RCM-2009 Reliability Centered Maintenance Managers' Forum - March 23-26 - Daytona Hilton Ocean Walk Village - Daytona Beach Florida Join or Manage Your Profile Posting Boards Machinery Condition Monitoring and Predictive Maintenance Posts About vibration/alignment/balance Cost of outside Tech/Consultant to run routes and analysis dat?
Join or Manage Your Profile Posting Boards Machinery Condition Monitoring and Predictive Maintenance Posts About vibration/alignment/balance Cost of outside Tech/Consultant to run routes and analysis dat?|
Go
|
New
|
Find
|
Notify
|
Tools
|
Reply
|
|
For all your information.
Once tested a fan undergound, normal readings for a good fan in this position was usually aroung 5mm/sec Velocity and 0.5g's max. This fan was 70mm/sec velocity and about 8g's. Probably the closest thing to failure I have ever tested. This fan was doing 3000rpm and being used to suck the dust away from a continous miner in a coal mine.I stopped this fan as soon as I finished taking the data. I was lucky as I had been given the authority to do this if I believed it was necessary. We pulled the coupling apart (what was left of it) and found that both the motor drive end bearing and the impellor bearing had approx 1mm of lift. I thank god the other 2 bearings held out while I was testing this fan. A little training for the fitters that night and I have never seen a problem like that again. It has been almost 5 years since I did that night shift. Hooch |
||||
|
Velocity or Acceleration....always use BOTH. If you are just looking at acceleration, you will miss mechanical faults. Just looking at velocity, you will miss bearing faults.
If your software/analyzer allows it, collect your spectrum in velocity (typically 2000 Hz, 1600 lines for me), but your time waveform in acceleration (4000 Hz Fmax typically). My meter shows me the spectrum (velocity) and the waveform (spectrum) when I'm done collecting a point. This way you see everything, and can collect 'extra data' on the spot to pinpoint a problem. Regards, Rusty |
||||
|
Late reply. Been out on sick leave with a bout of strep throat.
Your right, bearings are funny animals, I've seen bearings fail when running in the green, and bearings in alarm keep on running. I just continue doing my job, informing the maintenance supervisors of the condition. Thats all I can do. If they decide to keep running the machine they at least know the risk...
R. Bell Baton Rouge, LA |
||||
|
Hahaha.. I sent a report of a blower bearing at 70g yesterday. Still waiting!!
Quote:
I've seen bearings fail when running in the green, and bearings in alarm keep on running. . Doesn't this suggest the the definition of green category is wrong as are the alarm settings. 2 nominally identical machines can have entirely different vibration characteristics, and applying universal rules to all machines is not the best way to go. I've got machines that have shown high and steady vibration for several years (likely damaged but not failing) and other machines that show very low levels of vibration - the alrm requirments are different in each case. |
||||
|
Hey Buzz,
All of our machines were set up in the database using the parameters established by Technical Associates. Depending on weather it is a fan, pump or whatever. We even set the alarms for each specific style and type of pump. Some as low as .15 ips and 2.0g's Generally speaking over the years I've done this, I have observed this happen in regards to machines crashing weather they are green or otherwise. Sometimes you get no warning despite the monthly routes.
R. Bell Baton Rouge, LA |
||||
|
No one will ever get 100% results with periodic monitoring, that's a recognised limitation, but the trick is to make the best of what you have, and the most valuable resource is time.
The charts out out by those good and knowledgeable folks at TA are a good starting point, but they have probably never seen your machines and they certainly haven't see mine, so final responsibility for setting working alarms rests with you and I. As I said before, universally applying the same rules to machines is not the way to go - all machines are individuals and 2 nominally identical machines can exhibit very different vibration characteristics - for example, I have 4 'identical' pumps and the alarms range from 0.05 ips to 0.25ips and from 0.5g to 4g (all peak) and alarm settings are often different in the horizontal and vertical directions. I also make extensive use of frequency band alarms: these were set up just after the program began and are based on the 'normal' spectra. If anything, most of my rolling element bearing failures are picked up by the accleration frequency bands (500Hz to 5kHz~0.1g). It takes a lot of effort to get the program setup and running properly, but getting the alarms setup properly sure beats looking at all the spectra. This is for a monthly walkaround - anything that does slip through the net (two unexpected failures in three years out of 2500 machines) gets put into a weekly or bi-weekly route for closer scrutiny. I've got about 6 machines in these more frequent routes. Your plant and resources will most likely be different to mine, so what works for me might not work for you |
||||
|
Yes I would agree that everyones plant and situation isn't the same. When I mentioned we used TA standars this is just a baseline or starting point for our machines. After we initially set up all the routes we got a feel for the good machines and the ones which were shakey. We did lower some of the alarms at this point, and started checking these problem machines at least twice a month, some even more frequent. Our place is divided up into 5 parts and everything is checked once a month regularly, and I stress the word regularly.
As I said in my original post, even this will not catch all machines. Sometimes things can and will go south quickly in between the monthly routes. Hasn't happened much in the past 6 years. Only about 3 times. Our downtime is minimal. We've made great improvements since implementing the vibration monitoring. Using both ACC & VEL readings. We don't do spike energy as far as finding these super early bearing faults, but we will catch these in stage 1 etc.... Thats the best I can answer to your comments. I don't think our program is any better or worse than any one elses in this forum. I believe the vibration monitoring field is pretty much standard in the industry as are the schools. After all we all learned the same info and use the same standards pretty much. Regardless of which equipment you use to gather data, wew all persue the same goals and end result. So far we've not had a failure all year, (knock on wood) something which was pretty rare in the past. Last failure was October 2007... Guess we're doing something right eh mate!
R. Bell Baton Rouge, LA |
||||
|
I did not know this. From what I read on this forum, I do not believe this. Regards, Bill Bill.Foiles@bp.com |
||||
|
My intention was to leave this open for discussion. I'd like to hear some feedback on this, as far as the training, and criteria.
Some of the 20 year plus guys should know this. I wonder as the vibration technology advanced over the years and more companies started getting involved in it, did these places like TA come into being. Anyone know? I guess what I'm getting at is who set the standards... R. Bell Baton Rouge, LA |
||||
|
The machines sets its standard. Like Bill, I find it hard to find a set pattern of folks doing it consistantly the same or coming up with the same results or advise.
I typically collect vibration cap data for the most part using acceroleters and don't bother acquiring velocity data on those machines as a rule. I don't see the need; maybe some do. Integrating from acceleration to velocity is relative and not absolute. Do you use two transducers? A 6312 bearing set in a machine: there is not standard. One machine's mass and physical dimensions, etc... may vary greatly as well as the damping factor which results in different failure patterns and magnitudes. Let the machine talk - it will tell you what you need to know through the majic talking rocks. OK, add voice to your output. Noise travels through steel rapidly but putting one's ear to a rail may not be totally relavent to vibration analyses in all instances. ISO standards are a collections of ideas basically from qualified people and are very good but that standard does not tell you to take the pump out of service at 9.9 g's or 0.9 IPS. The wall chart may get it OK sometimes and miss it others. You get 20 experience in 20 yrs time not in a week's seminar. Cost of the outside consultant: how about qualified consultant? He/she will put money in your pocket and is cost justifable wereas the 'get the guy because his rates are lower' may cost you plenty. A good outside consultant will greatly add to your ROI increase. Cordially, Sam Pickens pdmsampickens@gmail.com |
||||
|
Quote:
I typically collect vibration cap data for the most part using acceroleters and don't bother acquiring velocity data on those machines as a rule.I don't see the need; maybe some do. Sam - I have noticed that you have made similar statements in other posts - can you expand on what you mean? Do you never take velocity readings on bearing housings? If so, how do you detect a change in 1X vibration on say an 1800rpm pump? Would you consider acceleration readings sufficent to detect this? |
||||
|
I guess when I said standards, what I was referring too were the actual schools, not the machines themselves.
When starting out in vibration you learn basics, then you go to more advanced courses. These basically teach the same methods. (I think) Thats what I meant when I said we all learned the same thing. I know what you mean Sam, about having the same machines, yet they have different problems or characteristics. We have that situation here at our place. Some of the problems are obvious, some not so. We use an accelerometer for both ACC & VEL readings. We need these lower frequencies for fan problems other than bearings.
R. Bell Baton Rouge, LA |
||||
|
Do I consider an accel reading sufficient - YES!!! General statement. Select the right transducer!!! If you are calling "Velocity" using an accelerometer and integrating to velocity; you did not acquire a velocity spectrum you merely took the voltage given by the accelerometer and integrated to velocity. If the accel isn't good enough, then don't use it. It did not majically become a velocity spectrum. 0.05 g's becomes 0.1 IPS at 30 Hz or 1800 cps. If you want a true velocity spectrum you need a velocity transducer. Else, you only have relative data. Having said this; ensure you are using the correct transducer! Is its application correct and adequate resolution and correct frequency range. When looking at a slow turning machine calculate voltages at 1X rpm and 0.4X as a rule. NOw look at the voltage the accel will generate at the frequency of interest and compare against the accel's spec sheet. Is that voltage in the noise floor? Is yes, then you must change to a different transducer. You may need a transducer down to 0.05 Hz and linear with a resolution of 0.0001 or better. Select from the OEM's available. I like PCB and Kistler but there are a number of good manufactures but don't expect to pay $150.00 or get it on eBay. Consider all enviromental factors and call the OEM if necessary as I think they will be glad to help you make the selection. Do you have low pass filters? Cordially, Sam Pickens pdmsampickens@gmail.com |
||||
|
Sorry Sam - I must be having a blonde day - I still don't understand
Quote: Else, you only have relative data Are you saying that for general purpose machines (1800 and 3600rpm) that using accelerometers for obtaining velocity spectra is a waste of time? I did a test a while back where I compared velocity spectra from a Bently Velocity transducer (the moving coil type)and Wilcoxon accelerometers integrated to velocity and couldn't see any difference over the range 0.5X to 20X for some 1800rpm pumps? For the purpose of condition monitoring I couldn't see any reason not to use velocity spectra obtained from accelerometers. What am I missing? |
||||
|
Are we there yet? Does your program provide planned/scheduled maintenance? Then you have achieved and your data collection must be adequate or so I would think.
Sorry, I re-read my post and don't see a rational problem or its difficulty. But I wrote it. If you have selected the correct transducer; its voltage generating capabilities above your threshold and sufficiently above the noise floor you must be OK. If you are wanting to see 0.03 IPS at 5 Hz using a 1V/g accel clean and clear then it must be above the noise floor and its characteristics such that 0.03 IPS is able to produce a clean signal 'out of the dirt'. If you had 0.03 IPS at 60 Hz then that should be = 0.03 V but at 5 Hz 0.03 would correspont to 0.0025 V so your resolution must be above 0.001 or thereabouts. Naturally you would be concerned about environmental things as temperature relates to many transducers and their frequency ranges. So on low frequency items I may select a magnitude of interest at a given frequency and make sure my transducer provides a clean signal. It if isn't clean in g's, it isn't clean in IPS. Hope this helps. Cordially, Sam Pickens pdmsampickens@gmail.com |
||||
|
Nope - I think I have an understanding of why signal to noise ratio important, but maybe you could help me understand by giving a quick overview. If I have say 100 pumps running at a nominal 3600rpm, and asked you as a consultant to look after my vibration monitoring needs, what vibration parameters would you monitor on the pumps? Would you make acceleration measurements only and no velocity at all, or what? |
||||
|
Buzz,
What do you guys have out there mostly pumps? Are you at a pumping station?
R. Bell Baton Rouge, LA |
||||
|
Sam, I was wondering, when your taking your ACC reading where are your FMAX's set I'm assuming around 50-100K...I was thinking maybe you set the FMAX lower since you don't use velocity.
Wouldn't that enable you to zoom in a bit at the lower CPM's Just like you would looking for motor faults. Just wondering. I know our SKF Microlog can take either kind, but we have it set up to take ACC and the info is converted to VEL as well. Thanks,
R. Bell Baton Rouge, LA |
||||
|
Motor driven pumps (100s)
Fin-fans (100s) Centrifugal compressors - driven by motors and GT (aeros and industrial) As you can imagine, there is a lot of effort put into vibration monitoring. I think I have a pretty good program in place, but I am always looking to learn more so it can be done better. There is a large time saving to be made if I could ditch velocity readings on the smaller machines - both in terms of time required to collect the data and the time needed to analyse it, but I need to understand the downside first. |
||||
|
Yep, roger that! Always plenty of ways to improve. We've been trying to impress upon our maint people the importance of doing quality work, and that reliability starts with them, and good sound work practices.
I was asking you about the pumps because I've found that the VEL reading assist me in vane pass problems, and alignment issues, as well as general mechanical looseness. I was picking Sam's head on his FFT settings thinking perhaps he used a lower freq range and higher resolution to catch these low end problems where these peaks would be more pronounced in the ACC mode. I also do a lot of phase testing with the strobe. I've found that very helpful as well to see how a machine is moving. Sounds like were all striving to improve. Good deal...
R. Bell Baton Rouge, LA |
||||
|
| Previous Topic | Next Topic | powered by eve community | Page 1 2 3 4 |
 | Please Wait. Your request is being processed... |
|
|||
| Site | Join or Manage Your Profile | ||
| Servlet | eve1da019 | ||
| Version | 1.3.0 build 9383 | ||
| Module | Forums 4.0.3 | ||
| Stylesheet | "Eve" | ||
| Wordlet Set | "Default Wordlet Set" | ||
