I've got dibs on that first route Danny
You can have the next one.

I forgot while I was writing my epistle above that I always question the one I'm working for about what was found on the machines I write up.
Once in a while I will get to see the part, most of the time not.
I put notes in my database on the trend line about findings.

Dave

Lots of information and different approaches. Here is mine. 1000 bearings with 4 bearings per machine (typical) is 250 machines (a machine is the motor/pump or motor/fan combination). I will collect a horizontal and vertical reading on each bearing, plus one axial reading per shaft (one axial per bearing on overhung fans). So that's 10 pts per machine, or 2500 points total. Depending on how the machine is constructed, some problems don't show up at all in the horizontal direction, or not at all in the vertical direction. So you really need horizontal and vertical readings on each bearing for the best results.

It will take a good analyst 2 - 3 days to collect good data on 2500 points, plus one day for analysis and report prep. Travel is a HUGE variable with the cost of gas these days. All things considered, for 2500 points, $5000 would be the low end I think, with $7500 toward the upper end. I would recommend monthly data for at least the first 3 months, and then possibly every other month after that. For bi-monthly surveys, I add 25% to the 'monthly' charge, and for quarterly surveys (not recommended) I add 50% because the less frequent you do the survey, the lower your margin for error is, so the more time you spend collecting data (including "extra" data in the filed) and doing the analysis. You just have to be more careful which takes longer.

Of course, you could just pay "time & material" but I think that will end up costing your more in the long run (and it makes management nervous). Most folks will 'under estimate the time required to do a survey, and so will bid cheaper than an actual "time and material" survey would cost.

Rusty,

According to your data:
You spend per machine on average:
- for data collection 6 min
- for data analysis 480/250 = 2min ( including reporting)

I spend per machine on average:
- for data collection 6 min
- for data analysis 18 min ( including reporting)

I can believe it only if you wholeheartedly trust your alarms.

Dave

I can believe it only if you wholeheartedly trust your alarms.

Why would you not trust your alarms?

For 250 machines (1000 bearings) I would give it 2 days of data collection. Still 1 day of reporting, as long as they are typical pump, motor, fan, etc.

3 days + travel. $3000.00 is probably a close figure.

Dave

Dave_man, 18 mins. per machine for analysis? For a 'typical' pump or fan? How long have you been in the business? I ask that, because if you take 24 minutes per machine (6+18=24), it's going to take you 100 HOURS to do those 250 machines! That's 2.5 WEEKS, assuming you don't do anything else. At $100/hr, that's $10,000 with no travel or living expenses.

I actually don't use 'alarms' per se. If I don't sense (feel, hear, smell) something is wrong with the machine when I collect the data, or see something in the spectrum or waveform, or there isn't a significant difference from the last overall level, then there is nothing to analyze. If anything "triggers" a reaction in my brain, then I enter a "note code" and/or take "extra data" (with a note code saying so). Back at the office, I work all items from the previous report first, and then print the note codes and work all those items.

To me, looking at (analyzing) data on "good" machines just creates cognitive "noise" in my brain and is a waste of time (usually).

I am a mechanical engineer, but probably not your 'typical' engineer in that I take a very intuitive approach to vibration analysis, and put a lot of trust in my instincts and experience (20 years of being in the field, and around two million points of collected data). I don't obsess over "missing" a call because I don't think that represents the best value for my customers. I spend a lot of time talking to folks in the plant, asking them about the machines and what's going on with them.

But that's just my time-proven approach, and it works well for me and my clients. Might not work for others depending on the industry, the equipment, the people involved, and the personality of the analyst.

I alway recommend you do what works for you, and do not worry about how "others" are doing it. The bottom line is results, and keeping the customer happy (as much as possible).

But hey, that's just my opinion.

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

I just can’t help to jump in. As an in-house guy, I just have to share my opinion on one missed point. From my experience, any vibration program, Pdm, Cbm, Rbm, etc start by a decision from management and that decision as to be supported all the steps down to the mechanics that do the actual works. Unless you do have a cheerleader to prove, educate and sells the program to the individuals (manager, foreman, planner, etc) you are domed to fail! Pdm is like a marriage, you need to commit, to remind yourself that being together makes you better and prove it! Having someone inside that is dedicated to the best interest of the company makes the in-house analyst a strong partner for a successful Pdm program that has the financial return as long as he is supported by the management. What is it worth? Another point is that most people I worked with, means well and are true professional (contracted or in-house) but the in-house guy have the advantage to work from the inside (he knows personally every individual) and help to improve every step of the maintenance process which helps to achieved the real goal…. Save money!
Best regard, Marcel

I look at every point and sometimes spend only a few minutes on a machine but on others maybe 2 hrs or a day or longer. I do not trust instinct on anything and doing this 41 yrs as of Aug 7th.

And regarding the program: decisions are handed down - people will 'buy-in' when they see results and that you are calling the shots right.

An example: went back into a plant just a few hours prior to start-up. Noticed a pump had not been replaced I wrote up for replacement. Asked why; the mechanic/MW told me it felt good. I said I know it felt smooth and seemed OK but it isn't. Saw the manager and told him if they started the pump would fail in hours. We went to the shop saw the MW and I explained the bearings were failing. The manager said, you're sure? YES!!! We replaced the pump and immediately tore into it ---- bad to everyone's eye with no room for doubt. This had happened a number of times in a number of locations but made believers out of everyone. Afterward the report is issued and followed to achieve planned/scheduled maintenance. It does work but few buy-in in the initial stages. It only appears that you're walking on water; a few inches below are alligators and you're walking on their backs - don't fall in -- only takes one slip to ruin a 1,000 ataboy's.

Tom,

200 Pumps! What do they pump(water, fuel, oil, sludge etc)?

Sounds like you have a couple of very expierenced people watching this tread with great interest. You may find the right person for the job here if you are not carefull.

Good luck to all.

Hooch

quote:

I do not trust instinct on anything

I don't make a call or recommendation based on instinct either. But if I don't sense something is wrong, or see something in the data I'm collecting -- if the machine appears to be "normal" based on the totality of the evidence -- then it's logical to assume it's normal and there are no problems. Of course, I am taking the data and doing all the analysis (I am a one-man shop). If I were analyzing someone else's data, it'd be a whole different deal. Truthfully, I don't care who you are, I'm not going to trust your data enough to make a call on a machine. That's really why I work alone I guess.

Case history: Flash Tower Bottoms Pump 3600 RPM AF brgs.

So smooth like touching you desk. Much smoother than normal. Analyzer (RTA Ono Sokki) was getting excited. I ran back to the machine and yes I remembered right - very smooth. MW's nearby so I ask them to feel - they said smooth. Back to the analyzer - bad! Changed transducers and checked cables. Felt good, looked bad; only one other thing like that. Anyway, on a machine everyone thought was good I went to the manager and said, it's failing as of this moment and won't last a day. That's trusting your data. We shut it down; I felt the brgs/shaft good. Rolled the shaft; could feel nothing bad. Pulled the plug and oil had metal in it. Pulled the brg cover and one balll missing and no cage - yet the thing felt good. Couldn't feel a bump or blip. I trust my data over all.

Many times I have people come and tell me, this one is bad - but I look at the data and say run it. Or, I say one is bad and they'll say, it seems to run fine will little or no vibration. I like using high quality transducers and low noise micro-dot cables still. If using a RG-58U or 1/4" coil I like to hold the cable near the transducer and remain motionless as not to strain the transducer. I even worry about the micro-dot cables. Never assume anything!!!!!!

Hi Rusty,

I agree with your methods...I do the vibration analysis for my company, and after 5 years of looking at the same equipment month after month, I have pretty much learned all their personalities, so to speak...I also listen, feel, and smell my way through those, so I can pretty much sense if and when something changes.

I by all means as you say, focus on the ones trending up..

Good assessment...Have a good one...

quote:

Originally posted by rustythevibeguy:
Dave_man, 18 mins. per machine for analysis? For a 'typical' pump or fan? How long have you been in the business? I ask that, because if you take 24 minutes per machine (6+18=24), it's going to take you 100 HOURS to do those 250 machines! That's 2.5 WEEKS, assuming you don't do anything else. At $100/hr, that's $10,000 with no travel or living expenses.

I actually don't use 'alarms' per se. If I don't sense (feel, hear, smell) something is wrong with the machine when I collect the data, or see something in the spectrum or waveform, or there isn't a significant difference from the last overall level, then there is nothing to analyze. If anything "triggers" a reaction in my brain, then I enter a "note code" and/or take "extra data" (with a note code saying so). Back at the office, I work all items from the previous report first, and then print the note codes and work all those items.

To me, looking at (analyzing) data on "good" machines just creates cognitive "noise" in my brain and is a waste of time (usually).

I am a mechanical engineer, but probably not your 'typical' engineer in that I take a very intuitive approach to vibration analysis, and put a lot of trust in my instincts and experience (20 years of being in the field, and around two million points of collected data). I don't obsess over "missing" a call because I don't think that represents the best value for my customers. I spend a lot of time talking to folks in the plant, asking them about the machines and what's going on with them.

But that's just my time-proven approach, and it works well for me and my clients. Might not work for others depending on the industry, the equipment, the people involved, and the personality of the analyst.

I alway recommend you do what works for you, and do not worry about how "others" are doing it. The bottom line is results, and keeping the customer happy (as much as possible).

But hey, that's just my opinion.

[QUOTE]Originally posted by Sam Pickens:
Case history: Flash Tower Bottoms Pump 3600 RPM AF brgs.

So smooth like touching you desk. Much smoother than normal. Analyzer (RTA Ono Sokki) was getting excited. I ran back to the machine and yes I remembered right - very smooth. MW's nearby so I ask them to feel - they said smooth. Back to the analyzer - bad! Changed transducers and checked cables. Felt good, looked bad; only one other thing like that. Anyway, on a machine everyone thought was good I went to the manager and said, it's failing as of this moment and won't last a day. That's trusting your data. We shut it down; I felt the brgs/shaft good. Rolled the shaft; could feel nothing bad. Pulled the plug and oil had metal in it. Pulled the brg cover and one balll missing and no cage - yet the thing felt good. Couldn't feel a bump or blip. I trust my data over all. ...QUOTE]

Sam - Out of curiousity, what kind of vibration measurements are you collecting that will indicate a failure within one day without any noticeable roughness (i.e. feeling the component)?

quote:

Pulled the brg cover and one balll missing and no cage - yet the thing felt good. Couldn't feel a bump or blip. I trust my data over all.

This reminds me of a bearing on a felt roll in a paper machine. Spectrum looked good, felt smooth, nothing seemed out of the ordinary, except in the waveform. I could see what appeared to be the cage frequency but could not see it in the spectrum. Cover was removed and there was no cage. All the rollers were jammed up together (side by side) rolling freely without a cage.

How can the absence of a cage give a FTF?

quote:

How can the absence of a cage give a FTF?

With or without a cage, the balls/rolls still travel at the same rate as long as no skidding takes place. I believe the purpose of the cage is to reduce friction and spread the load by providing uniform spacing of balls/rollers.

Walt

On every point with a rolling-element bearing my standard setup is to take a spectrum in velocity, BUT a waveform in g's at least out to 4kHz ("special time waveform" in CSI's meters). I quite often find "issues" with a bearing when it feels OK and the spectrum looks OK (almost.... usually there will be 'something' there if I look close enough). If you are not looking at an acceleration waveform when you collect your data, you are really "flying blind" in my opinion.

But Sam, as someone else asked, what "data" did you see when you were doing your point-by-point analysis that flagged a problem on this machine?

I believe it was an FFT rather than an FTF or maybe I had a typo somewhere.

On general machines I take cap data using accelerometers (hi-quality w/hi-quality cables). Use extreme care in placing the transducer and try to have no cable wiggle or movement with good mount. Since I use accelerometers I collect data in acceleration (do not integrate to velocity or double integrate to displacement). Those data are FFT's and TWF and whatever I deem necessary and/or adequate. So the data I see is machine components defective or.......... I have never been able to do this using velocity through integration. Integration of acceleration to velocity is not absolute; only relative.

More cake; some good icing - on a 1000 HP unit in Carrollton, GA at Southwire. Their in-house program (two tech units; one using IRD and the other using ???CSI or SKF???). They had been on the motor like stink on a skunk. I was call-in and did the analyses. Thursday evening as I recall. I asked if they could shut-down and repair on SAT as the AF bearings were going to fail. They said, "why do we not see this in our data as we've been doing it dailey for two weeks?" We stepped into my van and I showed them the spectra and such data as I had collected. They said, we believe you know but; are you serious that the machine will fail. I said, YES! This is a true story and something I probably would never do again - they said "when or what day will it fail on". I said Monday. They said, can you be more specific? I said Monday afternoon. They opted not to shut-down. I was called TUE and told the machine failed at 1:30 PM Monday. Lucky - you betya! But also based on what I was seeing I called it the best I could and really believed it would fail by then if not on the previous SAT or SUN. They say a blind hog will find an acorn every once in a while.

~1980 I started using accelerometers for acquiring data. 1983 used accels exclusively for cap data doing PdM programs and in that year I quit using velocity or integrating to velocity.

Many use velocity successfully and have great success stories. However it has been my experience that the acceleration based program working in acceleration has worked for me really good. I was teaching my alignment seminar at Macon Tech college in Macon, GA. In that class a client had sent two of the maintenance guys. During the second day one spoke-up and said, "this guy does our PdM vibration program and we have not had a failure in over five years and have worked on OT during that period".

When I started the program a number of years earlier (7+) I was seeing failures and would go out and find that the techs were out all night working on a break-down as they were afraid to be without a back-up unit. MTBF was 1 to 1.5 yrs on average. Now a number of years even later we have MTBF ~5-7 yrs and planned/scheduled maintenance.

If you can get lucky enough to call some good shots early in the program and have the people buy-in through confidence and also talking shop and sharing your knowledge so they know you can 'do' as well as 'talk about it'; you get cooperation and teamwork.

Let's face it; we collect data and write a report and mostly do nothing (we don't work on the machines to rebuild them or not generally). By nothing I mean our action of doing our job did not physically do anything mechanical or make corrections. That only comes by teamwork and confidence.

I wish all stories were good; but, they are not. Not all work out this good nor are all people wanting a program to function - lip service is enough. Sometimes you have to interface with managers that have a run-to-fail philosophy and don't even read your reports. I had one such that I would get calls and he'd ask, "what did you say about such machine?" I'd look it up and see that the bearings were going to fail and I had given them a time for repair by this date. I'd get "oh!" and nothing until they would kill the program - no one saw the reports and the manager didn't read them. There is frustration in the job. I can tell you those stories as well. But out plight is to keep on keeping on and not get discouraged.

Yes, with the accelerometer readings you can see the visible impacting of the balls/rollers, in a timewave form, even pinpoint the damage to the cage inner or outer race...I also look at the amplitudes of the HGE3, if it's way up there like 20g's or better, it's time to shut it down. I've seen blower bearings bet as high as 70g's plus in extreme cases. Not good considering the red alarms we set here are at 30g's

Point being you cannot rely on velocity reading alone. I've heard there are some out there that do. Velocity is good for lower frequency vibration on the FFT scale

quote:

Originally posted by Sam Pickens:
I believe it was an FFT rather than an FTF or maybe I had a typo somewhere.

On general machines I take cap data using accelerometers (hi-quality w/hi-quality cables). Use extreme care in placing the transducer and try to have no cable wiggle or movement with good mount. Since I use accelerometers I collect data in acceleration (do not integrate to velocity or double integrate to displacement). Those data are FFT's and TWF and whatever I deem necessary and/or adequate. So the data I see is machine components defective or.......... I have never been able to do this using velocity through integration. Integration of acceleration to velocity is not absolute; only relative.

More cake; some good icing - on a 1000 HP unit in Carrollton, GA at Southwire. Their in-house program (two tech units; one using IRD and the other using ???CSI or SKF???). They had been on the motor like stink on a skunk. I was call-in and did the analyses. Thursday evening as I recall. I asked if they could shut-down and repair on SAT as the AF bearings were going to fail. They said, "why do we not see this in our data as we've been doing it dailey for two weeks?" We stepped into my van and I showed them the spectra and such data as I had collected. They said, we believe you know but; are you serious that the machine will fail. I said, YES! This is a true story and something I probably would never do again - they said "when or what day will it fail on". I said Monday. They said, can you be more specific? I said Monday afternoon. They opted not to shut-down. I was called TUE and told the machine failed at 1:30 PM Monday. Lucky - you betya! But also based on what I was seeing I called it the best I could and really believed it would fail by then if not on the previous SAT or SUN. They say a blind hog will find an acorn every once in a while.

Hi Sam,

I really appreciate your confidence when you said that the bearing fail in the Monday afternoon. Itz a proud moment for any Vibration Analyst to specify that precise.

Blower bearings are a problem Rod...It sometime will not fail even we sent a danger report.. Hahaha.. I sent a report of a blower bearing at 70g yesterday. Still waiting!!

Have a nice day!!