What analyzers are available that use shock pulse method? What are the advantages/disadvantages between both methods? For a first time setup, starting a program from scratch, what would you guys suggest?
From what I have seen, most of our breakdowns are due to bearings failing.

Thanks

This message has been edited. Last edited by: Xracer,

Make a search for SPM and PeakVue respectively on the board and you will find some things to read. Olov

There are other methods besides shock pulse and PeakView that are widely used. You should also research demodulation, also known as envelope detection.

I would suggest you evaluate your equipment, operation and personnel. Let the plant dictate and listen, then temper per your budget.

Are you a large facility 600+ people with a lot of costly equipment? If you are wanting a good top shelf program versus an overall level program to include spike energy or shock pulse or?

A simple monitoring program may put you at an ME-42 (Conditech) or similar hand-held device recording several simple parameters. But if your budget can allow and/or your plant demands a comprehensive program you may want to investigate a number of vendors that supply full route based programs - there are a fairly large number of vendors but it will pay you to investigate most all of them to make a logical decission.

quote:

Originally posted by Sam Pickens:
I would suggest you evaluate your equipment, operation and personnel. Let the plant dictate and listen, then temper per your budget.

Are you a large facility 600+ people with a lot of costly equipment? If you are wanting a good top shelf program versus an overall level program to include spike energy or shock pulse or?

A simple monitoring program may put you at an ME-42 (Conditech) or similar hand-held device recording several simple parameters. But if your budget can allow and/or your plant demands a comprehensive program you may want to investigate a number of vendors that supply full route based programs - there are a fairly large number of vendors but it will pay you to investigate most all of them to make a logical decission.

Thank you for your replies. Yes we are a large facility with a lot of costly equipment. A rough est would be 40,000 pieces of equipment. We currently are using a contract service for doing our vibration analysis. I do know that they are using a csi analyzer. The word on the street is that the company would like to bring this service in house. In a discussion, shock pulse method was mentioned, and I had not heard of that method, so I have been searching and reading up on it. I would have to say, that a good portion of our failures are bearing failures. Most are at the too late stage or very close to too late. I have seen outer bearing races spun in the housings and inner races spun on shafts, followed by full bearing failures. What is hard to figure out is what happened first. Was the housing fits loose/tight. Was the bearing fit on shaft loose/tight. Correct me if I am wrong, but with shock pulse method, it detects ball to race defects but would not detect a loose race fit?

It appears to me that your outside contractor is not up to snuff. They should not miss this type stuff; so, that excludes phasing their service in and taking over their program as their program is inadequate and probable set up incorrectly. If you have the right technical people in-house then the CSI system will provide you with an adequate program IMHO. However is not my top pick even though I use it in-house now and have very good results. Having said that, you can probably find fault with any system to some degree so I'm not really bad-mouthing it. The problem you have is not with the CSI equipment but people. In addition, I don't think adding shock pulse or spike energy will add much; it certainly will not take the place of a good PdM programme.

I think any of the technologies for early detection of bearing faults (such as Peak View, Shock Pulse or demod) will produce similar results in the hands of a capable analyst; further more I doubt most folks in the VA trade will make a call (for bearing change out) based on Peak View or Shock Pulse only if there are no fault indications in the velocity and/or accel data.
From the faults you mention it may be worth looking at maintenance and QC practices/procedures for bearing installation re correct fit/clearances.

quote:

Originally posted by Sam Pickens:
It appears to me that your outside contractor is not up to snuff. They should not miss this type stuff; so, that excludes phasing their service in and taking over their program as their program is inadequate and probable set up incorrectly. If you have the right technical people in-house then the CSI system will provide you with an adequate program IMHO. However is not my top pick even though I use it in-house now and have very good results. Having said that, you can probably find fault with any system to some degree so I'm not really bad-mouthing it. The problem you have is not with the CSI equipment but people. In addition, I don't think adding shock pulse or spike energy will add much; it certainly will not take the place of a good PdM programme.

What would be your top pick, or top 3?

The brand of equipment being used is not the key factor in your program. The quality of the person operating it and analyzing the data is the most important aspect of your program. Any good analyst can make any vendors equipment do the job correctly.

quote:

Originally posted by Xracer:
Correct me if I am wrong, but with shock pulse method, it detects ball to race defects but would not detect a loose race fit?

I agree. It would seem to me that most methods will have a fairly easy time detecting and diagnosing ball and race defects because the patterns show up as discrete nonsync peaks in predictable patterns (because the impacting is fairly periodic). The motion for a loose fit will be much more chaotic and random... so you won't likely see any discrete frequencies in either regular spectrum or demod spectrum. You will see a raised floor in regular spectrum and that can have many possible sources. Likewise you may see somewhat random impacting in the TWF. These symtpoms might be mistaken for poor lubrication or cavitation or etc.

Xracer,

The following setup examples are used with CSI hardware and software but the fundamental princicples can be applied to any vendor hardware or software.

Start by dividing your machines into three catagories, (1200 - 1800 rpm), 3600 rpm & (>3600 rpm higher speed machines). These machines will basicly have the same analysis parameters but alarm parameters will be specific to speed.

Vibration spectrum and waveform units and configuration:
Spectrum and waveform unit: Velocity Pk
Overall: Digital overall (Spectrum Velocity overall from 3 to 1200Hz)
High pass filter: 3Hz
Fmax or low pass filter: 1200 Hz
Sample size to save: 2048



Analysis Paramters:
VHFD: Variable high frequency detection 300Hz to 5000Hz
Subsynchronous: .3 to .6 order (I prefere to use Hz)
1x: 1 order (20hz, 30hz, 60hz etc)
2x: 2 order (40hz, 60Hz, 120hz etc)
3X: 3 order (60Hz, 90Hz, 180Hz etc)
4x - 20X: 4x to 1200Hz ( spectral energy from 4x to FMAX)
Crest factor: peak in the waveform divided by the overall value
waveform Pk: waveform pk



alarm parameter
Overall: .3 ips pk alarm & .5 ips pk danger (slightly less for slower machines)
VHFD: 3600rpm(4 alarm, 7 danger) 1200rpm( 3 g's alarm , 5 g's danger)
subsynchronous: 3600 rpm (.1 ips pk alarm .2 danger)
waveform pk: 3600 rpm (.75 ips pk alarm, 1.0 ips pk danger)
I can't remember the other parameter alarm setting off the top of my head but take a look at your machines and set alarms to give an indication of change.

This will probably start another topic but I do not use demod or peakvue. I use the VHFD to trigger an alarm then I collect an addition waveform and spectrum using the VHFD parameters 300Hz to 5000Hz and 3200 Lines of resolution. Then I manually demod or peakvue the waveform, stripping out frequencies and lookng for them in the spectrum.

a 3600 rpm machine with a bad bearing of 7 to 12 g's VHFD will show 1x and 2x bearing frequency in the route velocity spectrum.

I agree that if your current service provider is missing these types of failures, then they are not up to snuff. This is just my opinion, but the bigger the firm (in terms of number of people employed), the less likely they are to do a good job for you. Experienced analysts are 80% of the solution because an experienced analyst can do a good job using just about anybody's equipment. The type of equipment used is only 20% of the solution. The reason larger firms are less effective (usually) is purely economic... as a firm grows there is more and more overhead, and more people on the payroll who are not productive in the sense they are not doing "billable" work. Usually there are several owners who expect to make a fair amount of money because it's their money invested and at risk. Experienced analysts are expensive to hire and retain (supply and demand) so the tendency is to hire less experienced people and train them, often "on-the-job", and perhaps on your job. Being less experienced, they tend to call everything that is not normal (excessive repairs) or they miss things (expensive failures). Experience counts.

The smaller firm, maybe even a lone consultant, will not last long making lots of mistakes. With lower overheads and more control over data collection, analysis, and reporting the "little guy" will usually do a better job, IF he is big enough for the task in terms of available manhours. Don't be afraid of the little guy... if he's any good, every customer he has will be glad to discuss his perormance with you.

The above perhaps flys in the face of popular thinking, but the larger firms generally get a contract because their "higher ups" are talking to your higher ups and that's where the deal is done, often based on a good job of "selling."

As for "shock pulse"... 80% of the time you will get a "bad" call on a bearing long before it is likely to fail. Often if removed, there will be little physical damage.

With "vibration analysis" many more things come into play, as others have mentioned. You get a better, bigger picture of what's going on with the machine, better diagnostic tools, better root cause analysis capability.

Remember, you can get PdM results fast, cheap, and right.... which 2 of those do you want. Another way of looking at it is, you can spend money on the program, or you can invest money in the program. Usually spending money gets you a "feel good" moment which produces little lasting value, whereas with investing your money, you get results, but usually later, down the road.

Doing the program in-house is always the best way to go for a large facility like yours, IF management understands it, supports it, and funds it; IF you have people available to do the work who will be committed and willing to do the "hot and dirty" work (if your present consultant is not the dirtiest guy leaving the plant at the end of the day, he's not doing his job); IF there is an effective accountability system for measuring results (whoever is doing the work).

With

quote:

A rough est would be 40,000 pieces of equipment

, how many people is the contractor sending in to do the job? How often is he doing the analyst? Is he analyzing all the data or alarming and picking all the high alarms to make his calls from or doing "90%" of his analysis in the field while collecting data or what is he actually doing?

I know these questions are not answering your original question, so tothe question: CSI, Commtest, Ect. are excellent equipment and shock pulse is also, but as some have said, the equipment is only as good as the person using it. Getting familiar with a new piece of equipment to collect the data is far fetched from rocket science, but learning the analysis part of the program is where lack of experience and on site training is the only hold back for a new person.

If your company is wanting to start an inhouse program, the initial cost of setting it up, equipment wise, is going to be high, but my guess would be that 3 months worth of contractor costs would cover it, what with 40,000 pieces of equipment.

I started in vibration analysis at a plant who started an inhouse program and we had a rough go for the first year, due mainly to the fact that we never had a skilled person come in to give on site training and database setups that fitted our plant. We had to build and adjust as we went along. I have since seen some of the databases setup by the equipment manufature and most are a far cry from what is actually needed for some of the plants. Using generic setups to try and cover all plants and pieces of equipment really is, in my opinion, not the way to go. Each plant has its own needs.

I am not sure what your plant products are, but getting someone in to help the new guys start out on the correct foot and direction is HIHGLY important. Like someone has already said, the larger a contractor is the more difficult it is for them to be able to provide "one on one" training at an affordable cost.

No matter what equipment you choose, you need to put people on it who are at a minimum of 110% dedicated to do the job and someone to help setup the databases who has the "time" to do it and not be in a big hurry to get to the next job. Someone who explains things, including how and why he makes a call or setups up an AP Set the way he does.

On site training, using your equipment and analysis tools is, in my opinion, the most sound way to start a new program or even improving an older program.

Sorry for being so "long winded". I know it is "hard" to remain focused on a long worded reply, so, I apoligize.

Originally posted by Xracer:
Correct me if I am wrong, but with shock pulse method, it detects ball to race defects but would not detect a loose

I might mention that I find pretty distinct harmonic amplitudes in a vibration spectrum when a bearing inner race is loose on a shaft. I think that the 4th harmonic is dominant if it is the outer raceway (not sure), but always there are harmonics in the spectrum with an "interfacing fit" such as bearing shaft inner race or collars are machining the surface of the shaft, also looseness in the housing and outer raceway surface. Some bearings come with a pin through the casting to the outer race, Sealmaster I believe carries these type of bearings.
A FFT spectrum analyser recorder would pinpoint the looseness problem also a stobe might help in identifying machine component looseness.