between the outer race and the housing or even just the machine as it vibrates up and down we might have impact impact impact impact so there's rotating looseness mechanical looseness but anytime we have for example a once per revolution impact we will see 1x harmonics so we'll see peaks at 2x 3x 4x and so on so in this animation hopefully you can see you know as it rotates we see this this impacting going on and that could happen also in the you know between the outer race and the housing I mean that's if it's loose and of course my animations are really exaggerated and anyway when there's looseness we get that impacting now I've mentioned the word 1x a few times and this question is can you elaborate a little more on 1x 2x et cetera and and running speed what is that and how does it correlate to being able to determine where on the Machine the fault is located so number 1 1x simply means 1 times the running speed of the machine so let's have we've got a motor and it's turning at a particular speed in CPM you often quoted in rpm but a lot of people work in Hertz as well so it has a speed well because a lot of fault conditions occur as the shaft you know rotates once for revolution so for example unbalanced we get this this force and you'll see it in a moment you know once per revolution rather than quoting what it is in CPM or Hertz which is meaningless unless you happen to know the speed of the machine it's easier just to say 1 X 1 times the running speed 2 X 2 times the running speed and for most vibration analysts as soon as you say others a big peak at 1 X their mind goes off in one direction oh there's a you know the vibration at 2 X has increased their mind goes off in a different direction 3 X 4 X 5 X you know same thing we know what that means we don't need to know what the speed of the machine is but if someone said to us oh it's going up at two thousand nine hundred and something you'd say oh well I'm gonna assume that I know what the running speed is and you know it takes an extra step so that's the first part of the question second part is okay you know if we do have unbalance the spectrum might just have a big peak at 1x if for example those misalignment I could and I'm gonna explain this a little more in just a moment I could see a peak at 2x and maybe one at 3x and even went up for X and it depends actually on the on the coupling but you know looseness and it's what we mentioned just a motor or I mentioned you know there's my 1x peak but I also see one at 2 X 3 X 4 X 5 X 6 6 and so on and they are the harmonics and in this case I also see the noise floor that's the bottom of the spectrum it lifts up often when we have a lot of impacting going on it excites resonances in the machine and the vibration the general sort of noise if you like the general sort of background vibration also increases in amplitude so number one in terms of the 1x vibration we often see that because of unbalance in some situations misalignment if the shaft is bent if there's a centricity and really anything where once per revolution we got some sort of force so you know normally even with just a normal machine we will see vibration at once per revolution at 1x you you'll never balance machine so well that there is no vibration it may not be obvious in some cases which peak is the 1x and I've got a question about that coming up but if for example we have a fan that you can see there and because of unbalance then once per revolution we've got this centrifugal force that makes it wobble if I can use that term as you can see there it's actually wobbling sort of up and down it that that impeller is actually turning in a circular motion but it's actually rocking back and forward axially as well so we would actually see vibration both axially as in if we measured in the in this direction of the machine but also radially so vertically and horizontally we would see 1x vibration in this case this isn't the best animation in the world be there but if there's angular misalignment then once per revolution what I'm trying to show there is that the bolts are kind of being stretched at the top now this is obviously very exaggerated but just to sort of make the point that once we revolution we've got this force in the axial direction and we can see yeah hi 1x in that case now we can have actual forces that generate the 2x and 3x vibration but again we can also say well due to the misalignment or for example cocked bearing or and a bent shaft we can see vibration at twice running speed but there's always this debate about well is the machine doing something or creating a force twice per revolution or three times per revolution or is it just that it's generating something at once per revolution but because of the you know non-linearity you know it's not a simple rotating force that changes as it rotates we see peaks at 2x and 3x so in this case you can see a very bad case of offset or parallel or misalignment and we expect to see 2x vibration for example in this case and perhaps in 3x and 4x in this case you know if the if the inner race was bent on the shaft then each summit rotates it generates a 1x vibration but we often see peaks at the 2x and 3x as well and of course I need to say that will also see vibration at 2x and 3x and 4x for physical reasons so we could have for example machine with two lobes you know or or a coupling that has three jaws or four jaws and that's going to generate vibration as well so that kind of answer the question I wasn't sure exactly what you wanted to learn with this question but the fact is that with time waveform analysis and especially phase analysis it enables us to determine what the fault actually is because all of those things we just saw physical motions within the machine the bearing wobbling the shaft wobbling the fighting over the coupling do but offset an angular it was all physical and that makes the machine and the bearing housings respond in a certain way so if we compare the motion vertically in horizontally we can get a clearer idea without unbalanced or or something else and that can lead us to certain conclusions we can see how the phase changes from one side of a coupling to the other side and look there's a lot of ways it's just gives us timing information and that is the way to really properly solve you know why you're seeing vibration at 1 X 2 X 3 X and particularly those frequencies okay so the next question is these I've ordered these in a certain order so they're also relevant to each other all these analysis depend on the assumption that we have the running speed estimated correctly yes it really does you you really do need to know what the running speed of the machine is because how do you know if it's unbalanced if you don't know what the running speed is and therefore whether you're looking at the right peak and all those other things that I've said yes you need to know but you know if for some reason it's challenging to find out from the spectrum because often the 1x peak is quite high and it's it's the dominant peak at this or the low frequency end of the spectrum if you don't know there are tools that let you measure it there's a variety of tools in little stroboscope and there are other ways and then also even with the spectrum itself you can kind of reverse engineer it now make it additional comments about that later but when you're looking at a spectrum you can say well if this p equals 1x then i'd expect to see some other peaks at multiples if I've got you know fan blades or pan things or gear teeth or whatever yeah there's a certain way that kind of did use which is the 1x peak it's not perfect but once you have an idea of what it is it's then possible to make it accurate because for example if I knew I had six vanes I can go to the 6x peak figure out what the frequency is and then divide that by 6 to get a much better estimative run Expedia if if I need it okay so how can we ensure that we have the correct running speed estimated so it was just another question and you know as I guess kind of mentioned ideally we measure it and we know for sure but you know if I knew there were six pump veins or 31 gear teeth or something like that I can look for those Peaks and then say well okay well if that's your mesh frequency at a peak that's 31 times running speed I therefore know what the running speed is there's a lot more I could say about that but let's move on oh here's a good question what is the best conference to attend if you're interested in vibration analysis and condition monitoring and reliability improvement well there is a conference called the International machine vibration analysis and condition monitoring conference it's a mouthful but it's a great conference and we run those you probably figured that out we run them in the United States in Australia in Europe so this we've got one coming up in June in in Antwerp in Belgium we just had the conference in agent which Singapore and in the future we're going to conduct them in Latin America China you have many events in England and we're running a lot of these conferences they really help people to learn anyway I know you want another proper question so I'll get onto it what's your recommendation for routine vibration readings number one you must measure a spectrum and the wave form a lot of people measure spectra not enough people measure a waveform as well this is a big topic but the simple answer is you need to determine what the best settings are for the waveform and I've got a couple slightly questions related to that so I will I will elaborate a little bit but it is important if you possibly can to measure the way form while you're out there because there are a lot of fault conditions where you can look at a spectrum and think I believe that's the fault condition but when you look at the spectrum it'll either convince you that you were right or that you were wrong and that you know you need to dig a bit deeper or the waveform will tell you what's going on but phase readings are really only for special tests so you normally don't do that on routine tests unless you're an analyst and you look at the spectrum while you're out in the field and think oh there's 1x 2x and 3x or just 1x or just 2x or whatever and you take a phase reading at that time now I would personally suggest that you should get two spectra one that gives you the vibration from about 0 to 15 times running speed and that makes it crystal clear what's happening with all those 1 X 2 X 3 X Peaks that I mentioned the 6x at the pump vein rate for example and other Peaks in that area also in bearings fail and we're talking sort of later stage failure here but in the normal spectrum you'll see some peaks there but a peak that's you know zero to 100 times running speed 150 times running speed in units of acceleration actually whereas this might be velocity we see the higher frequency vibration from gear mesh and the road about passing frequencies and these other frequencies but we can also see humps in the spectra from lubrication problems we see other things as well so it's kind of good and the bonus is that with this bedroom that has a higher max we get and particularly the resolutions right we get a really good time waveform that shows us just four to ten just depending on your settings four to ten revolutions of the shaft which is really good for time waveforms we can really see what's happening as the shaft turns and the gears mesh or the rolling element bearings roll along we can learn an awful lot the spectrum with the lower F max 15 times running speed will have a longer time waveform and you can see that's what this one's about you see beats and random problems coming from the outside you'll see cavitation you'll see you know random problems in gearboxes like a certain teeth come in to mesh then you only see the vibration at that instance and what I didn't mention because actually the question didn't prompt me but that's my fault you should have an high frequency measurement as well like enveloping demodulation peak view shock pulse HFD these sorts of things but there's some questions coming up on that soon so I'll mention more about that as we go along and so what would be the most important setting to use to have a nice tone waveform so I've mentioned that a little bit but there are really two properties the sample rates and the lines of resolution the sample rate it's it's kind of like if I was measuring a bullet smashing through glass or something like that and and oh sorry and I was trying to film it so imagine trying to film something like that you would want a lot of frames per second so that you could actually see the bullet flying through the air hitting the plane of glass and see all the glass shards you know fly off and and all the rest of it if and you've got a lot of frames per second but then you might watch it in slow motion if you use the normal sort of 50 or 40 frames per second or 25 frames per second I forget what it is normally you would only have a few frames that caught any of that because it happened so quickly well vibration analysis is just like that the analyser is inside digitizing the voltage signal that comes from your vibration sensor like this seller ometer and the sample rate which is often controlled by the F max setting determines how many conversions from that analog voltage signal to the digital waveform value and how many of those samples it takes per second just like the framerate in your camera the higher the frame rate and the higher the sample rate the more detail we see so as the rolling element rolls over that damaged area in the load zone of the bearing for example we get to see that much more clearly if we've got a high sample rate and that's why we need a high F max so that's all fine but that doesn't determine how many samples we collect this setting the lines of resolution setting determines the length of the time record now let's illustrate all this stuff so here's my time waveform that we saw earlier and if I use this lines of resolution setting of 400 lines it now this is just a simulation it's only kilt gonna collect enough of the time waveform so that it can produce the spectrum you asked for it just has 400 lines well the next setting up is 800 lines it needs twice as much information for 1600 lines it needs twice as much again and for 3200 lines it needs twice as much again so none of this affects the samples per second it just affects how many samples you collect the more samples you collect the high the resolution 'us of the spectrum but we're not talking about the spectrum we're talking at the time waveform and look how much time a form I collect with 3200 line the trouble is that you know if I was looking at bearing impacts and so on the longer the time wave form is more revolution so more impacts and everything else and you basically have to zoom in to see what you want and that's all fine we can zoom in as part of the analysis process but it's very easy to look at a time waveform and think that you don't need to zoom in so I kind of like getting my tongue waveform settings right from the get-go so that I can just look at my time waveform and say aha I can see signs of a problem or there are no signs of a problem and we're done so sometimes though it's good to have all this long time waveform because and I just mentioned it a moment ago we can see beating we can see cavitation we can see problems in gearboxes that are hard to detect otherwise now I hope this isn't too confusing I'm getting into a bit of detail but if you look at this closely the idea this is this is a shaft turning or if it's easier this is a gear turning and the vibration analyzer is taking a sample of the vibration from the sensor with a certain time interval based on that F max setting and so even though it's a digitization process even though it's converting from a voltage that came from the sensor to something digital that we can use inside the software it corresponds to a time delay which corresponds to the physical thing we're testing and there's samples taken as the gears emission together we can only see one gear but you can see the time between each sample and you can see then the little red dots represent the actual values that we can use to construct the tone waveform now this is a bit complicated and I apologize if you're new to vibration but I can demonstrate a couple of things with this so these settings you see here with the speed of this rotation of 25 Hertz or 25 times per second and an F max of 50 Hertz now I'm sorry 500 Hertz if I increase it watch what happens so I'm going to increase the F max and now if you look closely in there or closely here we are getting more samples per second we're getting more samples per tooth mesh if I increase the F max again we're getting even more samples per tooth mesh more samples first cycle as the teeth mesh together now if I really want to know what's happening is those teeth meshed together I need a high sample rate that's the basic idea here and I can achieve that either by specifically asking for it my software will let you do that all by increasing the F max so now I'm up to 2,000 Hertz but watch what happens as I'm changing this watch my waveform now this is for a specific resolution watch the spectrum down here and what's the waveform as I I'll just keep increasing it there's 2,000 Hertz 4,000 Hertz mean now means I'm getting much more spectrum and you can see that changes to the tone waveform so you drop it down to 2000 to 1500 to 1000 so as I reduce the sample rate this is dictating how many samples I actually get this is how many seconds there are or milliseconds between the samples if I now change the resolution I go from 400 lines to 800 lines to 3200 lines or 16 sorry I'm getting much better resolution like the peaks here are much clearer but I'm getting more revolutions so if you're not sure perhaps go back and play that bit back hopefully I said everything that was was important then on to a totally different topic does the key fazer not create unbalanced well technically yes it does it's a loss of mass at that particular point so it does create unbalance but when we balance it that you know unbalance is compensated basically but it is always worth understanding that yes it does create unbalance and inserting the key as we're coupling the shafts together also create some balance so you gotta be careful not to balance a rotor and then use a key that causes unbalance so here is just in case you were wondering what this question is related to the key phaser proximity probe is aimed at the key way or a key but a key way and our probes are sitting up near the bearings but you know you can see that this will create an unbalance there of course we gotta in a machine like this we got a great huge long rotor so the question is what impact does it actually have but anyway the balancing process deals with it another question what does it mean if one sees half of a specific frequency in a spectrum for example a fan has 14 blades but produces 7x that's a really good question and it's probably a bit more advanced it's not one that I've got a really easy answer for so number one if you have extreme distortion like the looseness I was talking about sometimes we get half of that frequency so in the case of running speed the looseness which is generated at running speed we might see peaks at half running speed and sometimes you know third running speed one possibility another thing is that if I haven't have a natural frequency at half now since a half exceeds like you know there's at 7x even though I got fourteen blades I can sometimes excite that resonance it's a long story but I can I can excite that resonance there's also this thing called common factors when you have interaction between for example the fan blades and the diffuser veins and so in a pump as well but we would need common factor of two now this is you know a complex point so I've got my simulator here so what I'm gonna do is so in this case I've got 30 teeth here this is 30 and I got 18 teeth here sits his 18 now look this is really complicated probably too detailed for this sort of a presentation but if we had pure prime numbers with these teeth so what I'm gonna do is bump that up to 31 bump that to 19 I'm gonna pretend to damage a tooth so there's our damaged two I'll highlight it to make it easier and now what happens if that tooth was damaged it comes into mesh here and kind of damages that tooth and then that tooth damaged tooth comes into contact with this key or damages that one now I'm talking about over lots and lots of rotations so let's speed that up let's speed it up if we watch for long enough basically all the teeth end up turning red and we therefore we get even where they all the teeth share though we're but if we don't have a prime number of teeth so I'm going to fix them all and I'm going to change this to a different combination just because I know this will sort of come up with the numbers pertinent to the question I'm going to damage the tooth and we're already highlighting it so if I speed that up a funny thing happens these teeth do not mesh with every single tooth on this gearbox so if I speed it up we see what happens if I go much faster we watch for a while what we will see is that every second tooth gets worn and so even though I'm referring to gearboxes here there's a similar mechanism that can happen with the interaction between pump veins and impeller veins it just depends on how many pump veins and how many impeller veins there are but if I slow that down you can see that every second tooth is damaged so what that means is that instead of like just getting a peak at 30 times the running speed of this shaft or 22 times the speed of this shaft that shaft which comes out to the same frequency to same frequency because they were meshing at a certain rate we also get a vibration that in this case is half of that frequency because it's only every second mesh that generates additional vibration so we can see peaks in the spectrum fact I can simulate it here and we have this gear assembly phase frequency peak but like I say we can see that in other machines as well okay I think it's enough said about that particular topic that's a possibility hope enough for getting another reason but anyway okay so here's another question what is HFD it stands for high frequencies detection shock pulse it stands for shock pulse it is it an overall value within a high frequency band well sort of sort of so what happens and I'll show you some animations in a moment with certain fault particularly as bearings calls impacts so for example the rolling element impacting something a damaged area on the outer race or inner race or on the rolling elements themselves all when gears mesh together and this or like metal to metal contact because there's a damaged tooth and in other cases in addition to the vibration that you might expect we also get these stress waves or shock pulses that's why it's called shock pulse yet these shock pulses and it generates high frequency vibration it's vibration is it's very low amplitude so we can't see it in the normal spectrum but we see it in the at very high frequency so we can specifically go hunting for it now one way with techniques like hft and peak view and just normal enveloping it's sometimes called or demodulation or peak impacting demodulation I've heard it called as well with all those techniques basically what happens is the vibration analyzer looks at the frequencies above a specific frequency so in some cases it might just be like when I was getting started in vibration analysis we used to just take all the vibration above two and a half kilohertz or 2500 Hertz and take all that vibration and go through this process to say I just want to look at the periodicity of the vibration we see out there in the case of shock pulse they have specially designed sensors let's just call them a seller ah meters but they're actually a little different and they they are designed to resonate so just like you can ring a bell and it will ring at a certain frequency these sensors ring at a certain frequency so whenever there's impacting or lubrication issues but we'll talk about that separately whenever there's impacting from the gears or the bearings or other fault conditions it Rings those sensors like a bell and that software is looking at a little band around that frequency and it can give it to you as well let's call it an overall value because that's what the question said but it can take it as a value and present that as something that can be compared against the lamb limits or trended or we can actually generate a spectrum and time waveform from that to see well how often was the bell rung so let's just have a look at an animation first if we have metal to metal impacting like the ball landing on the little bar there the blue wave you see and this this is an animation that I saw from SPM the blue wave is this stress way of all this shock pulse the second wave is the bar responding to the impact which is it resonating so the resonance part can be detected but the stress wave or the shock pulse can be detected as well and you know in the earliest stages of fault we may find it really hard to see the resonance but we'll see this distress wave so here it is in a bearing instead so it's all exaggerated okay I'm really trying to exaggerate things but as the rolling element rolls over the damaged area you can see a wave rushing out that's the stress wave or the shock pulse we can detect that and then you see the bearing you know shake because of the resonance so we can detect that when the damage is bad enough and when it is bad enough which it clearly is here you can see that periodicity in the spectrum so as you can see here there is a set time between each of those impacts and we will see then in a spectrum but only when the damage is bad enough that you can see it at the lower frequencies okay how can a lubrication problem be detected using vibration analysis - see techniques that we just talked about and then you can watch for the bearing failure because of lubrication problems often lead to bearing faults but the high frequency techniques now one way to do that is with that high F max I mentioned before I should explain what happens with the lubrication problem now this might be a bit confusing but imagine you could get smaller and smaller and smaller so that you could go right down where the gears are meshing together and the rolling elements are rolling over the inner race and we looked at it at a microscopic level right where those two surfaces are together and it's sort of moving relative to each other that's where the lubricant actually keeps those surfaces apart but that that you know and the surfaces are from a microscopic standpoint rough now if the lubricant is doing a tip-top job it's keeping those surfaces apart and it will generate a bit of noise and we can listen to that noise with ultrasound but from a vibration point of view it just generates well just noise there's no periodicity so there's no peaks in the spectrum and the tongue waveform is kind of boring but boring is good with vibration analysis but if the fault gets worse in others we don't have as much lubricant or this is contaminated with water or something like that the surfaces start getting closer together and okay exaggerated really close together now what happens now is that those rough surfaces are actually impacting each other as it's rolling along or the teeth are meshing together and but there's it's still not periodic it's like bang bang bang bang bang maybe they bang bang bang babbu just whenever these little little mountains and troughs if you like collide with each other we we get that bit of vibration now this is a basic way of describing it a little bit exaggerated in what we're seeing but that creates noise it excites the shock pulse sensor we can see it with our high-frequency detection techniques but we don't expect to see Peaks up here in the spectrum because there's nothing periodic we don't expect to see periodic spikes in the tone waveform therefore either instead it's kind of like the noise level goes up and we can see that it can be detected with you know shock pulse it'll tell you if this situation exists all the techniques can tell you or you can listen to it with ultrasound and just hear the change in sound and then perhaps look at the tone waveform okay changing gears completely what is your impression about how to quantify the ROI or return on investment in case of implementing this kind of technology okay this is you know potentially also a big topic to discuss but I think it's it's hard to do a very good job of this unless you understand the organization's goals like why is it that the organization doesn't like your machines to fail is it because of safety issues or environmental issues is it because of the high cost of repair of the machines is it because you know if failure occurs you can't provide a dependable service like if you're what a treatment plant or you're providing water to a city or generating electricity you know you want to be able to provide that's that service if you're in a manufacturing plant then the the the uptime is important and the quality of the products important you know on slowdowns and you you know you don't want those problems you know if your company has a lot of standby machines then you know a failure of a machine may not result in a loss of production and may not result in a safety incident but now you've got a machine that's failed so there are cost implications so you need to know that first like what is we're trying to avoid and so then you ask yourself well what's our current state like are we hitting all about production targets and we have no down time everything but probably not so you know what is the current state and what you therefore have to do what you can therefore answer is what is the cost of the poor reliability in the poor performance you know if your plant if your machete machinery just ran and ran and ran and ran and ran and ran well then your plant would operate it at full capacity now there are other reasons why you can have slowdowns and other things but just in terms of the sorts of problems that vibration analysis can help you avoid you know there's there's a good day when there are no problems due to faults that vibration analysis could have detected and then there's a bad day when you do have a lot of those faults and you want to compare the two and so that's you know that's what you're aiming for now again we could talk about this for a long time but you know vibration can help you detect the problems and avoid all of the consequences of the problems and what we really want to do is improve reliability so that the vibration analysts don't detect constant bearing faults and unbalanced problems and all the rest of it and therefore we don't need to have as many spares which is a high cost we don't need to have emergency maintenance which is risky costly and everything else you know we can avoid a lot of problems if we improve reliability but we can also avoid fewer but plenty enough to have a good ROI if we use vibration analysis and knowing all those things we can then use criticality analysis to sum up the benefits overall now I've got another question on that coming up so I'll say a bit more about criticality analysis in just a moment or how about we do it right now how do you utilize vibration analysis with the equipment criticality so let's go on from what we were just talking about to say if we had unlimited resources you could test every machine with all the condition monitoring technologies and you could test them once a week hey you could test them once a day and therefore you'd know exactly what the condition was nothing would catch you by surprise and therefore you could always plan maintenance well in advance and you know life would be wonderful well anyway but the reality is you don't have unlimited resources and so somehow you have to decide well okay well which machines should I test can I actually test them all do I have the resources to test all my machines as frequently as I really need to you know where can I justify using more than one technology you know oil analysis and vibration analysis and so on you have to have a way of prioritizing what you do and equipment criticality analysis helps you to do that you also need the PF interval to help you make these decisions as well so let's explain what those things mean so firstly from a PF interval point of view and I apologize if you've seen this before and I probably had it in my presentation but we're assuming that for a period of time the machine just runs happily and it's doing its job and something happens which causes it to ultimately fail so at a certain point we call that point P in the PF interval we can actually detect that there's a problem so now we can't detect it the question is how long do we have before it functionally fails so functional failure isn't a catastrophic failure where the bearing ceases and the Machine damages itself further um it might be just that the machines unable to perform its function we say well we got to stop this machine because it can't pump the way it's supposed to or whatever but the fact is there is a time between this point and this point and knowing that we can decide well if we test measurements frequently enough will detect this but the truth is that number one that curve is a little different to what I what is often depicted you know the condition doesn't degrade very much for a while and we can detect it but the actual health of the Machine hasn't degraded much but it's not until it really starts to fail and that's really in that sort of the last hours and days of the failure when the bearing starts to make a sound that you can hear or it gets hot and becomes loose because of the loss of material so it's really in this or the last stages so that's one point I want to make another point I want to make is that there is you know varying times it all depends on the load and the speed of the machine and and the failure mode like what's causing it to fail we could have years of warning we could have seconds of warning that's why you know big big critical there's that word critical machines that can fail catastrophically like big steam gas turbines and boiler feed pumps and so on that's why they're monitored with protection systems that when they see certain situations arise boom they shut them down carefully so there is no safety incident but I still haven't mentioned what criticality is so let's do that on a really simple basis we can just say well how critical is a piece of equipment is moderately critical you know is there a major consequence of failure is there an extreme consequence of failure and unfortunately most people kind of think of equipment in just this way or maybe even simpler than that it's just it's critical or it's essential or it's non-essential it's just as simple as that and so if you look at a production line you might say well if any one of those machines fails and then the production line stops so therefore they are all equally critical well no that's not the case so what I'm gonna do is I go I just need to jump ahead or we'll be here forever the number one thing I'm going to do is say well what do you mean by the consequence of failure well all these little statements here are saying well the consequence of fattie gets worse as I go from left to right so this gets a score of 1 this gets a score of 5 this is the saw the worst thing we can think of but just from an equipment failure point of view what about people you know the minor first aid or no first aid single or multiple fatality what about from the environmental point of view you know this is really bad what about from a production point of view no impact on production versus it's going to cost us more than 10 million dollars of loss if we lose this machine what about from a customer quality and the impact on the customer point of view you know here our food product that we're producing kills our customers that's pretty bad so I could go into a lot of detail here but what we need to do is sort of say well for each piece of equipment what is the consequence of failure you know and it's going to vary in each of these categories and secondly we have to say well how likely is it that it is going to occur because you know you may have a whole lot of machines in a row that you think well they can all stop production but are they all capable of poisoning the customer are they all as costly to repair do you have the same sort of spares for each one are they all as reliable as each other you may have one machine that's been running for the last 15 years and doing a great job and another machine that fails more frequently I need to pay more attention to that one that fails more frequently it becomes more critical criticality is the combination of the consequences of failure and the likelihood of failure but we need to go a step further because not just the likelihood of failure being initiative initiated it's the detect ability because I might have a machine that fails quite frequently but I'm a hundred percent sure under detect that it's occurring so never gets to fail and so yes we could be worried about these things but I'm so sure that my protection system will function that the detect ability is very high therefore it actually reduces the criticality of the equipment now without that protection system you know we would have had a highly highly critical piece of equipment and that's what justified the detect ability and that's what goes back to that ROI question recently Here I am it's already 50 minutes I apologize but anyway hopefully it's useful so we we justified the expense of the protection system because we needed the detect ability to be very high you know I could talk about this for a lot but this really helps you justify programs both you know from a cost point of view and to decide which machines need to be tested we need a combination of detectability reliability and consequence of failure which all creates the criticality ranking okay on a different topic how can that how the trends could be used to analyze the data something I didn't mention in my webinar actually and I should have because whereas we measure the vibration we can look at just the overall level and we can look at the spectrum and tone waveform and phase and orbits and things that I did mention with the overall level and with chunks of data extracted from the spectra that we're being collecting we can see how machines changing over time because to a very high degree if the vibration isn't changing then we sort of become less worried about it even if the amplitudes a bit high well it's not changing now it doesn't mean we don't need to consider whether that's you know a safe way to operate the equipment and we might be inducing failure but in terms of you know priorities you need to deal with the machines where the vibration is changing before you deal with the machines where the vibration isn't changing assuming that the criticality is the same so we always need to remember criticality we need to deal with our critical critical machines because the risks are much higher so with our trends we can see you know how quickly the problem is developing we can see the trend going up quickly means that the condition is degrading but we can also extract frequency from bands in the spectrum we can take the vibration just around that 1x peak that I talked about almost an hour ago if we see that changing we can say well the unbalance is getting worse now there is a special spectrum plot called a waterfall plot and we can see you know how different Peaks are changing in amplitude over time but a trend which sort of extracts its information from all those spectra gives us very good information and it's also very good to extract that machine condition information that I just talked about and compare it with other parameters and they could be temperatures flows production rates you know other parameters and we can say well let's see how the vibrations changing you know against these other parameters we might notice that the flow is changing in the vibration is changing now it could be the flow that's affecting the vibration of the vibration that's affecting the flow you would have to figure that out but it is a very good thing to do to look at trends of both those values and it really helps us to understand the history of the equipment health we can look back you know if you've been running your program for five years look back over that those times and you can see our here's the time when that bearing failed here's the time when the imbalance suddenly got worse here's a time you know we can kind of see over a period of time and learn from that okay hopefully that answers that question if a random frequency or a peak at a random kind of frequency shows up without harmonics will that be a concern the first question you always have to ask is did it do that and there was a high amplitude peak or a low amplitude peak generally speaking when low amplitude Peaks pop up you sort of say well I don't care so much now that all depends to you know because the peak that popped up might be related to a bearing frequency and hopefully we would have seen signs of that before it popped up in just your normal spectrum but a few things to think about then you can decide whether it's bad or not number one it could it have come from an external machine or maybe the plants operating differently now or for a variety of reasons now the vibration from an external machine can be transmitted through to the machine you're testing and that's actually the source of the vibration so you may want to investigate why it's changed you know why am I seeing it now not before might be a problem with that machine but that's one way to explain it in a rare case you will see a peak at for example the ball path at a race frequency because and that's the frequency that we normally see when there's damage on the outer race but damage on the outer race will create harmonics as well but if the outer race was out of round we can see a peak at this beep efo without all the harmonics necessarily anyway when you see any peach you then got up sort of put your mind into a certain state we had a look at is it synchronous like well there wouldn't be a random frequency of space because that would well so synchronous means 1x or 2x or 3x or 4x or 5x 6x you know an integer or whole number multiple of running speed so if I suddenly saw a peak come up and have to think hmm what would generate vibration at exactly 5 times running speed maybe maybe it's actually a forcing frequency maybe it's the pump vein rate that I've never noticed before and it's suddenly growing or you know it's usually something like that if it's sub-synchronous which means less than 1x I have to think differently you know belt wear will show up as a peak there and but sometimes you get harmonics in that case turbulence will show up although it's usually a nasty big blobby peak that's a technical term blobby peak but um yes we might might see that anyway we need to think about that and the non synchronous you know could be an external machine generating the vibration might be from the bearing as I described it could be from another part of the same machine you've got gearbox or belt then a peak at non synchronous could be a synchronous peak from the fan not from the motor or something like that and then look in a rare case you have something called inter modulation which is like a sum or a different frequency so let's say we have a peak at frequency a due to something in the machine and a peak at frequency be due to something else in the machine sometimes we'll see a peak appear at a frequency equal to a plus B and sometimes a minus B or B minus a which ever brings a positive frequency and so we can look at and say oh you know what does that mean is this new peak that showed up what does it mean well you don't become too allowed it's you might ask the question well why am I seeing it now like what make has changed these are interacting with each other anyway enough said on that I think if I see a peak of vein paths or blade pass frequency what would be the defect there are few possibilities uneven clearance between the veins and of all new to other you know sort of between the blades in the housing and so on can cause that you know if the blades or veins are bent or damaged in some way or the guide vanes of being tore damaged you know we expect the water or the air or whatever it is to sort of flow through the pump but as those blades turn or as the veins turn this should be you know in terms of a once per revolution point of view this it should be sort of even if I can put it that way that's the way it should be designed but with damages to the the guide vanes for example all the blades themselves or you know if there was misalignment or something causing you know changing those clearances then we can see it increase in amplitude at the peak it's common to see a peak but an increase in average is what we're worried about there and since we're talking about pumps what is the vibration analysis device well I would argue that you want a two channel analyzer to channel vibration spectrum analyzer that portable data collector measurement device whatever want to call it two channels is good because it allows us to measure phase much easier but we definitely want spectrum and time waveforms from that pump and we want high frequency detection just like we talked about quite a while ago you know we need to be able to do those things so now which actual commercial product I couldn't tell you that we work with everyone everyone has a great product but you know what we're going to do on our pump is we're going to use our vibration analyze we're going to put the sensor on there we're going to get a little spectrum from it and you know in a measure maybe in the horizontal direction as well and a measure over on the pump and detect all sorts of things good-day poor alignment technique be the cause of bearing looseness now I don't think about this and think could you a line of machines so somehow you made the bearing loose but what can happen if misalignment can damage the bearing so you with misalignment you're putting a lot of extra load on that bearing which will cause it to fail it it's inevitable it will fail prematurely no it's earlier then it really should have and so with that wear or loss of metal or you want to call the damage that occurs we can increase clearances within the bearing and we may diagnose that as as looseness but that's a real late stage a bearing fault I think that's a fair enough answer to that if familiarity with the equipment is very important do you recommend in-house vibration programs over contracting the service so first I should comment on the basis of the question and I absolutely positively believe that it's helpful to understand the machine and its failure modes you know if it's just this big metal box and you want to do one what's inside there it's not harder to do vibration analysis that's much much much harder if you know what's inside and you can say well I can imagine now if I'm seeing those harmonics that we talked about earlier or a peak at 6 X or whatever it is we're seeing you know you can sort of think about what's inside the machine or if we see an increase at the pump impeller rate you know the pump lane rate or blade pulse rate you can think about what's inside and make a better judgement and that's why you know it can be that a person who was a mechanic on those machines and repairing those machines can do very well at vibration analysis because at least understand what's going on inside so then the question is well if you're you know providing that as a service can you still do a good job well the answer is yes I mean sometimes the consultants may be more familiar with the machines at your plant because they've seen those machines at other plants you know more familiar perhaps with people at the site you know can be vibration analysts who never take the time to really understand what's inside the machine so that's one question but either way who's ever doing your vibration analysis should take the time to learn about the machines and ask all the questions that they can there are pros and cons to in-house over you know service providers you know service providers might have all the gear and experience and training and certification and you can get started more quickly if it's an in-house program they should have them they probably have a much better relationship with the maintenance people and production people they can respond very quickly if there's you know some someone hears something or you know there's a process parameter that changes or for some reason people suspect a fault condition it's easy to do follow-up special tests there's pros and cons ok what is your experience with algorithms software diagnosing vibration readings and if you feel these will phase out certified analysts and should I send remember this during the webcam look the bottom line is it appears to be inevitable that one day computers and the robots probably not robots probably just computers will be able to measure vibration and detect fault conditions now it takes a lot of experience with machines to be able to diagnose faults because you're really one way or another you have to know how the vibration changes and you do that either by seeing enough faults that the software kind of learns how the vibration change that you know led up to the fault condition or analysts have to sort of say well you need these features from the data and you need to interpret them in a certain way but I can tell you that actually way way way back in the 90s I was involved with a program like this and we actually developed a pretty darn good system there was a pretty good system we weren't trying to phase out vibration analysts the way we looked at it in the way I'd still look at it today is that these techniques can help the vibration analyst because if you're out there measuring lots and lots and lots and lots of machines and then you got to sit there and analyze lots and lots and lots and lots of data because so few people have really effective alarm limits set up that's a lot of work and it's really easy to make mistakes and miss faults because you're just churning through so much data so an automated diagnostic system that can look at all that data and say all hey I see a problem here here here and there that that focus Europe focuses your attention and now you can analyze that data first knowing what the computer said but you know you're using your expertise to really get to the bottom of it and then you understand that the process and the production demands and all those other things and and you can help make the decision about what to do you can go and perform special tests so I would just say that if well you know put this if your primary role is to test the machines and from an analysis point of view you're really focusing on rolling them at bearing faults then yes the computers might one day you know with online systems with wireless sensors and all these things everything has to be justified you know everything has to have an ROI but you know it could be that one day you know a lot more machines are being tested remotely but that's even that's a long way off and then of course software can then take that data either from the portable system or from these online systems and do some analysis so basically you need to learn how you can add more value by diagnosing a wider range of faults by being able to do special tests that enable you to diagnose a wider range of faults anyway I think it's enough on that topic what is your sense on trending condition-based meat and starter canis monitoring data and aligning that information with real-time process data that may include process hiccups like spikes the temperature variations in process flow etc it makes perfect sense it's a very good idea it definitely helps so I think I mentioned a little bit about that in the trending question but vibration analysts should be well let's say condition monitoring specialists should be looking at the entire health of the machine and the performance of that machine tells you about the health it tells you whether some mechanical fault is affecting the performance or the performance is affecting the mechanical health or a bit of both and the performance includes is just the way it's being operated so it is very useful information and for sure if we have an online monitoring system or even if we haven't if we can look back and say AHA this vibration change we observe it happened at the same time that some other process parameter changed and then we can investigate the rest more why did it change that way it really is good information to have during the presentation you said that the measurement has to be taken at the same conditions but what about different process conditions between measurements are those measurements useless no definitely not useless it just poses a bit more of a challenge the the best thing for a our vibration analyst is where the condition of the machine just does not change from test to test to test and so when you see a change in the vibration you say well the only explanation is that the condition is changing and now you have to interpret what that change means but if the the process conditions change and you see a change in the vibration you then saying well gee why is it so was that the Machine health mechanical health that caused the change or the conditions now look when you dive into vibration analysis more deeply there's an awful lot you can learn even with varying conditions number one you know on the one hand if you've got bearing faults for example you always see harmonics at non-synchronous ratios of running speed like 3.0 nine and six point one six and these sorts of frequencies so regardless of the machine running speed you will see these Peaks they are multiples of running speed that's why I talked about orders way back at the beginning 1 X 2 X 3 X if you know what the running speed is it even if it changes you can still figure all this stuff out so there are certain patterns that only show up in certain situations and we can use that information but otherwise what can happen if you know just generally the amplitude changes gets a bit trickier because of resonance resonance amplifies vibration just within a limited band so if under one set of conditions the speed is such that resonances are not excited then all the amplitudes of all the peaks you could kind of argue related purely to the condition of those individual components the pump impeller the gears the bearings and so on but if the machine speed changes now the vibration may coincide or be just close to a natural frequency therefore resonance occurs therefore the amplitude changes and people can easily jump to a conclusion say goodness me look the 2x peak is really high we better align this machine but it's only that the 2x happens to be closer to resonance now and it increases in amplitude it's got nothing to do with the underlining the state of misalignment it has everything to do with the resonance so you need to sort of get a sense for how the condition of your machine and how the vibration will change under different conditions because even if the speed doesn't change just the load loading can really affect everything from you know the magnetic fields inside the motor and the amount of slip - you know what's happening is the liquid for example goes through the pump and so on that's why it helps to really understand your machines and not just look for patterns and jump to a conclusion ah there's that terrific question again what the same question again what is the best way to be trained yes you are right see this training works now you know that Mobius Institute's the best way to be trained because we have all those training centers so there's one close to you you can learn online or offline you can use all those animations I mean let's face it all the things we've just talked about weren't they easier to understand with all those animations and simulations you bet they were and we use them all the way through our classes and no other training company has anything like it not not a shadow on what we have we have literally hundreds of animations and - what do we have - simulators gosh there must be 300 400 related to vibration something like that not that we use every single one in every single class or anything like that but anyway yep and accredited certification there is no higher standard to our certification and what's the next question ah the conference's yes come to a conference speak at the conference share your experience it's fantastic but you can't just rely on that one three or four or five day training class to be an expert in vibration and yes there are situations like the one you're experiencing right now where you can continue to learn and that's fantastic but you know we run those conferences so that you've got lots of opportunities to learn learn from the other attendees learn through workshops learn through fantastic case studies really interactive case studies it's it's a great way to learn and if you're up for it share your knowledge okay that's all the questions there were a couple of questions I left out just because of the nature of them I thought we could be here for a very long time trying to answer them and I thought that a little bit left of center so if you have more questions feel free to email them through but look I really thank you for listening to this whole video if that's what you did I thank you for listening to the webinar I hope it has helped you and the next time you think about training or a conference now you know where to go thank you very much on there's our website we can always learn much more by now

Channel: Mobius Institute

Views: 27,546

Rating: 5 out of 5

Keywords: IMVAC, vibration, analysis, training, certification, machine, balancing, unbalance, alignment, misalignment, looseness, resonance, bearing, ball bearing, sleeve, faults, wear, diagnostics, diagnosis, ISO 18436, mobius, institute, jason tranter, iLearn, iLearnVibration, Mobius Institute, CBM Connect, The CBM Conference, CBM Conference

Id: 5GAsfu06H-I

Length: 74min 7sec (4447 seconds)

Published: Thu Apr 19 2018