Fundamentals of Vibration Measurement with Screw Compressors

Hello. My name is Joe Pelis, Today's presentation is on the fundamentals of vibration measurement with screw compressors. This is one of a two-part. Screw compressors are quite large and have a lot of horsepower put into them. So there's always a vibration signature on machines and it's important to learn what we can from those signatures to know if maintenance is needed or if a machine needs some attention. So we'll start out with the basics today and work up to a little more specific interpretation of screw compressor details. We're gonna start off with a simple harmonic motion of a wheel vibrating. And we're gonna say in this case we have we'll support it on an axle that's lightly sprung and it has a heavy spot. So as the wheel rotates, it vibrates the axle up and down following that heavy spot. So if the unbalanced wheel is mounted on springs or on a flexible support, motion of the axle is easy to envision. So we will follow the motion of the axle from the heavy spot being in the top position here. To starting around till it reaches the center position, and then fully at the bottom. And then back up to central position. And finally, ending up at the top The displacement of the axle or distance it travels can be measured in any linear measurement system. So that could be inches or millimeters or mills or microns. All of those would be suitable ways of measuring the displacement. The amplitude of the displacement or distance the axle travels can be measured in any linear measurement. So that would be kind of the more correct language if we're looking at displacement. Amplitude can be recorded as a function of time, so we have a time waveform that would describe the motion of the axle. And if we plot the distance traveled by the axle over time, it would look like a sine wave in this case at least. So we would have our beginning point, the maximum distance up. Back to zero again, maximum distance down, and again back to zero. So the displacement of the axle can be recorded in different ways. So we could say, we could describe that motion as one inch zero peak. We can say it's two inches, peak to peak, total travel. We could say it's .070 inches RMS, RMS stands for root mean squared. A common term in in vibration reporting, or average. The average is just 637 times the 0 peak value as the RMS value is just .707 times the zero peak value. All are common terms in reporting vibration data. Okay. The time required to complete 1 cycle is the period. So we start here at 0 time. By the time we get back to the same position of the axle, that would be described as 1e period. In this case, we're gonna say suppose that takes a second, then we could define the frequency of vibration as one cycle per second in this case or 1Hz. Hertz is just another way of saying cycles per second. We could also report it as 60 cycles per minute. We want to put it in a slightly different format. Means exactly the same thing. You just have to be sure what is being reported for the frequency. Okay. So a car, I like to use a car as an example, travels a distance in miles, and the velocity or speed travels is measured in miles per hour. It's kind of intuitive to everybody. The velocity or surface speed with which the acts of travels can be measured in any linear measurement per unit time. So we could have miles per hour, inches per second, millimeters per second. So really, velocity is the rate of change of the distance the axle moves up and down. We can measure the velocity of the axle with the velocity probe mounted on the axle. So the velocity of the axle vertically is 0 when it's at its highest position. Then we start to travel down. We reach the highest velocity when we're at the middle position. And we return to 0 velocity at the lowest position of the axle where it turns around and starts back in the other and again, we have our maximum velocity upwards. At this 90 degree position. And then finally, we return to 0 velocity at the highest position of the axle. So if we plot the surface velocity of the axle over time, it would look like a sine wave with a time shift from the displacement, and the velocity can be reported in different ways. We could say it is three inches per second, zero peak. We could say it's 2.1 inches per second RMS or root mean squared. Or we could say it's 1.9 inch per second average. All of those really relating to the same data and the same relative amplitude of velocity. Okay. So we go back to our car analogy. He said the car travels a distance in miles. The speed it traveled is in miles per hour. When you push on the gas pedal it speeds up or accelerates. Acceleration is a rate of change of the velocity and is measured in, in this case, miles per hour per second. Acceleration of the surface of the axle can be measured in any velocity measure per unit time. So it might be miles per hour per second, and it might be inches per second per second or inches per second squared, millimeters per second squared, or g's. G's is also a common measure of acceleration, so that's just relating it to the acceleration due to gravity. So 1g is 32.17 feet per second square. And again, so the acceleration is the rate of change of the velocity. We can measure the acceleration of the axle with an accelerometer mounted on the axle, and we can plot the surface acceleration of the axle over time it would look like a sine wave with a time shift from the velocity. And again, acceleration can be reported in different ways, we could say we have 1 g, zero peak or peak. And it's also common if you hear someone say peak, they mean 0 peak. Could be .64 g's average on acceleration. Those are probably the two most common. And RMS is also common with acceleration. So again, .707 times the zero peak acceleration would give us G's RMS. So we have these three measures. Used in vibration analysis, displacement, velocity and acceleration. They are all related to other by integral calculus if you ever took those classes and didn't forget it the second you walked out the door kinda like I did. But you can calculate one from the other which the advantage of that means you don't have to have separate probes to measure each of these, you can have an accelerometer normally, which is what's used, and then a meter that would calculate one. From the other. And as we say, the measurement of the surface motion with a standard accelerometer allows calculation of any of the above units. Real time waveforms of rotating machinery are very hard interpret because there's a lot of different excitations, not just a simple sine wave against time. And the tool that's used to simplify that is called a fast 48 transform or FFT. It's a mathematical technique used to change time waveform data, as you see here. Into a more useful format plotted against frequency. So we quite often will see frequency on the X axis and amplitude on the Y axis, and it could be used for anything. In this case, I've shown velocity and have shown frequency in hertz. So that's a pretty common spectrum, as we call it, to look at vibration data. So we have common units of measurement, displacement in mills, peak to peak is quite common. Velocity in inches per second peak or 0 peak, and acceleration in G's RMS. Those are probably the most common units of measure. Frequency, I've seen hurts used a lot, but I've also seen cycles per minute used a lot. So You always have to look at the data that someone has taken to see which measure of frequency that they've used. So what is the value of the data? Generally, displacement is used for slow moving equipment. So like reciprocs or if you're looking for rotating in balance, displacement is a good tool to find that. Velocity is used more for I Speed machinery. It's best for new equipment now So if you're looking for imbalance, you can find that also with velocity. Alignment, pipe stress, resonance, To answer the question, is it safe to walk away from this piece of equipment? Velocity is probably the best tool to do that. Acceleration is more for bearing condition monitoring, metal to metal rubbing, pitting or or wearing parts. To answer the question, is it time to tear down the machine and and rebuild it? So that's really the main the value of these different units of of vibration. There are others G spike energy. It's kind of a derivative of acceleration. It's generally proprietary. It's owned by a particular company that puts it in all of their meters, but it is useful. G's enveloping is kind of the same. They take the G spectrum and do some funny math on it, maybe to exaggerate the high frequency content looking for maybe making it easier to see bearing damage. There's a few others, acoustic emissions, ultrasonics, HFD, as many companies as make vibration meters all have their derivatives of acceleration as a as a standard measure. These tend to be proprietary using filters and amplifiers to try to determine a better measure of metal to metal impact or rubbing. They're very useful for monitoring bearing condition but the best way to use them is to take a baseline and then trend it and look for change. So this two main purposes of vibration measurement. The first one is really new equipment analysis. So we put in a new screw compressor We just turned it on, and we wanna know, you know, is there is it safe to walk away from? So this would help us detect misalignment or imbalance. We could have a coupling problem. We could have a resonance in a structure. Or piping. We could have a poor foundation. We could have soft feet. We're gonna talk about all these things. But that's really the main first purpose of vibration measurement. Completely different from that is preventive maintenance, which is really I've already installed the machine. It's up and operating. It's been doing fine. Now I need to know when do I need to take the bearings out and replace them? So you're really looking for bearing fatigue over the long haul you'd like to detect it before it becomes severe and before you have much damage in the machine. So preventive maintenance is really a different reason for doing vibration measurement, then new equipment analysis. There's a few things that are important regardless of the reason that you're doing vibration analysis. Taking consistent data is very important. Calibrating the equipment on a regular basis is very important. I've seen people carry the same vibration meters from job to job and never check them. And sometimes they can be taking bad data for quite some time. And either condemning machines or not finding problems that they should find. So Whoever you've used or if it's you that is doing vibration measurement, make sure that you have your equipment calibrated on a regularly scheduled basis. And take the readings at the same percent load of the slide route, same speed, the same pressures as much as possible. Same probe locations. Quite often, I'll see people take a permanent magic marker and put marks on a casing if they're using a magnetic based indicator. So that they know they're measuring at the same place every time. This is important. If you are gonna take a a baseline and look for change, you need to be measuring at the same place steam pressures and all these things. And also use a wide frequency range. We'd like to see 0 to about 10,000 hertz will capture most of the data of importance. So if you're taking a baseline for future readings, put the probes approximately over the bearing position. Horizontal vertical axial is a good is a good way to start. On both ends of the machine and perhaps on the motor. So we all have a simple data sheet where you can just record the data that you take from your vibration measurement, and and then it's easier to compare over time. So what does a vibration spectra normally look like on a screw compressor? Here's a kind of typical one. So we'll see this is peak velocity in inches per second. Our frequency is shown in hertz, and you'll see a couple big peaks here. One of them is that 239.7. So that's about 240Hz. happens to be a four lobe male rotor and rotating at 60Hz means the discharge pulse comes out of the machine at 240Hz. So we would expect to see strongest pulse normally at 240Hz on a screw compressor. We expect to see harmonics of that. So we would expect to see the first harmonic at about four eighty hertz, and then we'll see additional harmonics beyond that. That is normal. That does not mean there's anything wrong with the machine. That just basically is telling you that it's running. This is to talk a little bit about the importance of probe location. So what I've done here, I've got peak velocity in inches per second, and you'll see in this case we've got about .1 inches per second. Our biggest reading is at 476, so I'm gonna say that's the first harmonic of load passing. I can see the load passing frequency here at two thirty nine, and then I can see the other harmonics of low passing. But in this case, the probe was mounted on a foot. So we had relatively low level, point one, and the First harmonic was the largest measurement. Now this is exactly the same machine. We just took the probe and moved it to the discharge pipe. So on the discharge pipe, we're not reading point one, we're reading .4 inch per second. The pulsation is much stronger on the charged pipe than it is on the foot. And you can also see that the the low passing frequency is now higher than the first harmonic. And again, that's just by moving probe position. So the the point of this is you have to be measuring at the same places if you're gonna take a baseline and compare data. Now here's some spectrums in g's. So look at our units, RMS acceleration. In G's, frequency in hertz. So again, we've taken measurements on foot of a machine. And reported it in acceleration. Again, we see this 476 is our largest reading at about .5 G's. And then there's quite a bit of additional signature level at the Harlanics, and that the 240 is fairly low. And then we'll shift it to the discharge pipe. And again, you can see that the levels go up considerably. And again, the 240 shows up stronger on the discharge pipe. And we sort of missed the first harmonic, but then we go out to a third, and it's it's considerably considerably larger. So again, the importance of taking data in consistent location or you really don't have anything to compare to. So here are those that same data measured on the discharge pipe and on the foot. In the same scale. Now these are RMS acceleration in G's, and you can see the signals look quite different from those two locations. And okay, so then that's kind of enough on the language of vibration measurement that we can get started a little bit on new equipment analysis. So we'll talk about why are we doing this. So typically people don't call in a vibration company. To take data very often unless something's wrong. So they might say there's parts breaking. We need to figure out why. The noise is unacceptable. We get structure borne vibration that's shaking the boss's office, or we just need to establish a baseline, maybe it really is a new machine and we just wanna get a baseline of what the vibration looks like before it starts running. So we assume we're going to do some problem solving, particularly if parts are breaking. How do we how do we do that? The first one I recommend is really just going around and touching components. Now they can be hot. So you don't wanna touch them directly but I quite often will take a coin between my fingers. And just run it around the machine and on pipes and on different places. Look for the places that are moving the most. Anything that that hurts your fingers or buzzes or you can tell the ones that are moving the most. So take the readings on those. Analyze the frequency and the amplitude. You have to know the compressor enough to know how many lobes are on the male rotor. What speed is it rotating? Is there a gearbox inside? Is there a VFD? Or you won't know where the frequencies are you should expect? And then a little bit of thinking is required. Viribration analysis is not is not always simple, so you have to use your head. So we're gonna start out saying what are some of the probable things you'll see in the spectrums? And particularly if the signals levels are high, and what are you gonna do about it? So let's say you've got a high 60Hz. That can be a concern because it could be the male rotor that's out of balance. It could be the motor rotor that's out of balance. More likely than not, it's a coupling balance issue, coupling Generally, are more likely to have sixty hertz problems than either a compressor or a motor, but it can be any of them. Soft feet, which really means that the feet aren't pulled down solid, so that they're they're not giving a good base for the machinery. Alignment, which is really motor to compressor. If the alignment is out, that can cause a 60Hz problem and also piping stress. If you have stress in the piping, it tends to pull the machine particularly as cold suction pipe cools down, it can pull the compressor out of alignment with the motor, so that would show up at 60Hz, generally. So screw compressors are very precisely balanced balanced to very low levels. So it's it's really unusual that you would find the 60Hz problem because the compressor is out of balance. It's unusual to see more than about .01 inch per second 0 P on a screw compressor at 60Hz. So if you see more than that level, look for a coupling that has a problem or an either out of alignment or an imbalance problem, check the keys. The keys quite often are the source of imbalance problems on couplings. So we talked about soft foot, you need to make sure the feet are tight and that they are in solid contact with the base. We need to make sure that the coupling keyways, well, there's two answers here. One, if you have a balanced coupling with match marks, you have to put it together according to the match marks. If there are no match marks, which is quite often the case, then a good way to assemble a coupling is with the keys 180 degrees apart from each other. So that There's always a little bit of imbalance induced the keys and the keyways. So if you have them a hundred and eighty degrees apart, it tends to offset and can reduce the overall recreational level. Proper key sizing is very important. So when we balance the compressor rotor, we use a square key. So this little empty space at the back of the keyway, you don't need to fill that because when we balance the rotors, we account for that. So we're take removing material from the rotor to account for this the imbalance of the the in mill slot. Another important point with keys is the length of the key is very important. The key needs to be sized to preserve balance. And size the key to fill the empty slot in the shaft and the coupling, the best you can. However, there can be issues. For example, if the keyway slot in the compressor is exactly the same length as the coupling hub, then it's easy. You just fill both slots with the key, and you know what the right key length is. That's very seldom the case. This is another case where in this case the slot in the shaft pretty long, and the cup and cup is shorter. So again, we need to shorten this key because that's a big rotating and balance of all that mass of keyway up there that isn't offsetting anything. So this key should be shortened, and we'll talk about that. There are two wrong ways to do this. One is to fill the slot with key. The other one is to fill the hub with key and leave a big rotating empty spot in the chat. Neither one of those is ideal. The best way to size keys, take the length of the hub, take the length of the inmail slot in the compressor shaft and divide by two and take 95% of that and that's how on the key ought to be. It's a real good approximation for the way to preserve balance in in key lengths. So there's my simple equation starts to a little bit fuzzy, but ninety five percent of the average length between the two is a very good compromise on key length. Okay. So that was 60Hz. Now we go to 120Hz. If that shows up on your spectrum, what do you immediately expect or suspect? Coupling alignment is also on this list, so that can show up at sixty hertz or one twenty hertz or both. Soft feet on the compressor will generally show up at one twenty. And it could also be piping stress. And I think I've got it applied on soft foot. So if you loosen the bolt and the foot rises, that's what we would describe as soft foot. So when that's the case, you need to loosen the foot, shim the foot, get all four feet on the compressor to the point that you can loosen the bolt and the foot doesn't rise and then you have to go back and do the alignment. Oh, once you're sure you've gotten rid of soft flip. Going up the frequency scale. We talked about 240Hz a little, that that is low passing frequency, one of four lobe male rotor, running at 60Hz, and the first harmonic is at 480, those generally just mean the machine is running. Now if the levels get too high, sometimes we have to do something about it, but we expect to see 240 and 480 on. So if we've got four low mail rotor, again, we would expect 480 on 240Hz in 480 harmonic, in a sixty hertz country. If you're in a 50Hz country, then 240Hz would be the load passing that we would expect. Her small screws tend to run five low mail rotors, so you would have to take your rotational speed of the shaft times the five lobes to have an idea where your load passing frequency is, and that can get a little complicated particularly if there's a gearbox. So like all of our small screws generally have gearboxes. So you'll have to know the rotation speed of the male rotor in order to know what load passing frequency to expect on the vibration data. Now we do publish all of those frequencies and the gear tooth ratios and all that stuff. So depending on which compressor you've got, we can give you a chart that's got all of those low passing frequencies on it for each size machine. Okay. So we're come back and talk about we've talked about load passing frequency and the harmonics, but that is what we would expect to see. That to how do we decide though if it's too high? And there's a lot of different material published around this. I just this is a lot of data here which I know you can't read, but there's lots of charts out in the industry that talk about vibration severity guidelines. Frick also publishes a vibration severity guideline. You can't take a lot of these for screw compressors because generally screws the the low passing energy and screws at 240Hz. Would put them into the category that might be described as rough in a lot of this type of material because a lot of these were originally prepared for things like centrifugal compressors that have no pulsation energy in the low passing. Safrik publishes this vibration severity guideline. And in the guideline basically at about 0.4 inch per second 0 peak, that's about the point that we would say that we go into alarm. And up to about one inch per second zero peak where we would say you need to shut the machine down. So that's at guess pulsation related frequencies, so that means accounting for a two forty and four eighty. If you take out those frequencies and you're looking at other general frequencies, then the levels can be somewhat lower where we would where we would go into alarm. And these are larger machines on the right, and the smaller machines are on the left. So this is general publication, that's pretty easy to get from Frick. Now this is based on ISO codes. So if we have a high 240, 480Hz, what could the problem be? Well we know the discharge pulse is the source of that. And we expect that to be somewhat I know. There's a few things we could say. The first thing I usually check if I have a high two forty is that the volume ratio is protect properly calibrated. If the VI is wrong, and somewhat isn't hasn't calibrated right and you're running off of the eye, that alone can cause a high 240 or 480 hurts pulse. So that's the first thing to check. The other thing I usually check is main oil injection. If the oil injection is too much, it's very easy to increase the 240Hz pulsation in a machine. So watch your discharge temperature, crank the main wall injection down until the discharge temperature is near target range, I'd say, for an ammonia machine, generally about 180 Fahrenheit. And if you can cut the oil back and stay under one eighty, lots of times it'll reduce the 240Hz energy. Other things can be wrong. The separator could have a structural resonance that amplifies this two forty or four eighty. There's lots of structural residences and all sorts of machinery and piping. So it is possible that that could give you a high level. If the foundation is supported poorly, it's not shimmed or it's on a resonant base or it's not properly grouted. All of those can give high vibration that will generally show up at these frequencies. Piping itself can be in resonance, I've seen suction pipes and discharge pipes not supported very well, or even just the wrong length, and it's possible that they can they can produce a pretty high amplitude at two forty or four eighty. And refer to one thing as this forced vibration. It can be high just from high energy. If you've got a 2000 horsepower machine, that's a lot of energy in the discharge pulse you should expect to get somewhat higher readings. And I'll say up to about .4 inch per second, 0 peak is pretty normal. At 240 and we're not gonna recommend any action until you get to that level. Higher horsepower, it may even be slightly higher than that, but anything much over point four than we start paying attention. I used the word resonance a couple times and I'll talk a little more about that. Residence I describe is everything is a bell. If you hit a bell, it wants to ring at the particular frequency it was designed for. So you it's me struck and it's unrestrained, it'll ring at its natural frequency. And everything in the world is a belt. When you look at structural piping, Piping supports, brackets, everything has a resonant frequency. And if it happens to be right at 240 or 480, then sometimes those can get excited and you might see a high level on some parts. How do you tell if you've got resonance when the vibration of a particular part is excited to a higher amplitude than the source of the vibration? That's a pretty good indication that there's resonance going on. So if your compressor is only moving point one and your separator is moving point four. It's probably resonance in the separator. Now, it wouldn't be a concern at point four if it's moving an inch per second, then it might be concerned. So if you got a oil pipe or a piece of tubing that's vibrating at a higher amplitude than the points it is attached to on the vessel or pump that also is a real good indication of resonance. So what do you do in a case like that? Add a clamp, add a bracket. Something to introduce damping into the system is generally how you change resonance in a part like this. Sometimes a bracket might come loose. You need to look for things like that. You tighten the brackets back up and you get the damping back in the part, perhaps the resonance goes away. And we'll separate whether it vibrates more than the compressor it's attached to is suggesting that there can be some resonance in the separator. And that's not that unusual. It's unusual for it to go to a high level, but there can be areas of the separator that will vibrate somewhat more maybe than in the compressor. So as I said, if we got a high 240 or 480, what do I do? I mentioned the BI calibration that's really important, try reducing or increasing main oil injection, generally decreasing reduces 240Hz. Make sure the package feet are shimmed, grouted, and bolted to the floor. We have a separate guide foundation guideline paper. It's very important that you follow that in installing these machines. Or you can expect 240 and 480Hz of vibration. If the high levels on piping, brace it to increase the stiffness, to drive the natural frequency up. That's usually the first thing we would try if we got a vibrating piece of pipe. How high is high? As I said, up to 0.4 inch per second peak is normal. A few spots on an oil separator up to one inch per second, though high, may be okay if it's not near any small bore tubing. All separators are very thick and strong and steel, and they're not gonna break. At that kind of a vibrational level. But small bore tubing, if it's attached close to one of those areas could possibly break. Two to three inches per second even on heavy metal parts is risky for breakage and requires changes. So when you see levels like that, it's time to shut machinery down and change something. I'd like to bring up the nature of vibration fatigue failure. If a part has not failed in 1 X10 to the seven cycles, it is not likely to fail in fatigue. Unless the amplitude increases. At 240Hz, that takes twelve hours. This is one of the things that helped me sleep at nights. So if you have a machine, let's say it's vibrating .5 inch per second on this separator or maybe even more, .8 on the separator and you're trying to decide, is it safe for me to walk away from this? The first question I'll ask is how long does it run like that? If it's more than twelve hour especially if it's been six months. I'm probably gonna say, you're fine. If the if steel doesn't break, and one time standard seven cycles, it's not gonna fail in fatigue. So unless the amplitude goes up now, if a bracket falls off and the amplitude goes up, Then again the clock starts over, but vibration fatigue happens very quickly. Fatigue in general happens very quickly, especially at these higher frequencies. So again, that's an important point that gives us some comfort on machines that might be running a little higher than we might like to see. If we see higher frequencies like 1500 up to 10,000Hz, usually that is friction like rubbing parts, It can be bearing fatigue that's one of the strongest indications that we need to do something with the bearings, or can be structural resonance. I've seen even housings, compressor housings, sometimes that'll have a funny peak in the spectrum out in this range. Sometimes it's a characteristic of the machine. So if it doesn't change and it's not going up perhaps it's not a concern, but can be from all of these. Another less common occurrence is torsional resonance So if you picture your compressor as one mass and the motor as another mass with the coupling in the middle, if it's actually possible for them to get vibrating against themselves and be going back and forth in in resonance. So that is condition that we have to be careful about, the answer to that is generally that we have to make sure that the coupling stiffness is right. This is not a common problem on smaller machines. It's rare with motors below about twelve hundred horsepower. With engines, it's a very real concern. So if you've got a natural gas engine, you need to do a torsional vibration analysis before you before the machine is built. And we we will do that. If we sell you the engine, If you're measuring machine that's being engine driven, then you should find out if someone has done a torsional vibration analysis, because the complex pulsations from an engine, it's very easy to get into torsional problems. Speed increasers, turbines, all of them, gas turbines, or steam turbines. It's possible to have torsional resonance at any frequency at which the equipment runs. So again, that's something to watch for, but not not on smaller motors. And so I'd say generally on larger motors, you can do a simple torsional vibration check, which can be done writing Coohr. Coohr has the ability free to put in your coupling stiffness and inertia, and the motor stiffness and inertia, and it'll do a quick check of where the first torsional resonant frequency is. So what happens if you hit this, it can cause failure of rotors, couplings, gears, motor rotors, It requires special equipment to measure it in the field. And as I said, the best defense is always perform a torsional vibration analysis on complex drive trains and and and engine drives. And also then motor drives over about twelve hundred horsepower. And that can be done during the engineering phase. Here is one example I've got a machine that had a torsional problem And you can see it was like pulsing at one particular spot of rotation on the drive train, and it actually fatigued in a straight line down the the rotors. So that was kind of unusual, but I like to show people what can happen with a torsional problem. Okay. So that's the end of part one. We're gonna we have a part two when we'll pay more attention to the measurement of vibration after we have our machine installed and running. So then it's really machinery condition monitoring, and that'll be in part two. Thank you.