Vibration Analysis 2

[raysmith480]

Gentlemen,

I'm fresh out of college and just took a position with Honeywell Automations and Control Solutions. My welcome aboard gift was a COMMTEST vb1000v. I was told "here you go...now have this program operational by July."

I have been given the task to determine predictive maintenance on pumps and compressors by using this instrument. We are responsible for supplying cooling and heating (high-temp. hot water boilers and chillers) for all of a military base.

If anyone has tips on how to use this instrument or any helpful websites, it would be a life-saver.

v/R

Ray Smith IV

 

[dgallagher]

Commtest (865-588-2946) can provide training in Knoxville, TN. Technical Associates of Charlotte, NC offer excellent seminars on vibration analysis and setting up Pdm programs (

http://www.technicalassociates.net)

iLearnVibration is an interactive PC based program that would also be of help:

http://www.maintenanceproductsdirec...n_Training_Software_p/mob-ilearnvibration.htm

http://www.reliabilitydirect.com/trainingproducts/ilearn_interactive.htm

 

[Dnicejr]

Ray,
To have your program up and running by July with no background in predictive maintenance is a big job. You might want to consider getting a predictive maintenance provider in your area to help you set it up and run it for six months to a year and then you take it over. This would give you time to shadow them and see what they are doing and why they are doing it that way. In the mean time as Dave suggested take some vibration classes.

 

[TPL]

In a previous life I used to set these systems up, and I think that this is can be done with some conditions.

The biggest difficulty that I find is that people in a similar position have been far too ambitious in their approach and try to cover the entire plant in one hit, with the result that any mistakes made at the beginning stay in the system.

I would suggest that you select just one or 2 machines and 'play' with the software and hardware at your disposal so that you fully learn how it works - if possible, build a very simple route, populate it with one or two machines (that are easily accessible) and take one or two (or more) sets of readings each day - you will soon get trends that you can investigate further.

Then, figure out how to set alarms to alert you to any changes from the 'norm'. Set these deliberately low in these early stages so that you figure out how they work.

For the software setup:
a) determine the speed of your machine
b) setup to collect:
overall velocity
velocity spectra to 20X running speed
@1600 lines, hanning window and 4 averages
velocity waveforms to say 0.5 secs

overall acceleration
acceleration spectra to 5kHz
@3200 lines, hanning window and 4 averages
acceleration waveforms to say 0.5 secs
enveloping measurement

+ anything else that your datacollector can accomodate

Apply this schedule to a vertical, horizontal and axial measurement location as close to each bearing as possible.

All of this is a little overkill, but if you have the time to master this, you can then use your experience to set up a suitable set of templates (one for each type of machine) that you can then copy. Its better to collect a litle too much data in the early days and then dispose of it later, if it proves to be of no value.

If you make mistakes at this point on just one or two machines, its no big deal - just learn from them, bin the data and reconfigure to put things right.

Populating a database with a large number of machines from properly configured templates, takes no time at all, provided the templates are correct - the early effort to get the templates right is worth it.

If the budget allows, get any training that you can. Read and understand the instructions that came with your kit.

Be prepared to miss a few problems whilst you are getting setup

Most important of all - work safely

 

[raysmith480]

Gentlemen,

Thank you for the feedback. I am going to attend Commtest's training in Knoxville sometime in the next couple of months. Until then I'm just going to play around with my new toy and take it one step at time. I will not lie this project seems overwhelming to me.

I do have a couple of questions though in the meantime.

I don't really know where to place the accelerometer on the machine, and don't know if I should mount it rigidly, or use it like a stethoscope.

The instrument I am using only has one channel. Is that going to affect the precision of my measurements?

Is there anywhere online or inexpensive book I can buy that will help me get a jump start on spectrum, waveform and demodulation analysis before I attend training?

I hope these are valid questions. I want to do the best I can with this program. Also, if anyone is interested I was assigned a mentor today by ASME. Maybe you gentlemen would like to join this program.

Best Regards,

Ray Smith

 

[GregLocock]

The book to look for is an HP one called Introduction to signal processing. Bruel and Kjaer also do some good books, - their green paperback on signal analysis is a pretty heavyweight book.







Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[TPL]

With respect Greg - those books seem to be a little OTT for a beginner, I have always found B+K Stuff to be very good, but much more academic than practical - although its explained in great depth, I'm not sure that a real beginner really needs to know the maths behind signal processing that's for later.

If Ray can come to terms with what a waveform is (what the transducer output looks like on a 'scope) and then understand in basic terms what a spectrum is, that will put him in a good position for your training.

There is plenty of information available on the net and some good introductory texts on machinery vibration measurements - do a search on Amazon or Barnes and Noble for 'machinery vibration analsyis' as as starting point

Ray - you should use a magnetic base to fix the accelerometer in position. This gives good repeatability (by removing the human 'push down' variation) and adds a bit of safety - you put the magnet in place and concentrate on the instrument operation.

You should also try to mark the measurement positions (a dab of yellow paint?) so that the transducer is applied to the same point each time.

Most rotating elements are supported by bearings - vibration changes due to machine faults and bearing problems are best found by measuring as close to the bearing as possible. Measurements should generally be made in the vertical, horizontal and axial directions, although a little experience will allow you to drop some of these.

No problems with a single channel instrument - it will do just fine for identifying problems and most diagnoses.

 

[electricpete]

"I don't really know where to place the accelerometer on the machine, and don't know if I should mount it rigidly, or use it like a stethoscope."

Loose contact will severely attenuate the high frequency components such as might be important in diagnosing a bearing defect.

I would recommend a magnetic mount designed for the purpose.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[GregLocock]

If you are measuring the same machine regularly why not stick on a proper mounting plate?

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[tmoritz]

Ray,

Training is the key to getting up to speed quickly. I was put in to the same situation, handed a HP 35670A spectrum analyzer and told to diagnose a noisy 300HP double reduction right angle gear box that had just been rebuilt. I learned a lot.....of what doesn't work.....and how to generate useless information.

I can recommend courses provided by the Vibration Institute,

http://www.vibinst.org.

You may just want to purchase their book, "Basic Machinery Vibrations" to get a good introduction to the field. There are other companies out there that offer training in vibration analysis but I haven't participated in their classes and therefore can't make a recommendation one way or the other.

One thing to note. According to ISO 18436-2, establishing a condition monitoring program is a skill attributed to a Category III analyst, a person with three plus years of experience in the field. Without experience expect to make mistakes. I like the recommendation of contracting with a vibrations analysis shop to establish the routes, database, and alarm levels as well as showing you how to collect data. You can then retain the company to assist in analyzing out of the norm traces and spectrums.

Good luck,

Tom Moritz
Mechanical Engineer

 

[JohnCochrane]

Hi Ray,

I'm from Commtest, but this is not a sales pitch!

All the recommendations above are good, feel free to follow what makes sense to you. But, please be aware that Commtest has prepared an introductory vibration analysis handbook that is availble for FREE download from the Commtest website. This should be of some assistance until you get your training in Knoxville, or bring in a consultant to help establish your program.

http://www.commtest.com/uploadGallery/pdfs/beginners_guide_030306.pdf

Best regards,

John Cochrane
Commtest

 

[svanels]

establishing a condition monitoring program is a skill attributed to a Category III analyst

, and if you don't have or are a Level III?

Educate yourself by:
1) training
2) books
3) networking
4) getting your feed wet

Some suggestions:
networking :

http://maintenanceforums.com

Books:
Vibration Fundamentals and An Introduction to Predictive Maintenance both are written by R. Keith Mobley.
Machinery Vibration by Victor Wowk

I wished I had these books years ago in school when I was reading through the scarce (and sometimes) exotic B+K stuff.

We are implementing a vibration monitoring program using a commtest VB3000 and it is a "constant education" traject.

Be prepared to "tear" equipment down, or it least find all the documentation and drawings you have on it.

You will need info about running speeds, type of impeller, fan blades, bearings, couplings, rotor bars, loads on the equipment, oke anything that is moving inside or outside the pump.
Moving outside are also operators and mechanics which also are big contributing factor, but that is an other story..

Strange enough, when it comes to vibration monitoring, the only thing that people associate with is bearings

As some specialists state, vibration problems are:
40% unbalance
30% misalignment
20% resonance
10% the rest

 

[electricpete]

I agree 100% that the first thing you should do is get onto maintenanceforums.com There are lots of rotating equipment predictive maintenance / vibration monitoring professionals there.

You will find on the net a lot of literature spread from very theoretical to very practical (observation-based). As an engineer you might be tempted to try to relate everything back to theory. I would suggest to first get familiar with the general empircal thumbrules (like the infamous vibration wallchart), and then slowly try to match your own theoretical explanations later if you want.

I would say one of the broadest most practical references on the web is here for free:

http://www.update-intl.com/VibrationBook.htm

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Monagas]

Hello my friends. I would like to know your experience in vibration analysis using Windrock.
I'd like to apply it on engine driven gas compressors.

thanks

 

[electricpete]

John
I looked through the beginners guide. There is some good info about sensor placement etc.

I noticed one thing misleading. On page 51 it shows US uses ips and the rest of the world uses mm/sec (I believe this is generally true for machine condition monitoring).

The problem is that the document indicates a conversion factor of 25.4. Unfortunately in the US overwhelmingly ips is generally expressed as pk/0 while in the European countries mm/sec is overwhelmingly as rms (for example in ISO standards). The conversion is then 25.4/sqrt(2) or somewhere around 17mm/sec rms = 1 ips pk/0

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[dgallagher]

I highly recommend you purchase the Database Development Tool (

http://www.technicalassociates.net/DDT-1.htm)from

Technical Associates to assist in setting up the data collection parameters for your program. If your program is of any significant size, you will need to be efficient as well as effective. Taking data at 1600 and 3200 line resolution as recommended earlier may prove counter productive for periodic survey (though fine for detailed analysis if a problem is suspected). Take the Commtest class and augment with sound practical experience from the likes of Jim Berry.

I have no relationship with Technical Associates, but with 27 years experience in vibration and predictive maintenance, I have found them to be the best non-denominational Pdm training organization.

 

[DAlbertson]

Hi Ray,

Here's a few more suggestions. I hope you don't mind, but I'm assuming that since you just got your degree that this is your first professional job.

1. Your management has really got you under the gun. You must communicate to them your plans, and make sure they realize that you will no-way-in-heck ever hit their July end-date. Be cool, enumerate what needs to happen, and communicate that plan firmly. Your management is very likely to be OK with this; certainly they will be much more OK with this approach than you showing up to a meeting in July and saying you're not complete. I know many, many engineers that are far better than I am who have been hammered because they failed to communicate. Devise a plan, then tell'em what it is.

2. I took the Vibration Institute's course back in the early 1980's, and I thought they were a very competent and knowledgeable outfit. I learned a lot. They get 1 vote from me. I can't tell you about the others that have been mentioned because I have no real information.

3. One of the two guys from the Vibration Institute said about 100 times (no kidding) during the course: "The number one cause of vibration problems is LOOSENESS!" He wanted it to be remembered.

4. If you took a signal processing course in college, I'd like to suggest you haul out the book before you attend whatever vibration course you select and re-read the parts on the Sampling Theorem, aliasing, IIR and FIR filters, and DFFTs. If you didn't take a signal processing course, then get your intro calculus text and review exponential growth & decay, and the Fourier Transform. This will let you focus on the vibration-specific parts of the lectures, instead of scrambling around with the mathematics and basic DSP concepts.

5. Since I'm already bossing you around and telling you what to do, let me gently mention one other thing. Don't use the salutation "Gentlemen" unless you are certain that there are no women readers, and maybe not even then. Lots of smart women engineers out there...don't exclude them; you may need help one day!

The very best of luck in your career!
David Albertson