I worked with Irvin in Saudi Arabia. It is good to see some serious thought in this area. He has shown his model with derivation.

The model does have a 1/2 coupling type arrangement that causes speed variation for parallel offset. Many couplings will look more like 2 of these with angular alignment at each half.

Hi All!

In case everyone forget, i'm the thread starter .

Hmm, after a tiring exploration of some research paper (like Dr. Irving's paper, yup, all with those scary algebra , and some supported with laboratory experiment ). I've concluded that misalginment is a very complex and nonlinear situation. A misalginment may and may not causing 2x rotation freq. The nonlinearities come from many things like the flexible coupling, the coupling bolts strain, etc. There's still many things to be researched to explain the effect of misalignment to vibration.

[/QUOTE]

To add to the confusion... A "well defined" M or W pattern in the TWF would certainly infer imbalance more so than miss-alignment. I do agree the pattern can not always be trusted in it's purity but when considered/analyzed with an absence of other know forcing frequencies relative to that machine, does provide a valid visual starting point. The attachment is a case in point.[/QUOTE]

I wouldn't consider the TWF you posted as being the "M" or "W" pattern. What you have posted is a Sinusoidal wave form (showing unbalance) with a lot of high frequency content.
I will search in one of my databases for a misaligned machine and post what I believe has been considered the "M" or "W" (depends on which way your head is tilted I suppose)

Dave

Opiq, for what it's worth, I agree with the others that misalignment is very difficult to diagnose accurately with vibration data. The vibration produced is highly dependent on the coupling type, size and stiffness of the machine and shafts, the base arrangement.

Over the past 20 years I have had limited success "correcting" a vibration problem by aligning a machine. Usually alignment has little effect on the vibration.

But I can accurately predict misalignment, because 80% of the machines I monitor will have significant misalignment, unless I aligned the machine myself. (It's not that I'm so good, it's just that the majority of plant people doing alignment work are so bad at it.)

Now 'coupling' problems, other than pure misalignment often do cause 2x vibration. But that's a little different scenario.

I was wading through the link posted by Bill.

One thing made sense. It said on the bottom of page 44 that offset misalignment provides cross coupling between torsional and lateral vibration.

I think I have a way to envision that. Hold each end of a pencil with one hand. With your right hand, try to rotate the pencil, and with your left hand, try to resist the rotation. The pencil will twist a little, but will not move laterally, because it is straight (no offset misalignment). Now repeat the same thing with a pencil that has a dogleg in the middle (similar to offset misalignment). When you try to apply a pure torsion, one end of the dogleg will go up and the other will go down as shown in attached. The torque is coupled to lateral motion when you have that dogleg.

So, what does it have to do with 2x vibration in presence of misalignment? Beats me. One scenario he mentions on page 45 is that parallel misalignment produces torsional forces at 2x, especially with large difference in H/V stiffness. (and the torsional 2x I presume is coupled to produce the lateral 2x). But how and why offset misalignment causes 2x torsional force... I can't quite imagine that part. Maybe someone can explain it?

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

TorsionalToLateralCouplingBasedOnOffset.ppt (40 Kb, 44 downloads)

The 2X arises from the offset and the coupling model. The coupling is in an unstrained position at one point but not others, because the coupling has no radial stiffness and the shafts are rigid. This requires a pin joint at the coupling.

The model is well detailed in this paper. It might be fun to play with or tweak.

In the old days, more companies sponsored this type of journal, and there were some good articles in them. One doesn't see this much anymore. GE, Brown Bavari, and Westinghouse had publications of note. Orbit by GE is ok now, at least when it comes out. This publication by Saudi Aramco has a variety of articles, real class.

An example that some might think to be similar to the offset alignment model and should be close to home for some would be parallel offset of two turbine rotors in a big TG. These rotors have rigid coupling, but the shafts are flexible. 2X is not expected, except on the devil's disciple, the 2-pole generator, if the mis-alignment occurs there. Which came first gravity or the misalignment?

Hi
Angular misalignment produces bending moment on each shaft and this generates a strong vibration at 1X and some vibration at 2X in the axial direction.
parallel misalignment produces shear force and bending moment on the coupled end of each shaft which produces high 2X as well as 1X in the radial direction.

The bending moment and shear are in one direction and do not rotate. To get this at 1X it needs to be coupled into rotation, perhaps some hysteresis.

In some cases the shaft to shaft coupling can provide this work.

Hi

Can you explain more Mr William_C_ Foiles

thank you

[quote WC Foiles]"The bending moment and shear are in one direction and do not rotate. To get this at 1X it needs to be coupled into rotation, perhaps some hysteresis.

In some cases the shaft to shaft coupling can provide this wo[/quote]"

Some definitions of a Constant velocity universal joint require the lines of action or contact to operating in a plane that bisects the shafts' angle (angular misalignment).
In this animation of a typical CV joint the balls are rotating in a plane halfway between the input and output shaft angles. http://upload.wikimedia.org/wikipedia/commons/3/3a/Simp...V_Joint_animated.gif.
I think a grid coupling's and gear coupling's kinematics are similar if they're in good shape, with a DC non-rotating force (no vibration) cross the coupling.
(If the input and output torque is ~ constant)

the ends of each "cross" in a (non CV) Hooke or Cardan joint stay in the plane of their respective shaft, like this (tan dots are cross ends)-
http://upload.wikimedia.org/wikipedia/commons/3/3a/Simp...V_Joint_animated.gif
There is a 2X rotation secondary couple induced perpendicular to the plane formed by the shafts, by the variation in geometry. The torque amplitude is proportional to the torque being transmitted (torque x Cosine?). It's a common source of vibration complaints in RWD cars and trucks.
I picture a 4 jaw elastomer or bolt disc coupling being similar.

I'm still struggling to understand 3 bolt and 3 Jaw couplings.

In several common coupling types parallel misalignment can only be accomodated with 2 angular misalignments. It's fairly obvious when there are 2 discs or gears with a spool piece.

Dan - was your 2nd link intended to be the same as the first... or is there supposed to be a different link?

Following up on Danny's comment

Attached is info on Cardan joint:
http://en.wikipedia.org/wiki/Cardan_joint

Attached is graph of output shaft position over one rotation of input position.
http://en.wikipedia.org/wiki/Image:Universal_joint_-_ou...e_to_input_angle.png

In the link just above you can see the ripple variation of output shaft angular position goes through two cycles during one shaft rotation. 2x. This represents 2x torsional excitaiton. And the explanation lies in the geomtry as Danny explained. The Cardan clearly changes characteristics as you rotate it (input bar at 12:00/6:00 position is different than input bar at 3:00/9:00 position). But if you rotate it exactly 180 degrees, it looks the same, so 2x vibration seems logical based on variation in coupling characteristics.

I didn't quite follow one thing – are you suggesting similar logic applies to grid and gear coupling? I don't understand why for grid or gear we would expect a similar change in coupling characteristics vs angle, repeating every half rotation .

Attached is self made simple representation of two parallel offset shafts and dynamic forces generated as result of rotation. It is assumed that both shafts (except for the overhung section of the driven one, which is soft when bending or stretching and is linear ) are infinitely rigid and are resting in infinitely rigid bearings. There is no coupling per say, instead a ball-socket connection is used.

This is to demonstrate kinematics and resulting dynamic forces which are being visibly produced as one watches motion of the flexible part of the driven shaft.

As could be seen various forces will produced at 1xRPM and were observed in real life by many practicioners.

There is a possibility (which I did not investigate) that forces derived this way are not sinusoidal, then higher orders will be also expected.

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

Misalignment_model.doc (30 Kb, 56 downloads)

In David's model the red shaft looks bent to me, in addition to being radially offset/misaligned.

If the hunk of flexible shaft connected 2 rigid offset shafts I think it would form a curve with end conditions appropriate for each shaft-to-shaft junction.

Similar to this 1963 Pontiac Tempest (as made famous in the movie "my cousing Vinnie") driveshaft.
http://www.sportscardesigner.com/corvair2.jpg
The engine and rear mounted transaxle remain relatively fixed, as they are not involved in wheel travel over bumps. But to provide a less intrusive driveshaft tunnel, the small diameter driveshaft was forced to hold a permanent curve by a few bearings dispersed along its length.

Yep Pete, I pasted the wrong Cardan joint animation link

[quote]the ends of each "cross" in a (non CV) Hooke or Cardan joint stay in the plane of their respective shaft, like this (tan dots are cross ends)-
http://www.heiszwolf.com/heiszwolf/hobby_johan/universal_joint_animation.GIF


Dan T

Pete commented on Timberlake's post

quote:

didn't quite follow one thing – are you suggesting similar logic applies to grid and gear coupling? I don't understand why for grid or gear we would expect a similar change in coupling characteristics vs angle, repeating every half rotation

Sorry if that was not clear.
I don't think grid and gear couplings have the 2X thing going on. Maybe even close to pure DC preload, proportional to torque transmitted and how slippery the grease is. In a typical gear type first because there are dozens (not 4) faces sliding all the time, and second depending on the tooth profile the line of action >>might<< even be on the bisecting plane, which means its a CV joint.
I think of grids being kind of like gear types, with the 2 sets of gear meshes all in one unit, allowing it to handle offset misalignment, but with smaller tooth count.

Of course after 6 years of running without maintenance things start to seize and wear, and, as sometimes happens, 1 side wears a lot more, to the point things can lock in a seriously mechanically offset relationship, and produces a whopping 1X cranking force.

Here is a link copyrighted 2007 with an article by the honorable John Piotrowski.

http://www.maintenanceresources.com/referencelibrary/alignment/importance.htm

Here are some quotes snipped from that article.

"Misalignment is not easy to detect on machinery that is running. The radial forces transmitted from shaft to shaft are typically static forces (i.e. uni-directional) and are difficult to measure externally.

Excessive radial and axial vibration. (*Note ... tests have shown that different coupling designs exhibit different types of vibration behavior. It appears that the vibration is caused by the mechanical action that occurs in the coupling as it rotates).

quote:

Originally posted by Dan Timberlake:
In David's model the red shaft looks bent to me, in addition to being radially offset/misaligned.

I have "straightened" the "bent" red shaft by using a disk. The concept did not change though.
The dotted black portion simulates the other half of the elastic coupling. By using kinematics the model shows all the forces acting at 1x on it - radial and axial - and transmitted to the bearings.

Is there anything wrong with the model?

Misalignment_model_updated.doc (31 Kb, 35 downloads)

I think the representation of pure offset misalignment should have the ball and socket connection on the center of the disk. Isn't that how the "real" machine would be made? The end of the black shaft would always be bent -Y to meet the center of the butterscotch disk.

quote:

So, what does it have to do with 2x vibration in presence of misalignment? Beats me. One scenario he mentions on page 45 is that parallel misalignment produces torsional forces at 2x, especially with large difference in H/V stiffness. (and the torsional 2x I presume is coupled to produce the lateral 2x). But how and why offset misalignment causes 2x torsional force... I can't quite imagine that part. Maybe someone can explain it?

I think it has to do with the H/V stiffness difference you mentioned. The bearing stiffness provides a resisting torque for offset shafts. If the H is higher than the V for example, you'll get higher torque resistance at the 12 and 6 O'clock positions thus giving rise to the 2X. Not sure why this would cause a 2X lateral vibration though.