Sign in to follow this  
Followers 0
Been Around

Vibration Monitoring

9 posts in this topic

Hi all, Using Rsview32 Works with ML1500 on a test stand to data log amperage and Hypot testing on air handler units being made on assembly line. Recently I have been tasked to integrate a vibration analysis to the test stand. We had a demonstration in plant and the links below are of the equipment that has been recommended to me. I currently get a digital signal from Hypot after testing occurs. I am using one of the serial ports on the ML1500 for Rsview connection. The below equipment has Modbus. I may not need all of the info in registers but would be nice if needed. Has anyone out there used this equipment before? Does anyone have or have used similar devices or devices that can accomplish same results? We are looking for a hand held unit for bearing testing on 1/2 to 1 hp ac motors with changeable sensors to use portable unit as backup to test station unit. Needs to interface with AB PLC via modbus, devicenet, etc... I want more info that just a 4-20ma or pass/fail digital output from unit. this is AB vibration analysis http://www.automationworld.com/products-1616 Thanks in advance, Portable unit www.spminstrument.com/products/t30/ Stationary MG4-12 www.spminstrument.com/products/mg4/ P.S. I did make up a temporary FLUID VIBRATION ANALYSIS DETECTOR Yep, I took a glass 1/2 full (or half empty) of water Made a straight line at water line with marker denoting PASS Made a erratic line at water line with marker denoting FAIL Patent pending Edited by Been Around

Share this post


Link to post
Share on other sites
That is awesome. Make sure it has a green "Calibrated" sticker on the side of the glass :)

Share this post


Link to post
Share on other sites
i checked the link and found this spec sheet: http://literature.rockwellautomation.com/i...td061_-en-e.pdf however I could not find the most important information such as vibration units, range etc. (it only says that indication is 0-100%). is there a document with more info? regards...

Share this post


Link to post
Share on other sites
None that I have found, However I did forward this to my local AB rep today and I am waiting for a reply . Patiently

Share this post


Link to post
Share on other sites
AB_Condition_Monitoring_XM_Modules.pdf AB reps on the ball Better docs attached

Share this post


Link to post
Share on other sites
I don't know what you are looking for but I'm not sure you know what you are looking for either. There are lots of portable units on the market. Right off the top of my head, even though they are a bit expensive, you can get a portable vibration detector from Mitchell Instruments (www.mitchellinstrument.com) for not too much money. It will give you average measurements in either acceleration (g's) or velocity (in/s or mm/s). Same thing the raw sensor with 4-20mA output gives you. These things tell you how good a particular system is, nothing more, nothing less. They are more than adequate for what you are trying to do. AB is feeding you on a sales pitch and they CANNOT DELIVER. I'll give you a related example that is easy to understand. If you want to tell how strong a part is, you can run a tensile test on it and stretch the part until it breaks. This tells you exactly how strong it is. But if you have a bad one, it doesn't tell you why. If it is a metal part, you can visually inspect it, perhaps with a microscope. You might even run an elemental analysis on it. All of these tests don't tell you anything about how strong the part is. They tell you WHY it might not be as strong as it should be. These are diagnostic tests, not quantitative tests. Diagnostic tests have value for determining the source of a problem, but have nothing to do with measuring how good a particular piece of machinery is. Now a little history...Entek bought out IRD Mechanalysis, which is a company that sold very simple vibration sensors such as transmitters that had simple programmable pass/fail digital outputs. They are purely quantitative in nature. Entek integrated multiple sensors along with temperature, pressure, etc., into a much more expensive piece of modular hardware and discontinued all the old IRD Mechanalysis hardware that was cheap and simple. The Entek stuff didn't sell very well because it was overpriced and did not interface well with PLC's. Essentially Entek was selling a condition monitoring SYSTEM as a complete package instead of individual pieces. They wanted to take over what used to be done with a handful of IRD monitors and a PLC. Then Entek was bought out by AB and AB's plan was to integrate this stuff into the PLC line. They kept the same kind of hardware. AB made it possible to connect the outputs to DeviceNet so that now you could get the pass/fail signals into your PLC. AB turned it into a very expensive condition monitoring system for people with 400+ HP fans and such that don't know any better but still do all the "analysis" for you. This system is for people that don't know much about condition monitoring that have very expensive equipment such as large fans (400+ HP) and centrifuges. This is what you saw a demonstration of. AB has been very poor at advertising exactly what it is they are trying to sell and this product line has never sold very well, partly because of the "secret algorithms". The target customers will usually talk to a vibration technician and find out the reality of what AB is trying to sell. Then they will find out that it is vastly overpriced. You can put together the same thing with your own sensors for a fraction of the cost. And it doesn't come with "secret algorithms". I have looked into their condition monitoring equipment before and I always dropped it because it didn't do anything that a PLC can't do directly, and there's way too much advertising about secret features but AB won't explain to a vibration expert what their secret algorithm does, so you can't really trust it to give valid results. Real vibration analysis actually uses simple 4-20mA or voltage outputs and for your situation, boils down to simple pass/fail. If all you want is a measure of how "good" something is, this is all you need. Real vibration analysis may also use an FFT and other calculations to produce a chart which can give an analysis of what kind of problem to be looking for. This information is purely diagnostic...it takes the simple "how much vibration" information and helps to pinpoint the source of the problem, nothing more. With machinery monitoring, the key metric is displacement. I'd suggest using the exact same sensors that most vibration technicians use. I like CTC (www.ctconline.com). They hold up better than anything else I've ever used. The basic device is an accelerometer. Ever since MEMS sensors came out, the size and reliability has taken a step change. Internally, a MEMS sensor is a very small mass floating on a capacitive field. The device monitors the mass and either adjusts the capacitive field to keep the mass centered, or allows it to float freely and just measures the voltage change. These devices are truly the first practical example of "nanotechnology" machines. The technology has been around in practical form for a little over 10 years now. This is what CTC's equipment is based on. The output of this device is acceleration, measured in g's. If you integrate the output once (a simple analog integrator), you get velocity, usually meaured in in/s or mm/s. This is as far as the low level hardware goes. Accelerometers also have specific bandwidth limitations. Sometimes depending on the situation, it is useful to buy a sensor with less sensitivity. For instance the 900 HP fans on the emission system in a plant in New Jersey are mounted about 100 feet in the air. There is a huge amount of low frequency vibration because of the structure that they are floating on (including that induced by wind). If you buy a super-sensitive sensor, you will be mostly measuring this. So we buy one with a higher cutoff frequency (about 150 Hz) to eliminate the low frequency signals and concentrate on the signals from the power train exclusively. Most bearings give specifications in terms of displacement, either in inches or mm. To get to this unit, you need to determine the speed of the equipment (RPM). Then it's a simple calculation. See this page for recommended motor vibration as well as the acceleration/velocity/displacement conversions: http://www.ckit.co.za/secure/conveyor/trou...sionfactors.htm The rest of the technology in a high end vibration analyzer is just an FFT which converts the raw digital signal into a frequency spectrum. If you have 2D or 3D signals, then you can also treat these as vectors and do various vector versions of the basic FFT. They often do have full blown text file type outputs, and they are very expensive and require quite a bit of training to understand how to use them effectively. This process is essentially pattern recognition. A vibration technician gets the FFT and based on some simple knowledge of the machinery involved, looks for certain patterns in the FFT plots. In addition to the information you get from the raw outputs (which are good enough for pass/fail), the FFT information tells you WHAT kind of problem you have. It can pick up whether the problem is related to the bearing or motor, looseness, soft foot, etc. It is not practically possible to turn do algorithmic FFT plot analysis...there is nothing on the market that will do the FFT and produce an output such as "bearing failing" directly from the information without a human involved in reading the patterns. In the real world, everyone who is doing condition monitoring uses a simple device to measure either velocity or displacement. If the value is excessively high, then they either try to guess at what the problem is and fix it (it is often obvious) or else they have a vibration technician attach a vibration analyzer and get the FFT plots. Usually this is unnecessary because the reasons for the vibration problem are often very obvious if you inspect the equpment mechanically (which is usually done simultaneously with a vibration technician). If you buy a good high end vibration analyzer, expect to spend a week in a class learning how to use it and expect to pay about $20,000. At this point they are very PC-oriented. You have some analysis capabilities on board but most of it is in the software (PC) side of things. The real thing takes the vibration analyzer inputs and does an FFT on it. All the charts and such that it displays from this point forward are based on the FFT input. If you have a 2-D (XY) or 3-D (XYZ) accelerometer, you can also do a vector FFT and calculate additional information based on that. An example of the best known company for vibration analyzers in the world is CSi. OK, now let's consider the specific device you referred to, the XM-720 from Rockwell. If you head on over to literature.rockwellautomation.com, you can download the manuals for the XM-720. The XM-720 gives you either relay outputs based on specific vibration analyzer set points or 4-20mA outputs. It can read acceleration and do the integration internally. If has some sort of analog type filtering available (high pass, low pass, etc.). I'm not sure (it is not clear in the manual) but it appears that since it takes a tachometer output and supplies displacement numbers, it probably combines the two signals to provide displacement. Doing the filtering in PLC math is easy. Just use a FIR filter calculation, or an IIR if you just want to do analog simulations. Consult google for "FIR" and "IIR" if you want to get some simple tutorials explaining how this is done. If the filters supplied in the XM-720 are anything more than simple single pole IIR filters (Y=(old Y)*(1-alpha)+X*alpha, where alpha is the adjustable parameter that is based on the cutoff frequency), I would be very surprised. So far this describes precisely what you said you didn't want. You wanted something more. Well, here's the rest. The XM-720 also has some sort of mystery filter that somehow is supposed to provide output information about "spikes" which relates somehow to bearings and gearboxes. I don't know what they are talking about since the algorithm is super double top secret. This automatically precludes it from use in any of my equipment since I can't rely on their algorithm either (who knows what it really does). Real bearing and gear noise most often shows up as harmonics of the fundamental (base speed). It can also show up as a subharmonic (fractions of base speed) with bearings when you have a specific rolling element going bad (because the rollers turn at a fraction of shaft speed). Either way, this has nothing to do with "characteristic spikes" and other very mysterious forces at work. The closest thing to that is if you have a soft foot or looseness problem, you will see massive amounts of noise as the nice clean peaks of the FFT have a bunch of random distortion. In practice you will also have lots and lots of harmonics and your basic average/peak displacement will be quite high compared to normal anyways so even though you might not be able to get the detailed information that you get from a vibration analyzer showing specifically that the problem is looseness or a soft foot, the result will be close enough. Real vibration analysis isn't all that scientific anyways. You are simply looking at a spectrogram (FFT result) and based on the pattern you see and some simple information about the equipment, you are interpreting what you see. This is diagnostic. I suppose it is probably possible for AB to run a DFT at a specific frequency such as calculating the first 2 or 3 harmonics and the fundamental if you have a tachometer attached. They could add the harmonics together and compare it to the fundamental to calculate an estimated machine condition. This is pretty close to the first few steps that a vibration technician does to check for balance issues. But I highly doubt that they are doing this at all because if they were, they'd come right out and tell you that, and because this only detects one of many types of problems that the machinery may be having. It might miss a looseness issue or a failing spider or rolling element in a bearing altogether.

Share this post


Link to post
Share on other sites
Paulengr, Thank you for the detailed reply, I know in this busy world it's hard for someone to take time to teach others. Everyone here should be grateful to you!! I am. I will take your recommendations and read more. This is what I have, We have had some HVAC air handler units being returned for excessive noise and/or vibrations through the ductwork on new installs. Most of the returned units have had damage to the fan itself, with the blades on the wheels being damaged. After looking at new wheels before the assembly process here we have found new, uninstalled wheels with damaged blades. We are working with OEM of wheels to remedy damaged units from them. There quality control procedure is redundant, but some of the damage may be in shipping. Some of the worse damage may be from installers removing damaged wheel during replacement and subsequent shipping. Some of the units being returned have had bad bearings on motors. Now, after meetingsssssssssssss, we have been asked to build new test stand for running motors and using equipment equilivent with the portable SMC T30 using shock pulse transducer to check bearings. We have been asked to add a stationary unit SMC MG4-12 or equilivent with a vibration transducer to a test stand I built 4-07 with a ML1500 and Rsview32 that is currently checking Hypot and amperage of motor during test of unit on line before packaging and data logging for comparison in the event of a returned unit. There is significant vibration in units with damaged wheels as the water in glass test showed. The AB XM-720 was looked at for cost analysis. To be honest it just popped up during search for Vibration monitoring. I see where the transducers have 4-20ma . And some of your reply tells me I can probably plc to gather info from transducers and do math to come up with a range of acceptable readings during burn in. Correct me if I am wrong. Transducers measuring range is 0 - 25 mm/s and the frequency range 2-1000 Hz. This is where more research on my part is needed. Using Vibration analysis with controls is new to me, but I will accept challenge. The diversity I have seen in the last 25 years doing controls, retro fits and upgrades have been nothing less than amazing. The simple answer to what is wanted, is if a unit has excessive vibration during burn in testing, Flag that unit from test stand. We Not trying to determine real time cause now, just stop units from leaving for install. Thanks again, Hope this helps your understanding of my understanding. Scorn me if needed. My current status: Read Paulengr reply Repeat

Share this post


Link to post
Share on other sites
Buying an accelerometer with a 4-20mA output hopefully with a magnetic mount and sticking it on the unit being burned in should do everything you want. NEMA publishes specs for motors (it's an MG spec but I forgot which one...they will probably be happy to sell you the spec if you E-mail or call). The bearing manufacturers publish specs for bearings. Between the two, you should be able to determine the proper limits to look for. Later, you may want to expand the test stand to include multiple accelerometers if you want to narrow down which component has the problem and plan on taking it apart and replacing only the defective components. For instance, you may want one on each bearing plus one for the motor itself. Bear in mind that a 200 HP induction motor has roughly a 1 mm air gap between the rotor and stator coils. You are talking about something much smaller than that, so the corresponding air gap is going to be very tiny. Most of the time, a simple megger test detects failed motors because any problem with a bearing or any kind of mechanical coil imperfection will fail a megger test. You simply measure insulation resistance with a DC test signal approximately the same size as your nominal voltage (230 VAC means test with something close to 230 VAC). The general IEEE rule is approximately 1 Million ohms per kilovolt nominal voltage, with a minimum of 1 million ohms. So in your case, you're looking for an insulation resistance of at least 1 million ohms for all cases.

Share this post


Link to post
Share on other sites
Thank you Again, I will order accelerometer from CTC online. I have two spare analog inputs existing in panel. This should get my feet wet with this particular process. I will post some results and assumptions I make for clarification or a slap on the hand. Training being researched also. Honeywell's web page on Accelerometers http://content.honeywell.com/sensing/senso...elerometers.asp Analyzing the vibration signal from an accelerometer does not always require expensive or complicated analysis equipment. Built in signal conditioning in the model MA15, MA321, and MA322 accelerometers provides a 4-20mA output that is proportional to the RMS vibration value. This built in signal conditioning means that with only a 4-20mA meter hooked up to your accelerometer you can be monitoring vibration, quickly, easily and cheaply. The MA15 provides an output proportional to g vibration, the MA 52 provides an output proportional to velocity vibration levels and the MA 321/322 provides g vibration down to zero hertz. This is a good site and info https://www.ctconline.com/__ctc_university.aspx Best Regards and wishes Edited by Been Around

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0