Maybe stirring up the hornets once more but anyway, dynamic and static forces must be carried by the bearings, increased dynamic forces will increase movement of the bearings and detected by absolute bearing transducers. I prefere to use relative eddy probes to monitor shaft bending and then as far as possible from the bearing as the bearing if designed and working properly would be a node for shaft bending. So monitoring in the classical way inside bearings with eddy probes in my view generate very small signal, close to the noise level of the system to be monitored and give difficulties to define alarm levels defined to be able to shut down the machine w/o false alarms. So go for monitoring absolute on the bearings for good system operation and most likely a lower cost. Olov

Keep in mind that you can monitor for impacts and rubs within the bearings using envelope detection methods with accelerometers.

Measuring both the relative and the seismic is important for many machines. I don't think both are required for most large slow fans though.

I would stick with a velocity transducer on the housing for most of these fans. As some of the others stated the bearing supports for these machines are generally flexible. In this case you might get more useful data without confusing the issue by avoiding the whole shaft relative vs. seismic vs. absolute shaft vibration discussion.

In the worst case, you could get very low shaft relative movement and a lot of housing movement. I see this often on turbine generators that only have relative transducers. The customer is smiling due to low readings on their color monitor in the control room and then they walk around their LP's and generators and get their fillings rattled out by the seismic that some genius thought was not important to measure. This might not be the case for your specific fan, but maybe you could take some shaft absolute data with a fishtail to compare to the housing reading. Then you can subtract the data vectorially (at least the 1X) to get a calculated relative vibration value.

I think prox probes are important and useful for many machines, but many tend to view them as the ultimate vibration parameter. This opinion is just not based on facts. In many cases a proper shaft absolute measurement (via shaft rider or via combining relative and seismic) is far superior to a shaft relative measurement.

Also, if the bearing is located at a node, wouldn't the displacement and therefore the damping be essentially zero. I have been taught that locating the bearings away from the nodes is required to obtain adequate damping. I'm not a machine designer though, and I'm sure there are many other considerations.

Michael Titone

quote:

as the bearing if designed and working properly would be a node for shaft bending

Not true for any bearing that I would design on a flexible shaft - at least on purpose. This would be a bad design to have a node at the bearing.

quote:

So monitoring in the classical way inside bearings with eddy probes in my view generate very small signal, close to the noise level of the system to be monitored and give difficulties to define alarm levels defined to be able to shut down the machine w/o false alarms.

Interested to know what you monitor that behaves like this.

One issue that can occur if not careful is placing the transducers away from the bearings but at a node point. Assuming the shaft in question is relatively rigid, this should not be the case at both ends, unless there is 0 relative motion.

When specifying the monitoring it is often called out to avoid measuring at nodes.

One of the vendors that visits here, http://www.reliabilitydirect.com/appnotes/cvr.html, has a quote "Journal or Sleeve Bearings are designed so that the oil film provides damping. The shaft is free to vibrate within the bearing. Due to the damping provided by the oil film very little of the shaft vibration is transmitted to the bearing cap." Not always true, but it can be the case. It's not really teh damping that does it, because damping generates forces that get transmitted.

How well would you see oil whirl through the casing transducer? Difficult to tell without measuring directly and comparing. How could you get oil whirl with a proper design? Wear, if the bearing clearance opens it may be possible to obtain whirl in your system. That may be ok, because in the long run (perhaps not too long) the machine may shut down for a maintenance item or two.

If the supports are flexible then the rotor system can be close to or above a resonance.

You can look at the oil film bearing coefficients (Dodge can generate these, or you can go to a consultant.). What does the dynamics look like with the supports, bearings, and rotor? Where are the modes, and how does forced response look?

Another good idea was too take measurements on the existing machinery. I hesitate to recommend fish tails because of safety concerns ... but then I am not chairman of the board. Temporary shaft probes would work; although, I would expect some excessive runout on an unprepared surface.

This message has been edited. Last edited by: William_C._Foiles,

You are correct, EP
I am going to bow out of this thread. I know that Bill, coming from petro chem has a very important need for non-contact probes, but my history in the field (34 years) has exposed me to more than a few individuals (I am not accusing you yet, Bill) of having orbits branded on their foreheads.
Bill, how did life exist before Don Bentley? An Atlas Copco compressor has pinions and compressor wheels that weigh 1 pound turning at 80,000 rpm in housings that weigh several tons. Without a non-contact probe you are toast.
I have used shaft sticks (yes, I am that old), and also have taken the output from the non-contact probe outputs and compared them to my acceleration data. I have always had a spectrum analyzer. Early on it was a real time compression analyzer (Federal Scientific UA500).
I learned early on that a non-contact probe was WORTHLESS above the seventh harmonic of running speed. The dynamic range is gone. So, if the sleeve bearing machine has ANYTHING that can generate vibration above that harmonic, gears, blade pass, etc., forget it. You can't measure it.
I have always given the proximity probe it's due where it is essential, low rotor weight to high housing weight, rotor dynamics analysis for bearing design issues, high speed turbo machinery etc.
But, please Bill. Don't push it here for 700 rpm fans. You sound like a Bently salesman.

quote:

Bill, how did life exist before Don Bentley?

Before Don Bently there were shaft riders (and still are), which are still used. There's an option??? Of course, Bently did not invent the shaft relative transducer; they were around prior to him.

7 times 700 cpm is 4900 cpm; I think prox probes can handle that if they can handle vibraton on your 80,000 rpm machine. Will there be much blade pass on the rotor? The damping provided by the bearings may reduce the shaft relative measurement. The probes are certainly not WORTHLESS above the 7th harmonic of these machines running at 900 rpm and 700 rpm.

You still find turbine manufactures using either shaft relative or shaft riders on their turbines. Shaft riders have severe frequency limitations along with basic mechanical issues. These machines are not light rotors by most standards. In many cases the supports are relatively stiff.

Do you detect many failures using blade pass, especially for monitoring and protection?

No one considers temperature measurements?

I don't think the quote from ReliabilityDirect came from GE Bently.

Bearing stiffness depends on speed for a given bearing design. With lower stiffness at the lower speeds, one can have large relative displacements depending upon support stiffness.

Not even curious enough to look and see?

Why do you need to balance to 0.2 mils on the casing? Is this overkill, or could there be significant shaft relative vibration that concerns you? There would be no reason to justify this type of level on shaft relative vibration, because you have a measure of the forces at the bearings. A possible exection would be support resonances; in which case one should correct that situation.

How do balance sensitivites compare to rigid rotor balance tolerences?

This is my Friday off; so I probably have no more to say. This is Rusty's consulting job. If it were mine, I suppose we would do it differently. Temperature would be included, probably shaft relative vibration before casing measurement, and I would decide if this is a run to failure machine. Portable measurements occasionally if this is a low consequence, low failure, low cost machine perhaps.

I've seen a few shaft probes on fans, but only a few since many fans have rolling element bearings. I think the ones that I have worked with were justified.

quote:

You sound like a Bently salesman.

Does one of us have a marketing degree? Perhaps, but it isn't me.

This message has been edited. Last edited by: William_C._Foiles,

I think part of the problem is that some folks have no idea what kind of fan we are talking about here.... think about it... a 10,000 rotor is pretty massive, and the bearing supports are very stiff. This is a rigid rotor machine at 900 rpm. It runs well below the 1st critical, with a 'lag' of only 25 degs. (would be 90 degs. at the 1st critical).

"Portable measurements occasionally if this is a low consequence, low failure, low cost machine perhaps." These are decidedly NOT low consequence, low cost machines. They ARE low failure... I've been monitoring these machines monthly now for about 4 years and we have never lost a fan, wiped a bearing, or even had a motor bearing fail in service. We haven't had any incidents at all. By comparison, a sister plant has similar fans that have continuous, remote diagnostics performed on them by "experts" and they've had all kinds of problems with their fans.

Instrumentation doesn't make equipment reliable. Do we really want state-of-the-art instrumentation just so we can say "it's about to fail" or "it just failed." What kind of sense does that make? The real value that is added by the lowly, walk-around PdM consultant is he gets to know the machines, understands how they are operated, makes recommendations on maintenance and repairs, and follows up on the machines.

My opinion, based on 20+ years of experience with fans like this, is you don't need prox probes on them, period. It's just not necessary. Oh, I guess if you only see one of these every once in awhile, then maybe you need prox probes (and whatever else you can drag out to the machine).

If a person doesn't have experience with a particular type of machine, then he knows little about it really, despite his training, education, pedigree, or resume. I wouldn't think of jumping into a discussion of gas turbines, or high speed compressors, or lots of other stuff, because I have little experience with some types of equipment.

Some folks obviously have little experience with ordinary, run-of-the-mill industrial equipment, which as I've said before makes up about 99% of what is out there.

Thanks Ron, Michael, OLI, & Mike (to name a few) for your insightful, experience-based input.

Rusty,

I monitor several bag house fans and FD fans that are in the 35K# range running in the 900 RPM range. Some are on Cooper splits, and others are in babbitt (sleeve) bearings. There are no prox probes on any of them. With the exception of those fans that are plumb worn out (holes in leading edges), we have not missed any problems with them. Cleaning usually solves their slow growing problems (again, with the exceptions of those with holes in the blades), and the Coopers can be seen to be failing way before they are in much distress.
I don't think it is necessary to have prox probes on them if you are doing a regular wald around.

Dave

"If a person doesn't have experience with a particular type of machine, then he knows little about it really, despite his training, education, pedigree, or resume. I wouldn't think of jumping into a discussion of gas turbines, or high speed compressors, or lots of other stuff, because I have little experience with some types of equipment.

Some folks obviously have little experience with ordinary, run-of-the-mill industrial equipment, which as I've said before makes up about 99% of what is out there."

I am a bit curious who you are referring to Rusty...and if you are implying that those of us that don't gather data on your style of fan everyday don't have anything to contribute, then you are flat out wrong. But if that is how you feel, then I won't bother reading your posts anymore.

As for this particular machine type I have some experience with them (although I don't gather data on them every day), and while I agree that "you don't need prox probes on them, period" that does not mean that prox probes are useless on this type of machine. I still would like to have them, but it would not impair my ability to diagnose this machine if they were not there. I am a prox probe guy, I like them, and so I use them. To each his own.

Youre opening question was in regard to the usefulness of prox probes on this style of fan. If you had already made up your mind that "you don't need prox probes on them, period" then why did you bother with the post? Just tell your customer not to expend the extra money

quote:

Originally posted by Ron Brook:
exposed me to more than a few individuals (I am not accusing you yet, Bill) of having orbits branded on their foreheads.

I also am a prox probe guy but I do not have an orbit on my forehead, it's underneath the hair behind my ear and there are three orbits together like the sign of the DEVIL .

As Steve mentions, prox probes are not necessary but they can offer data that is impossible to gather with case readings. The same is valid for case readings, there are some things that cannot be resolved with prox probes. But if you have both you will be able to solve any vibration problem and understand your machine's dynamics better. This is why I collect periodic case readings on machines that are fully instrumented. Sometimes I have to look at case, shaft, temp and process data to make a call.

quote:

one of the vendors is proposing using a pair of prox probes at each bearing. While this is certainly desireable, there will no doubt be extra expense involved. ...there is some question as to whether or not they are needed...

We don't really have any problems with the current fans. We balance them to 0.5 mils (pk-pk) at the end of each shutdown (2x per year) and they are cleaned when they reach 4-5 mils. ...

My question to you guys is, do you think the X-Y probes are needed? Is the extra expense justified? Personally, I'd be fine with them as they'd give me a nice platform for "research" (the only machines I monitor with probes is Centacs, and they won't let me 'play' with them). But I question if they will be worth it, and if they will just confuse my client. But what do you think?

Did you really think that they were desirable at some point? With only Centacs having prox probes, it didn't and doesn't sound like you have much experience with shaft relative measurements, and you may be the person confused by the information. It's probably good that they don't let you play with the Centacs. Some of the old machines had monitors that didn't have isolation between the input and the signal; this has been know to cause shutdowns, but perhaps they have updated these in Arkansas, or the machines aren't that old.

Given your knowledge, you are probably correct the probes may be of little use to you, especially if it is just a research platform. I assume the customer is paying for expertise in this matter. It doesn't look like you really wanted the advise of anyone that has used shaft relative measurements.

quote:

think about it... a 10,000 rotor is pretty massive, and the bearing supports are very stiff. This is a rigid rotor machine at 900 rpm.

Exactly, stiff supports (if you truly mean that) with fluid film bearing at low speed – a perfect scenario for shaft relative measurements.

Now to the reason for my writing,

quote:

Some folks obviously have little experience with ordinary, run-of-the-mill industrial equipment,

Was this aimed at me? If it was, I assume that you have my resume, because otherwise this statement was made out of ignorance. My most recent exposure to run-of-the-mill equipment has been having the responsibility for specifications for equipment with 800 to 1000+ motors, hp from 50 (or below) to 25,000; most will be around 75 hp. OK, 25 000 HP equipment may not be run-or-the-mill. Are the 4,500 HP units?

I have taken measurements in sewage treatment plants, building to nuclear HVACs (including some at a Star Wars facility), high speed equipment at Sandia labs, \fin fans (I suppose our industry has a few more of these compared to your ordinary industrial facility.), cooling towers, pumps from nuclear to chemical and some in between, paper machines (long time ago), coal mine fans, a quarry fan (really just needed cleaning, but they wanted a balance), and I suppose a few more that I don't recall like gas turbines (large frame units, aero, light industrial – I've even balanced engines on the wing of aircraft.), steam turbines (nuclear, fossil, process, waste heat, combined cycle), generators, motors, minor involvement with potato chips (Althought in all honesty I my main experience is in having eaten too many. I trust this is in keeping with your copy of my CV!. I've also looked at equipment from the tropics to the artic, from the desert to offshore locations – almost as diverse as Arkansas – did I mention valleys and mountains?

OK, I guess I crossed the line with my post above. Bill, I feel like you take pot shots at most everything I say, even if my post or reply has nothing to do with anything you've said. I feel like you sort of look down your nose at me. I know I pop off sometimes, but not more than some others. Yet, you are the only person here I've ever had a problem with, or who seems to have a problem with me. What is the deal exactly?

Originally, I really did want opinions on whether these fans need prox probes. Folks who seem to be the most familiar with them seem to think "No" or "it'd be nice to have them, but they aren't necessary". Towards the end, I just made up my mind that they are in fact not needed. That's based on my month after month, year after year involvement with these fans, balancing them all the time, aligning any that need it, etc. But hey that's just my opinion.

As for my 'expertise', I'll just let the marketplace vote on that.

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

quote:
as the bearing if designed and working properly would be a node for shaft bending



Not true for any bearing that I would design on a flexible shaft - at least on purpose. This would be a bad design to have a node at the bearing.

quote:
So monitoring in the classical way inside bearings with eddy probes in my view generate very small signal, close to the noise level of the system to be monitored and give difficulties to define alarm levels defined to be able to shut down the machine w/o false alarms.



Interested to know what you monitor that behaves like this.

Comment:
Nodes.
The modeshapes I have seen on turbines the first bending mode for LP to HP stages and generator is very close to have bending using the bearing as a natural "stiff" fix point so it does happen, not that I claim it always does, but surely is not the case that it never happens. It seems naturally to me that it would be normal for a flexible shaft with relatively stiff bearings. For other modes nodes can be present in other places, and I agree for good information from eddy probes, don´t place them in a node for the mode you are interested in having data about.
For slow speed machines like 700RPM we suggest 1 or 2 eddy probes on vertical hydroplants with long shafts as they will alarm if the generator walks away, that does happen at least once and velocity probes on bearings.

Sensitivity.
Since there is a significant struggle to get decent data from eddy probes with shaft anomalies etc. is it not the case that one reason is low signal level from the signal of interest? Due to among other things transducer placement? It also depends on what use you will have of the data from the probes. My interest is to be able to see the shaft bending so I want to have the transducers in a place where it´s possible to mount them and where the bending of the shaft is largest if it exists. I would like to be able to balance the rotors to get low absolute vibration and minimum shaft bending. I can´t see that happen collecting data inside a bearing. If on the other hand you are interested in measurement of the oil film thickness inside the bearing, in the load zone, I wish you luck, compare the total error from all sources in the data to the distance you are trying to measure even with probes inside the bearings.
And yes, I have seen such problems in various sizes of turbines, RCP´s and other hi speed pumps, also in petrochemical.

I will not argue that some machines benefit from having eddy probes but I can´t find it motivated in this case. I argue that a set of probes should not be put on a machine just for good measure or by chance, (not that it doesn´t happen) you should look at what problems are common and put probes on the machine that give the best data to monitor those problems in the best way, what probes or combinations of probes that may be. Olov

Rusty,

Bearing housing accelerometers (one per fan bearing in horizontal direction) with velocity or displacement dispay/alarm would be adequate for permanent monitoring, especially when supported by your periodic monitoring and balancing serveces. Selecting proximity probes for this application is unreasonable.

Walt

I agree with Walt. Most of the ones I monitor that have on line are using the old IRD 544 accel mounted below the bearing on the pedastle. They normally set alarms around 3 mils warning, alarm (a couple shutdown) at 5.

Dave

Bill,
I reread your last comment on this thread and felt I had to respond. You are definitely a road warrior. The kinds of jobs you have been involved with obviously call for cutting edge thinking and sometimes state-of-the-are equipment. I don't think you will disagree with me that your present job and situation has zero degree of mistakes or missed events, especially after the big accident last year. Your business segment has always been one that paid maximum dollar for whatever it took to get the job COMPLETELY done.
I think Rusty is just asking that you open your mind back to some of the other places you have worked (NOT Star Wars, again, unlimited budget) and consider business segments that are getting by on a thread because the profit margins are never going to be there or the competition is so great that they always must consider the best bang for the buck.
Sometimes this board brings out what I always shudder about; going to solve a problem and finding the customer has hired two or more 'experts' to be there and solve the problem! Usually end up with more than one solution, they both would work, but until the blood letting is complete, the war wages full bore!

Hi Rusty,
I didn't have time to read everybody (sorry guys) but I like to share my experience with you. I used to work for a mining company that had 12, 2000Hp fans that runs from 600 to 900 rpm, 3 of which had prox X and Y. First Dodge RTL brg are very effective and reliable so they really don't break that much and standard acc is enough to diagnose unbalance. Personally I recommend prox because it is the only sensor that will give you confidence of what really is going on with the fan's shaft. You cannot do advance analysis unless you have those probe. I missed a lot of problem with fans or I raised flags when that really wasn't that bad just because every things with fans on sleeve brg happen at turning speed where you need phase measurement, shaft position relative motion of from acc versus prox which all give you insight of what is moving and in which direction. It is also difficult to upgrade an existing machine so to have those prox already installed is a bonus. There is also a very important safety aspect of having prox with an automated shut-off device especially with a 1500Hp machine; the implication of a major break down could be devastating.

Your call, best regard, Marcel

quote:

consider business segments that are getting by on a thread because the profit margins are never going to be there or the competition is so great that they always must consider the best bang for the buck.

Whatever this business, they spend money on a consultant for a machine that is very reliable. Why would anyone do this? Probably because this machine has a large importance to the business.

With very stiff pedestals and a more or less rigid shaft (Phase shift alone doesn't tell you that it is flexible, only that a natural frequency is near.) one expects to have larger vibration on the shaft relative measurements than the casing. How much more? Without measuring this one cannot know.

I wouldn't be surprised to see 2 or 3 times greater vibration on the relative measurement in a case like this with a stiff support, relatively low speed (less stiff bearings), and a rigid shaft. The actual relationship could be greater or less than this – Remember, it is not measured.

Balancing to half a mil on the casing is not the same as balancing to half a mil on the shaft relative vibration. Similarly, the shaft relative to the bearing motion will not be 2 or 3 mils when you shut it down for maintenance or balancing at 2 or 3 mils on the casing. If the transfer function between shaft relative and casing vibration is 2 or 3 then the shaft relative would be from 4 to 9 mils; the uncertainly is large without a measurement. These levels could be harmful, depending upon bearing geometry and loads.

The force transmitted to the supports is a function of the relative vibration at the bearings, and with small vibration this is an approximately linear function. The bearings are a point of potential failure. Why not measure the force at the bearings or equivalently the relative vibration?

quote:

I don't think you will disagree with me that your present job and situation has zero degree of mistakes or missed events, especially after the big accident last year. Your business segment has always been one that paid maximum dollar for whatever it took to get the job COMPLETELY done.

The big accident of which you refer, I assume, occured in the downstream business two years ago; I am in upstream. Also, regardless of what you pay at the pump for gasoline, the refineries aren't where we make that much profit; ask someone in downstream about this aspect.

We designate some machinery as run to failure when appropriate, contrary to the statement. Much of our major equipment is not spared, and similar to other businesses that have unspared critical equipment, one takes the appropiate measures as needed.

"consider business segments that are getting by on a thread because the profit margins are never going to be there or the competition is so great that they always must consider the best bang for the buck."

Ron, does that apply to a bag house fan in what I assume is a power plant?

Rusty never really gave us a good criticality analysis of this machine, what the safety, environmental, economic impacts are.

In my star wars world we often determine what kind of monitoring is necessary by using a FMEA assesment. What has happened in this thread is that we have all assumed we know how critical this machine is, have done our own FMEA based on very little data, and decided what monitoring approach is best.

Rusty knows the machine, I am more than willing to trust his judgement. I would probably choose a different approach.

The other aspect that I find interesting is the assumption by a number of people that prox probes would be horribly expensive on this machine. These are brand new machines, and I expect the fan vendor already has provision for probes in his design, as a result the price for prox probes is not going to be much different than accels....That might be different on a retrofit, but this is a new fan.

You maybe save a few hundred dollars with accels, in the price of a new fan that is nothing....

This message has been edited. Last edited by: Steve Schultheis,