I agree with the sentiment expressed by Steve on 18 May 2007 10:34 AM. I also have to mention that I respect Bill and Steve's opinion very highly, although I don't always understand or agree with them.

On a different subject, in this case where the machine is reasonably far below resonances (stiffness controlled), we should be able to approximate the ratio of response of housing as compared to absolute shaft position according to the stiffnesses (neglecting the masses) as follows::

HousingDisplacement / AbsoluteShaftDisplacement = Kb / (Kb + Ks)

Where:
Kb = bearing oil film thickness in lbf/inch
Ks = support stiffness in lbf/inch

I would be curious to know if anyone agrees or disagrees with this particular relationship?

ElPete,

... and continuing on this line of thinking we have (calling housing displacement - support displacement - Ds and bearing relative displacement Db, absolute shaft displacement Da).

(Ds+Db)/Da = 1 since Ds+Db=Da

ElPete gave
Ds/Da=Kb/(Kb+Ks) This implies the following.

Db/Da=Ks/(Kb+Ks)

Now if Ks>>Kb this gives Db/Da ~ 1 or all shaft relative motion as is common sense.

From the equations for Ds/Da and Db/Da we can obtain Db/Ds as (Db/Da)/(Ds/Da) or

Db/Ds = Ks/Kb which simplifies the questions previouly explored

If Ks = 2 x Kb then Db/Ds=2, etc.

Some typos corrected

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

"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."

Before they started spending money on a consultant to monitor these fans (and all other pumps and fans in the mill), they weren't nearly as reliable. They were doing the program "in house" and the position was basically a revolving door. Someone 'bid into' the job, spent a year getting competent (training, experience, etc), spent a year doing a pretty good job, and then spent the last year getting bored and burned out. Then they 'bid into' a different job, and the cycle started over. This is typical of many in-house programs, and is the reason they decided to go 'outside' and hire a consultant.

One problem with "continuous" monitoring systems is the people doing the monitoring. If the plant people are not knowledgeable about vibration and the monitoring equipment, confusion often results. Instead of the system giving them confidence, it gives confusion and worry, often over nothing.

I think the expertise of the plant people who will do the monitoring is a critical issue when specifying a continuous monitoring system.

Bill,
Touchy, aren't you? I never said anything about the profit margins of your business. I also was not pointing fingers at your part of the business, just making a statement that you deal in a very hazardous process that requires maximum reliability.
Maybe you haven't been all over the world, as you state, or you would know that there are 10 businesses that can't afford, or refuse to see the value of, higher end monitoring for every one. Should they just shut down because they can't afford it? Maybe, if environmental issues are involved. If they can still manage to make a buck and not break the law, then they must do what is prudent with what they can afford. Rusty has repeatedly stated that this customer is cut close to the vest and does what it needs to get by.
While we're on the topic of the wonderful proximity probe, why not discuss all of it's weaknesses as well;
Shaft magnetism will cause errors
Shaft scratches will cause errors
Shaft runout, if not subtracted will cause errors. You need slow roll runout subtraction, Wait a minute!!! That's what it turns!!!
Shaft material properties, if not known, will cause errors
Unlike your belief, once you use most of the dynamic range of the probe for 1x, you can't see past the sixth or seventh harmonic. If you could, do you really think Don would have traveled down the REBAM road? There's a real winner product for you!
If a proximity probe could see out to 80,000 rpm as you stated, then why not just use it for ball bearings and most other applications? Because what you missed is that it is good up to 80,000 rpm FOR 1X ONLY.
So, now that others have chimed in with wiring these probes to panel meters, what are we talking for a typical installation for one fan? Remember that the world can purchase a reliable, inexpensive accelerometer for approx. $200, with a 4-20 ma output relative to any vibration parameter that can be wired into any existing process controller. Add the readouts to the software, plug in an fft and you're done.


Sleeve bearings that are not giving 'infinite' life are usually the result of poor maintenance and/or poor operation (heat, dirt, etc.). We all know that a sleeve bearing will last forever if the vibration is kept low, the heat is kept low and the oil is kept clean and up to spec. Has anyone at this facility overlooked those issues? Probably.


Why don't we settle this and get Rusty out there to instrument one of these fans with BOTH prox probes and accels, get the data and then analyze the heck out of it?

Hear, Hear. Olov

Shaft magnetism will cause errors

If not fail the machine. Shaft magnetism is a problem for the machine as well as the measurement system.

Shaft scratches will cause errors

Oh yea, don't use chain wrenches or hammers on the shaft as well. One prepares the shaft for use.


Shaft runout, if not subtracted will cause errors. You need slow roll runout subtraction.

Not if your properly prepare the shaft. At 700 rpm with a prepared shaft this should be of little importance.

Shaft material properties, if not known, will cause errors

This should not be an issue for a new machine. If you calibrate against the shaft (Experienced consultants should know this.) one has no issue. Unknown or improper shaft materials have caused bearing failures as well - not an issue with a new installation.

Unlike your belief, once you use most of the dynamic range of the probe for 1x, you can't see past the sixth or seventh harmonic.

I don't think that you will find your failures at the 6th or 7th harmonic, and the physics will not have much relative vibration at these frequencies.

If a proximity probe could see out to 80,000 rpm as you stated, then why not just use it for ball bearings and most other applications? Because what you missed is that it is good up to 80,000 rpm FOR 1X ONLY.

Because shaft relative displacement is not the correct measurement. Ball bearings have little shaft relative displacement. This machine has a different design and will have significant shaft relative displacement.

From the simple examination of the transfer function between shaft relative/support vibration above, we have a rough estimate Ks/Kb, where Ks - support stiffness and Kb - bearing stiffness.

At 700 rpm (and cpm for frequency) which do we think has the greater stiffness, the support or the fluid film in the bearing? Given the geometry of the bearing we can calculate the bearing stiffness. Not knowing this I would state that the bearing at low frequency has a significantly lower stiffness than the support. This means that the shaft relative vibration will be significanly greater than the support vibration if you believe the physics.

The part about being good for 1X only doesn't make sense. Why have you used prox. probes in these applications? If you don't want them remove them; unless, this removes the warranty as well.


If you could, do you really think Don would have traveled down the REBAM road? There's a real winner product for you!

You do know that REBAM is a specialized proximity probe system? I have never contended that REBAM should be used in a specific application. Also for those who are unfamiliar with REBAM, this technology is not relevant to this discussion.


Remember that the world can purchase a reliable, inexpensive accelerometer for approx. $200, with a 4-20 ma output relative to any vibration parameter that can be wired into any existing process controller. Add the readouts to the software, plug in an fft and you're done.

I think the price just increased for the accelerometer installation with all this. Also, what signal are you sending into the fft from the 4 to 20?

Well there are versions that also output the raw AC signal beside the 4-20mA, cost some extra but only a fraction of the shaft treatment required maybe?
I have just written a quote for a turbine retrofit/upgrade with eddy probes, 3000RPM 120MW or so, where my question how the shaft was to be treated (as they would lift the rotor anyway as they lost a complete blade stage) was answered by "as the rotors will be out, the status will be perfect when assembled and any runout will be compensated at slow roll". So that´s how it works IRL. It will be compensated at slowroll. Olov

Bill,
You didn't answer the question. What does a prox probe system, with panel meters cost to instrument both fan bearings and, if the motor has sleeve bearings, both motor bearings?

Most suppliers will happily provide you with a quote if interested for a 90 day period. Design criterion and limitation can also be answered with the OEM design engineer for a specific application.

All veiwpoints may not be the same but the hardcore characteristics and specification with engineered acceptable practices are mentioned and should be taken for good advise. Not all philosophies may not be the same but they do not alter physical data that I think Bill shared with us as well as Oli and Ron. Good topic and useful to many I think.

"don't use chain wrenches or hammers on the shaft as well. One prepares the shaft for use"

"Not if you properly prepare the shaft"

Most plants I work in are a 'rough and tumble' environment (understatement ?) with little in the way of technical sophistication. Whatever you do, you have to keep it very simple and basic or you just create problems when they start "working on it."

I promise this is the last time I will ever mention it, but Bill, you just don't live/work in the world that most of the rest of us do, in spite of your statements to the contrary. (And, no, it's not just Arkansas).

Ron,

I'll leave the non-technical sales and marketing aspects to you; you've probably studied and worked in those areas more than I have, which is none. As Sam says, one needs a quote, and depending upon what you desire the prices could vary substantially.
----------------

As for compsesation of slowroll, there have been various approaches. Most people (at least that I have worked with) would caution against compensation for slowroll in a monitoring system or before you get to the monitoring system.

While this can be helpful for diagnostics, it has many pitfalls for monitoring. Basically, if one can change the signal for slowroll, then it may be possible to change it for other reasons. One of the things we see in slowroll is shaft bow, thermal or otherwise. Tracking slowroll is a recommended CM/diagnostic practice since it can indicate a change in shaft bow.

Recommended practice is to prepare the target area on the shaft and not to use runout compensation. Vector compensation of slowroll is a simplified partial solution and is recommended generally for balancing among other things. There were a couple of journal papers from the 70's

Effect of Residual Shaft Bow on Unbalance Response and Balancing of a Single Mass Flexible Rotor. Part I: Unbalance Response by Nicholas, Gunter and Allaire has some analysis of different methods of balancing, using compensation or not using it. The reference is from Transactions of the American Society of Mechanical Engineers, Journal of Engineering for Powerl Vol 98, Number 2, 1976, pp 171 -- 181. Part II on balancing followed this on pages 183 -- 188 - both give some insight into runout as would be measured on shaft relative probes. Although the analysis is technically for absolute rotor motion (rigid foundation - similar to this example with a very stiff pedestal).

Understandably, one must deal of a different body of knowledge to analyze fluid film bearings.

I have no idea the actual bearing, but given a 5000 lb load to an 8 inch diameter and 8 inch long bearing with 14 mils clearance one can calculate stiffnessess as attached.

Kxx = 1.33 X 10^6 lb/in and Kyy = 3.16 X 10^6 lb/in.

Places a 10,000 lb rotor with a resonance well above running speed of 700 rpm. Oil used had properties of a Shell lubricant, Tellus T 32.

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

Well, there goes the neighborhood. No more plumbers wrenches and come alongs. What will they think of next. Removing the hydraulic jacks and timbers. Oh well. Rusty, I'm with you. Jim

Always good to have an engineer around when you have to do the math. As Bill points out, physics will bear out that prox probes would be of benefit to an analyst with experience in the application of prox probes.

Being a new machine, there is no issue with getting the probes on the machine, or problems with mechanical runout. Is it cost justified? A probe system (probe, cable and driver) will run approx. $600.00 per point. Rusty's initial comment was that the probes would be incorporated into his monthly survey. No need for panel meters or continuous monitoring. A couple thousand dollars seems like a small investment with respect to cost of a new machine. If the customer springs for the probes it would be a good opportunity for Rusty to get the most out of his two channel analyzer and add to his substantial knowledge of machine diagnostics. Never too late for an old data dog to learn a new trick.

My experience with prox probes is minimal. My advice to the customer was to include "vibration monitors" as a seperate line item under "options" of the RFP and let each vendor propose their preferred system(s). We will get more honest pricing that way and maybe a range of options to consider.

Certainly prox probes are a good thing on equipment for which they are appropriate. But the GE 800 MW turbines I worked with back in the 80's had a single shaft rider per bearing with no prox probes, except axial at the front standard. Those managed to get balanced, and we never lost a machine. So prox probes are not the holy grail of vibration analysis.

I wish I had the head for numbers that Bill does, but I don't, in spite of an engineering degree. I can't even tell you what horsepower most of the equipment I monitor is. For what I do, the horsepower is irrelevant.... each machine I monitor is just as deserving of my attention as the next, and each problem is considered important. The customer doesn't prioritize the equipment they have me monitor. They expect me to find every problem. So I really don't have time to get into all the numbers... I just find the problems and diagnose them. I'm not their engineer, and they don't expect me to be.

Using Dave's quote, and assuming eight channels, the price for prox probes is $3200 more than the price for accels.

Ron I am not buying the idea that this customer is so strapped for budget that $3200 makes any difference on the capital purchase of a fan. That amount of money is lost in the noise. The installation crew will waste that on coffee breaks...

Rusty is this a power plant or something else?

And since none of the other 9 fans presently in service have them, there is some question as to whether or not they are needed

How about a sort of compromise, where the bearing housings are drilled and tapped for probes and then plugged until (if?) required.

You then have the option to install probes at some future date (even on a temporary basis) if your customer feels they might help. You might want to consider proper preparation and testing of the shaft area that the probes would look at.

quote:

But the GE 800 MW turbines I worked with back in the 80's had a single shaft rider per bearing with no prox probes, except axial at the front standard.

This certainly doesn't make it right. Today, you can still buy a large steam turbine with shaft riders from at least one major OEM(I am not sure that GE would offer them, but if you want to buy a 800MW turbine I bet they could find some.). I have balanced a large steam turbine with shaft riders. It appeared to go out of balance after a couple of weeks, but when you held down on the offending transducer the vibration was ok - just needed adjusting the shaft rider.

Shaft riders are definitely not the way to go for shaft measurements. Note there are some differences; shaft riders give a shaft absolute measurement with severe frequency response limitations. Because they are an existing technology that has 'proven' to work for many years, shaft riders are acceptable by ISO standards for machinery protection. The standards allow for their frequency limitations.

The axial probe on your turbine got used, because the old back pressure measurement is just too difficult to make reliably. Many would argue that this, the axial position indicator, is the first priority for measurements - relative growth measurements can be important also for these large machines.

Before the early 80's you could have had an arguemnent with GE as to whether you could balance steam turbines using proximity probes instead of shaft riders. We had to have a formal discussion with a power plant and the utility engineers after retro-fitting dual probes (shaft relative with an integrated velocity measurement for shaft absolute in addition to shaf relative and casing absolute measurements available) convincing them that the turbine could be balanced with the new instruemention (in the80's). The world is filled with mis-preceptions.

These fans are at a steel mill, and money is really not a problem. But I don't recommend anything just because the customer can afford it. I want it to be appropriate for their needs and their ability to use it properly. These guys are surprisingly smart about what is and is not needed... they hate vendors who try to push stuff on them they don't need or want.

As for turbines, prox probes are of course the way to go. I thought GE was very backwards about this and couldn't imagine why they took so long to get up to speed on this. We had an older 600 MW Westinghouse unit that had shaft riders also and it developed a vibration problem that would just suddenly appear a while after a cold start. One of the first things I noticed was the 1st critical on rollup was at a higher frequency than it was on coasting down. My theory was that one of the bearings was "unloading" when the unit heated up, effectively lengthening part of the rotor, making it more flexible, and thus the critical at a lower speed.

Westinghouse came in and installed temporary prox probes and you could clearly see the orbit for the suspect bearing change as the unit heated up, eventually becoming almost perfectly round (shimming the bearing solved the problem). I thought the Adre' system they used was extremely slick, even back then.

I am actually fascinated by prox probes... I've just never had the opportunity or need to really learn about using them. IF the need presented itself, I'd be all over them.

quote:

vibration problem that would just suddenly appear a while after a cold start.

This couldn't have been an instability could it? The symptoms aren't entirely consistent with just being a forced response problem.

I guess Bently Nevada was glad to see Westinghous use their ADRE equipment.

Did anyone try to solve this problem with a portable walk around device like a CSI or other data collector? 600 MW, something like a BB-44? I guess this was in Arkansas.

Bill,

Is Arkansas to Texas like West Virginia is to Virginia?