This is a topic that has been discussed a few times before, at least on the old reliability-magazine.com board. I'm not sure if we have talked about it recently here.

It seems to me that as a first approximation, shaft misalignment on a flexibly-coupled machine creates primarily static (rather than dynamic) stresses which can be destructive but will not necessarily produce vibration symptoms.

Let's say we can characterize the coupling reactions using
F = K * X

As a simplest interpretation of the above equation, consider that we are looking only looking at offset X and the resulting force F in one plane. As a more general interpretation, we may consider F to be a vector representing forces and moments in both planes, X to be a vector representing offsets and angularity in both planes, and K to be the matrix of stiffnesses and cross-stiffnesses giving the assumed linear relatinoship between X and F. But this generalization adds complexity and doesn't change the outcome of the argument, so lets stick with the simpler interpretation that we are looking only at offsets and forces in one plane.

There is no reason for offset X to vary over time given that we don't expect the alignment to change as the machine rotates. Therefore, no reason for the resulting reaction force F to vary over time. So the force generated by the stretching of the coupling spring has no inherent time variation, as long as the offset doesn't change when the machine rotates.

So with a static (non-time varying) force F, we can have stress on the bearings, but no vibration would occur.

One way that time variation can come into the picture is if the stiffness K changes as the shaft rotates in comparison to the stationary force. This may be the case due to a variety of reasons:
1 - Keyway in the shaft creates a strong variation in shaft stiffness at 2*phi and a weaker variation in stiffness at 1*phi where phi is mechanical angle of rotation.
2 - The coupling may have inherent variation in stiffness as it rotates. For example Lovejoy 3-jaw coupling appears to have stiffness varying at 3*phi and generates 3x in the presence of misalignment. Also for shim pack and disk type couplings, the bolting pattern might cause variation in stiffness as the coupling rotates.
3 - The coupling may be assembled/tightened in a nonuniform manner that causes assymetric stiffness and dynamic forces, possibly as a result of the misalignment of the shafts.

So #1, #2, #3 all provide possible ways that the static force may translate itself to dynamic movement. But these seem like secondary effects to me. It is certainly possible for misalignment to cause large forces without creating noticeable vibration, isn't it?

So the main question is:
Can a destructive misalignment condition be present without causing vibration?

The practical signficance of the question comes in a number of different contexts, including:

One context is that we have a family of machines with cracked foundations and a fair number of unexplained bearing problems. We just found one machine out of alignment by 17 mils offset vertical, but there was no vibration pattern correlated to that misalignment (vibration didn't change significantly when the alignment was corrected). Some say that the absence of misalignment-correlated vibration indicates no action is required with respect to the cracking, but I think the absence of misalignment-correlcted vibration doesn't mean much here).

Another context is evaluation of vibration severity. There is of course usually a fairly large unknown of dynamic stiffness. But if we get past that and allow ourselves the audacity of attempting to guesstimate severity based on vibration magnitude, then we should consider vibration suspected to originate from misalignment more severe than a similar magnitude of vibration suspected to originate from unbalance, right?

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

By the way, CSI used to have some good articles on reaction forces associated with coupling misalignment, but I can't find them any more. Does anyone have a link or a copy of those articles?

quote:

then we should consider vibration suspected to originate from misalignment more severe than a similar magnitude of vibration suspected to originate from unbalance, right?

That's exactly right. You can setup two machines of equal magnitude; one have soley imbalance while the other is misaligned - the misaligned machine will fail much quicker.

"It seems to me that as a first approximation, shaft misalignment on a flexibly-coupled machine creates primarily static (rather than dynamic) stresses which can be destructive but will not necessarily produce vibration symptoms"> I have to grossly disagree on this one - flexible couple takes in a lot of territory (Woods type to Zurn gear type). Gross misalignment will take on a heavy pull and may omit the axial completely along with 2X while the less misaligned machine may show all characteristics. Depending on the machine and its design the coupling can go before the machine everytime.

I may not have understood your thread completely; so, take mine lightly.

Suppose you have a shaft with 3 keyways at one axial location spaced equally around the shaft, what type of shaft (beam) stiffness assymetry do you have? What type of vibration would you expect from misalignment on this? The Lovejoy coupling brought this to mind.

quote:

Can a destructive misalignment condition be present without causing vibration?

Yes, absolutely. And a severe misalignment condition can be present that is not destructive. And a severe vibration can be present which is not destructive. And a machine that has very little vibration can fail overnight, for no apparent reason. And some machines with imbalance present along with severe looseness in the bearings can be balanced fairly easily, while others not at all. Some machines running within a few rpm of a critical speed can be balanced fairly easily, while others can't.

I have seen exceptions to just about every "diagnostic rule" or rule-of-thumb. If you only follow the diagnostic rules, you will be right about 80% of the time. The other 20% of the time, you will be wrong, and often embarassed as a result. You have to always be thinking about the machine, how it's built, supported, operated. There is no substitute for first-hand experience.

Just another question to throw into the pot.Could it also be said that a machine with imbalance might only affect one part of the machine train while misalignment is more likely to affect the entire machine train.Imbalance is created from outside for the most part ie.dirt build up on the blades or impeller yes it is in a housing but a good portion of imbalance comes from the process or a failure of a component this can be very visual where misalignment can be visual such as a soft foot but the forces are acting on the internals of the machine train
making it hard to see the progression of resultant damage.If the misalignment is so severe that it results in component failure then I would like to think that it should be noticable with one of the many technologies availible today.Is this something that is a rare occurrance Pete or are you thinking that we are missing the damage induced by vibration more times than we would care to think about?As stated by Rusty regarding the diagnostice rules
there are exceptions to every rule.If it is the exception regarding the issue then I would agree that this could and does happen on occasion.We have all sat down and scratched our heads saying how could this have happened and I was unable to see it coming.My belief is and I coud be totally wrong is that somewhere in the vast amounts of data collected with the numerous technologies availible the asnwer is there. The only thing holding one back would be the lack of access to all the types of tools applied to our industry.The old adage of nobody is perfect also applies to machines after all we created them so if you do the math yes we will not see everything all the time.There is no such thing as 100% correct in any industry or for that matter life.We can only do the best we can and try to explain to the powers that be why thing blew up when being monitored on a regular basis.So to answer the question yes but only when all resources have been accessed and applied showing zilch vibration otherwise no matter what amplitutde is seen it should be considered as a force that will over time be destructive.The only good vibration is from the beach boys,vibrating recliners and the rumble of 1000HP doing the quater mile.

quote:

Originally posted by electricpete:
So the main question is:
Can a destructive misalignment condition be present without causing vibration?

In the attachement I am trying to show how vibration results from misalignment and that its destructiveness directly depends on coupling elasticity. (In any case misalignment is destructive for the coupling.)

For simplicity purposes I am taking a case of a homogeniuos shaft and just offset misalignment trying to prove that even in this case vibration will occur theoretically.

TIOMOAICBTW

David

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

Misalignment.doc (28 Kb, 90 downloads)

Dave
I can see no reason for this to vibrate even if the shafts are rotating as long as one does not surpass the couplings ability to recover from the cyclic forces it will see. On the other hand once the couplings ability to recover is passed then you would see vibration occur.To me this looks somewhat similar to a drive shaft which can have misalignment to the eye but does not vibrate unless the angles recommended by the manufacturer are not adhered to. Go beyond these limits and wow do these units light up.

Hi Pete,

I have observed stiff coupling - rigid bolted couplings behaving and showing the classical misalignment spectra due to small misalignemnt. In case of flexible couplings like metastreams and lovejoy - metallic shim couplings the spectra rarely shows signs of misalignemnt but on inspection of the transmission units I have seen signs of distress - like cracked shims in the unit.
Further i am attaching a few spectra i found in the net - on the various misalignement spectra for various coupling types and they seem to match with my obsevations made in the field.

Regards

Ramesh Rao

I did some searching and dug up 5 articles.

1 - http://www.ulb.ac.be/polytech/laborulb/couplings/couplings.pdf
Reactions of flexible bearings and vibrations.
(By maintenanceforums.com member Gerald D’Ans)

2 - http://www.mhm.assetweb.com/drknow/aplpapr.nsf/06b6f5a4...05e816e?OpenDocument

Characterizing Shaft Misalignment Effects Using Dynamic Measurements (CSI)

3a - http://citeseer.ist.psu.edu/cache/papers/cs/15639/ftp:z...lignment-bearing.pdf

Motor Shaft Misalignment Bearing Load Analysis (CSI) (10 pages)

3b - http://www.compsys.com/drknow/aplpapr.nsf/19fcc759cbee3...05ab6c4?OpenDocument

Motor Shaft Misalignment Bearing Load Analysis (an expanded version of 3a) – 32 pages
(this site is sloooooooooow)

4 - http://www.spectraquest.com/resources/misalignment.pdf

Observations Concerning Misalignment Vibration Signatures (Spectraquest)

5 - http://www.eng.monash.edu.au/uicee/worldtransactions/Wo...actsVo2No2/Tan30.pdf

A practical approach to learning vibration condition monitoring (Queensland University Au).
(Figure 5 on page 4 is result of misalignment on test rig)

Sam – You are absolutely right that I shouldn't be lumping all couplings together - they each have their own personality. And thanks for reminding me that it’s not just the bearings that can be challenged by misalignment, but the coupling as well. Good points.

Bill – I’m guessing 3 keyways would give 3x but we don’t have any machines like that. Do you?

Rusty – Without exceptions life would be too easy. But I’m thinking maybe it would be the exception that misalignment at 0.2ips creates the same bearing stress as unbalance at 0.2ips. The rule would be misalignment creates a lot more stress for the sam vibration level. I’m wondering whether we have a lot of misaligned machines that vibration doesn’t tell us about. But we don’t see it that often because we don’t usually go in and check as-found alignment on machines that don’t show misalignment on vibration (at least not at our plant… maybe others have comments). As I mentioned we did check one recently as a result of other work and found 17 mils offset misalignment, for gear-coupled motor-generator set, but never showed up on vibration.


Thanks Ramesh - It looks like the graphic you providedd comes from the end of #2. We have also found cracked Thomas Shim pack couplings fairly routinely recently without abnormal vibration present. Up until recently we haven’t been regularly performing a careful inspection/replacement of our shim pack shims, but we have started doing it now.

Lee – that’s a lot to think about. My opinion (which might be wrong), is that it’s not just a fluke to have degrading/destructive misalignment without any significant vibration, it is somewhat expected. You’re right I should also clarify I am assuming no hystersis type effects in the coupling. Otherwise I think we are looking at it the same way.

David – In trying to predict how the coupling will behave, a lot depends on your assumptions. I assumed that several “second order effects” were negligible: negligible variation of stiffness with angle for either the shaft or the assembled coupling. Also in view of Lee’s comments, I should add another assumption that there is no hysteresis in the response of the coupling. In short: we assume the coupling acts like a linear time-invariant anisotropic spring.

With these assumptions, we have the analogy of a simple spring. You apply a constant displacement (constant X) to a spring with unchanging stiffness (constant K) and you get a constant force (constant F = K*X)

The displacement is the displacement from the initial misalignment. In general that doesn’t change over time. OK, maybe it will change due to thermal effects and due to torque loading and it’s effect on coupling, but when the machine is up and running at steady state (under the assumptions above), the misalignment of the shafts is constant.

As I sit here in my hotel room (business travel this week), I have a cheap flexible plastic hotel pen at my desk. It is 6” long. If I press upwards with my thumbs 2” from each end and down with my fingers at the end, I can create a bow. If I stop and rotate the pen 20 degrees and try again, I create the same bow with the same force. If I use alternating thumbs pushing up and tangentially to roll the pen while keeping constant force downward on the ends with my fingers, I have a pen rotating in a constant shape with constant force applied to my fingers. Even though the pen is rotating, the force is constant.

All – I believe article #1 tends to support my view. For gear couplings, on pages 14 and 15 it shows the shear force associated with a misaligned gear coupling at low load is 158 Kgf static and 10 kgf dynamic. Under full load that increases to 416 kgf static (almost 1000 pounds) and 35 kgf dynamic. For disc and elastomer couplings on page 30 he indicates the dynamic forces for his misalignment experiment are almost negligible.

So with the dynamic forces much lower than static forces, vibration resulting from dynamic forces will not be a very good indicator of the total forces. Once again for similar level of vibration, I would be much more concerned about the machine that appears to be misaligned than the machine that appears to have misalignment.

Just thinking some more, some other possible reasons to be more concerned about symptoms of misalignment than symptoms of unbalance at comparable vib levels (beyond the fact that higher forces are involved):
1 – it tends to create more of a moment on a bearing which can be more destructive than a pure radial load. Unbalance often creates more of a radial load.
2 – Misalignment puts stress on the coupling, unbalance usually doesn’t.
3 – For unbalance, the shaft will take a shape and that shape will rotate with the shaft, creating no fatigue cycles. For misalginment under the assumptions above, the shaft rotates but it’s shape remains stationary, meaning the shaft stresses are continually changing creating the potential for fatigue. Just worth a mention, even though I can’t say that shaft failure is a common occurrence, and there would probably be many components dieing long before the shaft.


Tiomoaicbtwa

Keep those good comments coming.

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

quote:

http://www.ulb.ac.be/polytech/laborulb/couplings/couplings.pdf

The above reference is the most comprehensive out of the rest since it addresses the cause of vibration due to misalignment.

According to the authors, when a flexible coupling is employed, there is no vibration even with misalignment in place at zero torque. This is a bold but a well proven experimentally statement. Eccentually at a normal torque two originally offset shafts acquire an S-shape. This is not only causing shaft flexing with associated stresses but also vibration.

quote:

According to the authors, when a flexible coupling is employed, there is no vibration even with misalignment in place at zero torque

I believe that statement is in error. Of course the amount of misalignment and type is not mentioned. But large mass and high speed equate to imbalance produced by offset masses of a centerline and will produce high 1X from 'bad' alignment and probably no 2X or 3X with no imbalance exhibited from either machine. You can setup an experiment in the shop when you have a 'real world' machine in or a new one on a skid package. This has been done in increments of 5 mils going to gross misalignment. You can watch the 1X, 2X & 3X components and see the 2X & 3X disappear with gross misalignment leaving a high 1X. I've been called to balance a number of times where misalignment was the real problem.

That's why misalignment is enemy no. 1 to machineries?

josh - that's what I'm suggesting - that misalignment can be a silent killer.

David - If you look again at the sections I referred to on pages 14/15 and 30, both low-load and full load are addressed and the conclusion is that the dynamic shear stresses are much lower than the static. You are correct that as load increases, the coupling stiffens and creates more shaft bending (S-shape). Under the assumptions I outlined above, this leads to static (vs dynamic) stresses. This is the same logic that tells us that when we apply a static displacement to a time-invariant spring, we get only a static force response and no vibration F=K*X.

The assumptions are that the shafts and coupling are isotropic (properties don't change as the shaft rotates) and non-time varying. I'm not saying these assumptions are always true, but I hope you would agree that IF we accept the assumptions for sake of discussion, THEN there would be no dynamic stresses produced by vibration. The purpose of the thought excercize of the assumptions is to lead us to question another "common" (?) assumption that vibration would be a reliable indicator of misalignment and stresses from misalignment.

Sam - that's a good experiment. It would be interesting to hear more of the details. What type coupling, how much misalignment, and how much vib changes.

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

All - I have heard from an experienced ex-Westinghouse turbine engineer that vibration from misalignment often has a "multiplicative" effect to vibration from unbalance. If we have only misalignment (on turbomachinery), often there is no vibration. If we have only unbalance, there is vibration. If we add misalignment to the vibration, the resulting vibration is often higher than it would be with unbalance alone. Has anyone else heard this? (I think it applies more commonly to turbomachinery).

I thought for awhile come up with some possible explanations why machines might act this way. The first two fairly simple and the last a little more subtle. I think the last is more correct based on what I understand from the literature.

#1A - Misalignment creates a preload which shifts the resonant frequency up and amplifies the vibration. I don't think this is true because no-one has shown shift in resonant from misalignment by coastdown. Also shifting the resonant frequency up might increase or decrease vib depending on where the machine is operating compared to resonances.
1B - Misalignment creates a preoad which results in more vib transferred to the bearing. But the vib monitored on turbomachines is primarily shaft position by prox probe.

2 - One aspect of a rotor which may not meet the isotropic assumption above is a state of rotor unbalance. If we violate that isotropic assumption through unbalance, then we have a mechanism for misalignment to create time-varying reaction forces, resulting in vibration. To visualize, imagine that the unbalance causes one rotor to bow and we have angular misalignment. With the rotor at one angle, the bow introduced by the unbalance would tend to reduce the force of the misalignment to a minimum, with the rotor 180 degrees from there, the bow from the unbalance would tend to increase the force from the misalignment to a maximum. To see that this is a multiplicative/non-linear (vs additive/linear) combination of vib from misalignment and vib from unbalance, remove the unbalance and therefore remove the bow and the vib from BOTH the unbalance and from the misalignment would go away (under my assumptions that other rotor/shaft/coupling properties are all isotropic and non-time varying). This mechanism would generally produce vib at 1x I would think.

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

quote:

According to the authors, when a flexible coupling is employed, there is no vibration even with misalignment in place at zero torque

This statement seemed to apply to the 'lubricated' couplings and to a lesser degree the elastomeric couplings. Other couplings like shim pack and diaphram don't show this torque related effect as noted in the article. Rigid couplings like on a TG also do not show this torque related effect. Of course this assumes that the centers of rotation or centers of mass for the two shafts are not offset in the coupling.

quote:

#1A - Misalignment creates a preload which shifts the resonant frequency up and amplifies the vibration. I don't think this is true because no-one has shown shift in resonant from misalignment by coastdown. Also shifting the resonant frequency up might increase or decrease vib depending on where the machine is operating compared to resonances.

One of the needed factors in doing rotor dynamics on large TG sets is the load on each bearing. The loading of the bearings does affect the rotordynamics and hence the response. Alignment (or mis-alignment) affects bearing loading, and thus can affect the vibration.

Alignment has been known to cause instabilities in fluid film bearings. Sometimes the loading of exiter bearings is somewhat heavy because of fear of whirl; a better bearing design might help.

quote:

1B - Misalignment creates a preoad which results in more vib transferred to the bearing. But the vib monitored on turbomachines is primarily shaft position by prox probe

I don't know why this would be true. Mis-alignment can unload the bearing as well as load it.

quote:

2 - One aspect of a rotor which may not meet the isotropic assumption above is a state of rotor unbalance. If we violate that isotropic assumption through unbalance, then we have a mechanism for misalignment to create time-varying reaction forces, resulting in vibration. To visualize, imagine that the unbalance causes one rotor to bow and we have angular misalignment. With the rotor at one angle, the bow introduced by the unbalance would tend to reduce the force of the misalignment to a minimum, with the rotor 180 degrees from there, the bow from the unbalance would tend to increase the force from the misalignment to a maximum. To see that this is a multiplicative/non-linear (vs additive/linear) combination of vib from misalignment and vib from unbalance, remove the unbalance and therefore remove the bow and the vib from BOTH the unbalance and from the misalignment would go away (under my assumptions that other rotor/shaft/coupling properties are all isotropic and non-time varying). This mechanism would generally produce vib at 1x I would think

I don't get this one. Imbalance would create a rotating bow (perhaps asymmetric). Rotating bows don't look like misalignment (except for things like couplings put together with axial runout [cocked], which is a little different again).

What if your shaft had three (or more) stiffness asymmetries equally spaced around the shaft, and the shaft still acted like a shaft (beam)? There is no beam asymmetry; hence no 2X.

Hi Bill. I appreciate your comments.

Do you agree with the premise about the observed relationship between unbalance vib and misalignment vib discussed in the first paragraph of my post 18 November 2006 08:24 AM?

Regarding 1A - I didn't mean to say that misalignment has no effect on resonant frequency behavior, only that I believe this is a small effect and could go in either direction (to increase or decrease vib), and therefore does not explain the observed behavior.

1B - I agree. My main point on 1A and 1B is these are not the explanations for the observed behavior.

What is it you don't like about my #2? That it doesn't predict a 2x component?

What other mechanisms are there for shaft misalignment to cause vibrations?

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

quote:

Originally posted by rustythevibeguy:

I have seen exceptions to just about every "diagnostic rule" or rule-of-thumb. If you only follow the diagnostic rules, you will be right about 80% of the time. The other 20% of the time, you will be wrong, and often embarassed as a result. You have to always be thinking about the machine, how it's built, supported, operated. There is no substitute for first-hand experience.

That 20% is what keeps guys like me in business. Checking the other 80% is just the icing on the cake

Ramesh Rao - very nice FFT examples, thanks!

electricpete,

No matter how advanced the diagnostic field becomes, there is simply NO substitute for "hands on" inspections. I have several customers that require annual alignment verifications as part of the PM's, regardless of other indications.

The first time I had this requirement presented, I was skeptical. I verified the alignment on a small (40HP) compressor and found it had a 0.035 inch horizontal offset (max spec for this was 0.003). Vibration amplitude before AND after correction was 0.02 in/s. I was impressed and a little disenchanted with my vibration meter.

And then I recalled that this same unit lost a coupling bolt 4 months prior, at which point the vibration went to around 0.2 in/s.

I can't explain it, I've stopped trying to figure it out.

Misalignment kills? YES
Silent killer? sometimes

Take Care,

ElPete - Much does depend upon the coupling. TG's have rigid couplings - at least large ones do. I don't see a 'multiplicative' effect.

I don't see how imbalance creates an anisotropic system, and I am not sure the meaning of anisotropic is correct in this situation. The varying reaction forces only vary in time because of the system response; the forces do not have a time independent nature. As such the system has no independent time varying components such as a shaft with a rotating stiffness asymmetry.

Gear couplings (and the paper mentioned other 'lubricated' couplings) are notorious for 2X vibration and axial vibration due to misalignment. A stiffness asymmetry in the shaft like a keyway or a 2-pole generator (although gravity will suffice to elicit 2X on a horizontal 2-pole generator all by itself, which makes using 2X response difficult to diagnose misalignment on 2-poles. - Before asked, I can't remember seeing a vertical 2-pole.).

On large TG's with solid couplings, if only using vibration misalignment would be very subtle. The difference in bearing loadings could change the vibration from what would be a proper alignment, but without the reference to a proper alignment one could not tell the difference.