I am working with Joel Levitt to update his book on Maintenance and Reliability Internet sites.

Please post any links to your favorite sites here.

I do have a deadline so would appreciate a quick response.

Any and all maintenance and reliability web sites are encouraged.

Thanks
Terry O

Terry or Joel
Please consider my web site...www.wrenchtime.com

Take care of yourself and your family,

Fred J. Weber, P.E.
Author of Wrench Time... using the RPM method to manage maintenance

http://maintenanceforums.com/eve/forums/a/tpc/f/3751089...301059451#2301059451

better late than never?

Amidst a somtimes "vanilla" offering, I may sometimes broach "esoteric" bits and pieces regarding reliability, PdM, vibration...

www.vibra-k.com

23 years of fun, frivolity and frolic? Not quite!

Hi Xo. That's a great site with very interesting reading.

One tidbit caught my eye:

"Another item tied to misalignment: the case of the misalignment-induced energy-loss. One paper reported there were no such mechanical losses from misalignment. We will protect the innocent, and prove them wrong: whenever an asynchronous AC electric-motor vibrates, two things occur, in varying degrees: the RPM DROPS slightly, and the amperage draw rises slightly. This is common knowledge, but not necessarily common to just anyone as extensive experience with balancing and other field procedures should likely be common trait of those who know this for a fact."

I have never heard this before and at first glance I'm skeptical, but always interested to learn. Do you have any additional information on this? (For example: Any example of how much change you have seen? Any reference to discuss this? Any possible explanation for the physical mechanism?)

Thanks.

The observation is purely empirical, over the balancing of hundreds of rotors, and really ties in best with the reverse of my statement...

Let me explain: while balancing, the slight RPM variation is typically known in high-precision work as a source of error. Not a huge one by any account, but still present. If using a balancer with a photocell or lasertach, the increase in RPM can be notable as one alleviates the 1X vibration amplitude. In lighter cases, it may only be beyond the decimal point. Worst case I've ever noted: abnormally vibrating (0.7 ips pk or more)light rotor (boiler fan) mounted directly on 2-pole motor shaft, 45 RPM variation from start to finish. Usual variation will usually be between 2 and 10 RPM (4-5 is a good target).

Obviously, VFDs will obliterate signs of this if the control is sophisticated and the RPM fixed to a preset: you'll go right back to "nominal" RPM = last setting (within control precision).

The mechanical energy wasted by unwanted motion must come from somewhere... For the motor, there are only two possible variations to consume energy away from the main task and dedicate it to "shaking": RPM and power consumption.

For RPM, the "hit" is also power consumption. For purposes of this example, I'll use "neverland" RPMs (reality never has 1800 RPM unless synchro or VFD, and this neglects losses): take a 200HP, 4-pole motor at 1800 RPM, reduced (gears) to 900 RPM. What is the power output? 400HP. A drop of 9 RPM on that motor considered in a different application would account for (close to) 10 HP... (9 for 1800, or 5% of speed applied to 200HP). Loose terms. This is not a thesis...

Do take note that my observation predates the wide adoption of high-efficiency motors. Namely because no one has wanted to pay our fees to cover balancing tasks in quite some time Thus, I can not positively affirm same behavior, but one could suppose same principles apply.

Being aware of the clauses of industrial power supply contracts (expected levels, penalties for spikes above steady consumption, and the like) and given our care to often seek better documentation of exact costs, not to mention the sensitive units, we had monitored many motors during balancing procedures.

Observations often placed Amp draw difference at fractional to 5 A, depending on motor.

One of these days, on a large synchro motor, I'd like to get the exact power consumption before and after a balance job, but it'll likely be as an observer busy on other items.

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

quote:

Xo (from the other thread)
Oh, also, I had replied overly quickly to a previous query of yours (unbalance versus motor RPM and amps). Had meant to say that power wasted to shake structure can almost be calculated directly from quasi linear relationship of slip (say 20 RPM) for 100 HP: 5HP per 1 RPM of slip... As mechanical vibration must be "powered", it is not surprising for unbalance to cause losses through appropriation of energy meant to rotate the machine. That energy must be acquired from kW or HP @ RPM relationships.

I agree 100% that induction motor power is proportional to slip. But I am still skeptical of your comments which suggest that there is significant power savings that result from balancing machines (enough to cause 5 or even 10 rpm change in induction motor speed). If it were true, then there is a tremendous potential for the balancers on this board to promote their services which is being grossly overlooked (balancing is undertaken primarily for reliability, not for efficiency). But I have never heard any studies or experience to support it. It is true that there is varying kinetic and potential (spring) energy associated with vibrating systems, but the losses are only associated with the damping which is very low for most real world machines. For a SDOF damped mass sprng model vibrating sinusoidally, we can calculate the losses as
P = 0.5 * c* V^2 where P is loss, c is damping, V is pk/0 velocity, appropriate unit conversions required. I will try to see if I can scrounge up some reasonable values of damping to apply to an example.

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

I've heard about this many times over the years, so I decided to investigate a little further upon reading this post...

I found "Energy Losses Caused by Machinery Misalignment and Unbalance" presented at IMAC XVII : 17th international modal analysis conference (Kissimmee FL, 8-11 February 1999

Author(s)
GABERSON H. A.; CAPPILLINO R.

Abstract
Tests were conducted to measure the energy lost due to misalignment and unbalance of rotating machinery. Several articles have appeared in maintenance publications that warn of misalignment energy losses of up to 15 percent. Energy losses of this a magnitude would dictate major machinery maintenance upgrading, precision alignment, and use of vibration and infrared diagnostics. Controlled tests were conducted to measure energy losses due to misalignment and unbalance. The greatest loss measured was 1.2 percent at 25 percent power for gross misalignment; most losses were far less. Losses due to unbalance were under half of that. The tests were conducted with a 30-hp. 3-phase motor driving a 20-kW generator connected to a resistive load bank. Power into the motor and from the generator to the load was measured with precision power measurement instrumentation. Four popular couplings were tested. This paper describes the tests, presents data, and comments on the ramifications.



I can't find the article, but would be interested in reading it, if anyone can get it...

As mentioned previously, check your RPM with precision with balancing... Have a look at amps if possible.

The energy must come from somewhere, and the structure rarely behaves like a perfect spring. The classic spring-mass system calculation involves "return" and cycling at natural frequency (for which stiffness is compromised).

As to paper / article, great care was taken to measure the quantity and type of misalignment introduced but no simultaneous measurement of vibration was made.

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

MRC at U of T offers the following:

hines97university.pdf (68 KB, 40 downloads) Motor Shaft Misalignment Research Project

Here's another:

motor-shaft-misalignment-versus.pdf (47 KB, 38 downloads) Motor Shaft Misalignment Versus Efficiency Analysis

Jamie - I didn't look at your second two attachments yet. Referring to the first one and focusing on vibration from unbalance (misalignment is a different subject worth discussing, but I prefer to start with vibration from unbalance) - I agree it's tough to draw any conclusions from the abstract without having some more details. One detail - how much unbalance/vibration was involved. Probably an even more important detail - how accurate was the measurements and how constant was the load/voltage etc... was the experiment controlled tightly enough to measure losses below 1%?

Taking the abstract at face value, we see they're saying half of 1.2% or around 0.6% losses attributable to unbalance in this particular case. (by the way is it 0.6% of full load or 0.6% of 25% load... I give the benefit of the doubt and assume 0.6% of full load). Two quick comments on that result:

1 - This 0.6% of full load would still a long way from causing a 5 rpm change in any motor I know of. A 5 rpm change corresponds to a lot more than 0.6% losses. A typical motor I see might be 1% full-load slip, 1800 rpm, 5 rpm change in speed would correspond to 28% of motor load! Absolute worst case, 3600 rpm motor, 5% full load slip (highest allowable under NEMA MG-1), even in that case 5 rpm still corresponds to 2.8% of full load.

2 - I personally don't even believe the 0.6%. As stated above, we don't have the details. Furthermore, I think a theoretical analysis can show how insignificant the losses associated with vibration is. More below:

================================
Here is my analysis based on the single degree of freedom model which suggests that losses from vibration and associated dampening are insignificant and far less even than suggested in the abstract above:

F = m X'' + c X' + k X
[Let multiplication by s correspond to differentiation (corresponds to assumption of solution exp(s*t)]
F = m s^2*X + c*s* X + k X
H(s) = X/F
H(s) = (1/m) / [s^2 + s*(c/m) + (k/m) ]
= (1/m) / [s^2 + s* 2*sigma + sqrt(sigma^2+wd^2)
= (1/m) / [s^2 + s* 2*sigma + w0^2]
= (1/m) / [s^2 + s* 2*zeta*w0 + w0^2]
where
sigma = c/(2m)
w0 = sqrt(k/m)
wd = sqrt(w0^2-sigma^2)
damping factor = zeta = sigma/w0 = cos(theta)
where theta is the angle between the real axis and the pole in the complex plane.

The response to an impact (bump test) might look something similar to:
x(t) ~ X0* exp(-sigma*t)*sin(wd*t)

Define ratio K such that the waveform drops factor of K between peaks (time difference 2Pi/wd)
exp(-sigma*2*Pi/wd) = K
-sigma*2*Pi/wd = ln (K) = logdec
let "logdec" = ln(K)
sigma = -logdec*wd / (2Pi).
Zeta = (sigma)/sqrt(sigma^2+wd^2)
= -logdec*wd / [(2Pi) *sqrt( (logdec*wd / (2Pi))^2+wd^2)]
= -logdec*wd / [wd * sqrt(logdec^2 + (2PI)^2)]
= -logdec / sqrt(logdec^2 + (2PI)^2)
For lightly damped systems logdec^2 << 2PI^2
Zeta ~ -logdec / (2Pi)

Now look at the attached bump test of a machine.
The ratio K ~ 0.95
LogDec~0.055
zeta = Log Dec / (2*Pi) = 0.0087
I think 1% or less is typical for a damping factor. Maybe a little higher for sleeve bearings which have viscious dampnig from the oil. Let's be generous and assume a whopping 5% damping factor

Let's try out these calculations on a 100 hp 3600 rpm motor. Weight ~700 pounds=318kg. Resonant frequency 4500cpm=75hz, zeta = 5%=0.05

c = 2*m*sigma = 2*m*w0*zeta
c = 2*m*w0*zeta
c = 2* 380kg * (2*Pi*75/sec)* 0.05 =14900 kg/sec

Let's assume a hefty vibration level of one 1.0 ips pk/0 = 25.4 mm/sec pk/0 = 0.0254m/sec

P = 0.5 * c*v^2
P = 0.5 * 14900 kg/sec*(0.0254m/sec)^2 = 5 Newton*meter/sec
P ~ 5 watts.
The motor rating is 100hp which is about 74600 watts

The losses as a fraction of motor full load are 5/74600 = 0.006%!

A factor of 100 lower than the 0.6% in the article even with all those generous assumptions.

I am open to comments on the math (maybe missed a few decimal places somewhere?) or the model.

=======================

Finally, let's take a step back from the math and look at the physical interpretation.

An idealized undamped system can vibrate and dissipate zero energy.

A real system has damping and dissipates energy in the damping. But where does the energy go when we have damping? Easy answer. Damping dissipates energy from the mechanical system by converting that energy to heat. So we shouldn't equate losses with vibration, we should equate losses with heat.

Where's the heat generated in a vibrating machine? The bearings run hot due to friction losses from rotation REGARDLESS of vibration level. Has anyone ever noticed a trend for bearings to run hotter when the machine is vibrating due to unbalance (again talking about unbalance, not misalignment)? I have never observed anything like that myself.

==================
By the way, I hope you don't interpret my comments the wrong way Xo. I like your website and you bring up a lot of interesting topics there. This one is also an interesting topic for me but I just don't see it the way you described. I am always interested to learn if I am missing something.

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

DampingFromLogDec.ppt (98 KB, 25 downloads)

I'd still like to see the article... Or maybe some tightly controlled experiments. There seem to be many variables.

Xo - - - do you have any specific data? 5A seems like a large difference... could there have been other factors involved???

I agree experiment would be a lot better way to answer the question than analysis.

I would still be interested to hear if anyone has any comments on the math analysis.

I have balanced a few motors in motor shops with no load, thus no external influence. I think I had good results. One thing that I never noticed was a change in motor speed, even for the bad ones. Maybe, I wasn't paying attention. Induction motors included horizontal and vertical motors from a few hundred hp up to around 5,000 hp; bigger ones were generally synchronous motors.

Some motors were above the critical, and some were closer than desired, particularly on test.

The SDOF model by ElPete was a SDOF. For MDOF substitute Mi=phi^*(M)*phi and Ki=phi^*(K)*phi for mass and stiffness where phi is the mode shape and phi^* is the complex transpose of the mode shape. M and K are the mass and stiffness matrices respectively. wi = sqrt(ki/mi), rest as usual.

For the SDOF model to work well, all the mass must move at about the same amplitude (or some other special set of circumstances).

Not to worry... I never take objections to my musings as an objection to self, and there is no fun nor any progress in discussion if everyone agrees... "It's red!" "Yup" "All agree" - start new thread

I think the issue is whether a rotor-supporting structure behaves like an ideal spring system. It doesn't... And calls for a switch in assumptions. I'll get back to this after another example (and then I'll just go all over the map)...

Years ago I learned that the load on a centered-load supporting shaft held in a single bearing at one end and dual bearings (back to back perhaps) at the other end gets distributed evenly... 50% (downwards) / 25% (downwards) / 25% (downwards)... Simple, elegant, and generally untrue as this presumes the absence of any shaft deflection. When deflection IS taken into account, a whole bunch of variables must be considered: shaft size, span, load, exact journal location, etc. You may end up with a top loaded bearing on the paired bearings. Slightly different, no? Engineering teachings and assumptions can often use a good reality check.

Back to the regular programming: shake a very flexible member supporting a load of several hundred pounds. You are capable of a steady output of 1/10th of 1HP, and a short burst of several HP. But not several HP at 1800 "actions" per minute. You would need to develop a huge (and continuous effort) to shake anything significantly even at 10 CPM. If you attempt the same thing at a resonant frequency, you'll work a lot less.

The stiffness of the supporting member (pedestal or other) AT resonance behaves like the idealized spring-mass system. Stiffness is presumably close to zero when oscillating (mobility is not infinite, but as you know, is very high in relation to Q or amplification factor). When doing an FRF, you will have a much lower mobility at that frequency corresponding to rotation, unless you have a resonance per say.

What is the difference with the pedestal? The stiffness at RPM is not void. The spring effect WANTS to devolve a portion of the accumulated potential energy, but the system must supply the remainder. Any WORK performed (a force over a distance) will ALSO be a loss. Damping losses are not only converted into heat: they also "move" fluid (air, which is quasi negligible, lubricant film effects, which can amount to something as you know from sleeve bearings).

You can calculate using Work where each cycle loses some distance on final position, and the system must make up for that loss. Compare maximum displacement with return position: the loss will be significant. The problem when you try to do this while stopped is you're now again looking at the pedestal or structure's favored frequency: push and let go yields dominant natural frequency oscillation.

During the forced movement at a different frequency, the structure was worked on: a mass was accelerated and moved for so many watts... (W = F x, F = m a).

It really boils down to the difference between moving something in the "easy" region (resonance) and the not-so-easy region of non-resonance. Damping is an active concern mostly at resonance. Stiffness (and spring inconsistencies) take over elsewhere.

There are many non-linearities and discontinuities in mechanical systems. We've seen differences (in rolling-element bearings) between (temporarily installed) shaft-targeting prox probe measurements and accel bearing cap measurements. Not everywhere of course. Cradling within the pillow block will make a difference (relative radius, etc).

The bearings themselves already generate some heat from friction. G. Louradour in France had an exacting model of element power losses and behavior in virtue of position (Guy, now retired I'm told, is the most knowledgeable bearing person I've ever met, and very practical in his approach).

As to 5 amps, it was for a large motor (think it was in the 400-500 HP range) and driven rotor vibrating abnormally and taken back down to quasi NIL vibration (I was always much fonder of API, MIL-STD than ISO 1940). So that observed extreme is just that, though certain cases might possibly show even more.

For RPM, when balancing, you can have your RPM with one decimal precision, and this has been the case for some time (almost going back to first photocells).

Afterthought: on magnetic cradle (bearing), spinning a non-working rotor such as a shaft will last an impressively long time. A radial oscillation dies down much faster (of course, from damping).

On a soft bearing machine, imposed horizontal movement (push to farthest position and let go)dies down in far less than half a cycle. Very low spring constant and nothing to do the work of returning...

quote:

For RPM, when balancing, you can have your RPM with one decimal precision,

Precision or resolution? For balancing one needs a once-per-turn trigger, and the time period may be averaged over several shaft revolutions.

Xo, I don't understand what point(s) you are making. You mentioned balancing changing the speed by up to 45 rpm earlier not some small fraction of an rpm. I think that I would notice a 45 rpm change, or the motor would pull out. Even the typical rpm change of 2-10 rpm for a good balance would be noticable.

quote:

Originally posted by Xo:
It really boils down to the difference between moving something in the "easy" region (resonance) and the not-so-easy region of non-resonance. Damping is an active concern mostly at resonance. Stiffness (and spring inconsistencies) take over elsewhere.
.

It would be helpful to the discussion if you could be more concise and to the point with your postings.
The discussion is about energy dissipation. The quote above implies that stiffness can be a factor also. But stiffness, even if non-linear or complicated, is a passive effect, never an active effect - it does not dissipate any energy.

Well, I'm not likely to balance any rotors in the next little while whether in the field or on a test bench, so I suggest someone who still does might make precise measurements of RPM and amperage (all 3 phases) before, during and after, and hopefully report their findings...

Given the field variations on the quantity of unbalance, machine configurations and the slip relationship of the different motors involved, sliver thin to wide variations would make sense.

As to the extreme of 45 RPM, the 2-pole motor was sorely out of shape, and was left finely balanced. Perhaps the mechanical energy wasted also contributed to a motor overload which then disappeared, enabling a greater RPM difference. That is quite possible and plausible. My interest was "before" and "after". Nothing more.

As to "stiffness, ... does not dissipate any energy.", granted. But applying a force over a distance does. In a spring, the energy is then stored as potential. And the spring (or more pertinently, a machine's structure) will not devolve a perfect passage of that stored potential to kinetic energy. Merely trying to illustrate that point and I obviously failed.

Bottom line: what costs more energy between a) turning a rotor around its center, or b) turning the same rotor with the shaft center describing ellipses, and the casings, pipes or ducts, and parts of the structure being taken along for a ride?

The various comments seem to tend towards "b), but only due to damping". I find that overly optimistic...

quote:

Originally posted by Xo:
... In a spring, the energy is then stored as potential. And the spring (or more pertinently, a machine's structure) will not devolve a perfect passage of that stored potential to kinetic energy.
...

OK, that's a valid viewpoint. But the losses you refer to are still, in effect, a part of the overall damping. Their effect would be included in a damping estimate such as from an impact / decay test etc.