Converting from WPII to TEFC Motors 3

[JJPellin]

I will start with a disclaimer. I am a rotating machinery engineer with expertise regarding pumps and compressors. I know relatively little about motors.

We have been converting a number of motors from the original WPII configuration to TEFC. These motors tend to be in the range of 300 to 1200 HP, two pole (3600 rpm) and 4160 volt. These motors are all sleeve bearing designs. We seem to be seeing two sorts of problems after these conversions.

Some of these motors have been experiencing chronic axial shuttling. Our couplings are disk-pack designs (Thomas Series-71). We see a persistent axial bounce in the motor shaft which would typically be associated with a motor fighting to get back to magnetic center. We have repeatedly verified mag center and coupling hub spacing. But, the problem continues.

Some of these installations have experienced repeated coupling disk-pack failures. The failures include cracks in the outer disks at the edge of the washer. The coupling manufacturer’s literature would classify this as a failure indicative of excessive radial misalignment.

I suspect that two issues may be coming into play with these problems. First, I am suspicious that the vertical thermal growth of the new motors is significantly different than the old motor. The arrangement with the cooling fan on the non-drive end blowing toward the drive end seems like it would have a strong impact. Just placing my hands on the motor feet and end housings, I can feel a big difference in temperature between drive end and non-drive end. We should be able to take support temperatures and recalculate our cold alignment offsets to adjust for this. The flow path of the cooling air may also be affecting the thermal growth of the driven pump. The air blowing against the coupling end pedestals is probably cooling them down more than the thrust end pedestals. With the motor hotter on the inboard and the pump hotter on the outboard, a significant misalignment may be occurring in service.

The other issue I have considered is axial thermal growth of the motor rotor. Some of these motors are very long (perhaps as much as 8 feet long shafts). Even through the motor is a sleeve bearing design with about one-half inch mechanical float, there has to be some issue with axial growth of the motor rotor and shaft toward the coupling. If the magnetic center locks the rotor at the center point, then half of the axial growth would be directed toward the coupling and half would be directed away from the coupling. We do not pre-stretch these coupling spacers to accommodate this axial growth. I am beginning to think that we should. I will probably ask our mechanics to stretch the coupling spacer on one of these machines by 0.030 inch to see if that stops the shuttling.

I am interested if anyone else has seen similar problems when converting from WPII to TEFC motors. And, I would like for motor experts to comment on my assumptions regarding vertical and axial thermal growth. Any help would be much appreciated.


Johnny Pellin

 

[rhatcher]

JJPellin,

Axial hunting is not uncommon on large 2 pole TEFC motors with sleeve bearings when running uncoupled or running coupled with large axial clearance in the coupling. As described in the Reliance motor primer, the ODE cooling fan forces the rotor away from magnetic center. As the rotor moves away from magnetic center, the centering force increases and the fan aerodynamic force decreases until the point that the centering force overcomes the aerodynamic force. The result is usually an oscillation rather than operating at a point of equilibrium. This is due to the non-linear nature of the forces involved, the relatively large inertia of the rotating assembly, and the fact that the sleeve bearings offer no resistance (friction) to axial movement.

As a note, in the extreme case the fan will completely overcome the magnetic centering force and the rotor will thrust completely towards the ODE and will run against the bearing thrust face. I have seen a motor fail on commissioning because the millwright performing the alignment took this to be magnetic center.

 

[electricpete]

But magnetic force is not a linear spring. In the latest link it suggests Fmag = Kmag*(x-xmagcenter)^0.75. The slope of this function Fmag(x) vs x (representing the effective spring constant from magnetic force) is KmagEffective = d/dx(Fmag) = 0.75*Kmag*(x-xmagcenter)^(-0.25). This effective magnetic spring constant KmagEffective is highest when x-xmagcenter =0 (when actual position x is near magnetic center xmagcenter) and decreases as x moves away from xmagcenter. If the combined action of coupling and magnetic force is an equilibrium position x different than xmagcenter, then the effective spring constant from magnetic force is lower. If coupling spring force is weak and we're relying on added magnetic centering force to limit response to perturbing forces, then it is not as effective when the equilibrium position is off of magnetic center. That represents a more tenable explanation although I'm not sure it's the whole story.

Attached I tried to provide a graphical depiction of above under assumption the magnetic force is given by the expression above the coupling spring is linear. The particular numerical values I picked out (*) were xmagcenter = 0.2 (offset from the coupling force acting at x=0), Kmag = 1, Kcoupling = 1, also coupling spring completely linear (big assumption... I imagine it stiffens as it deflects). * These values are 100% fictional just to make the plots look good for illustration, I made no attempt to figure out realistic values. The first graph attached is Fmag vs x. It looks fairly much like a straight line everywhere except around x=0.2 where there is an inflection point. The next graph is the slope of the first labeled KeffMagnetic... the effective spring constant of magnetic force. It has a peak at x=0.2 which is MATHEMATICALLY infinite according to this form of the equation. Obviously the form of the equation must in reality change somewhere near this singular point.. The next graph shows magnetic force in red (equilibrium if this was the only force would be the zero-crossing at x=0.2), coupling force in green (equilibrium if this was the only force would be the zero crossing at x=0), and the sum/combination in blue (equilibrium is the zero crossing at x~0.1). The final graph is again the spring constants where the combination spring constant is sum of the two.

EDIT - I should have defined the forces with a negative sign: F = -Kcoupling*x and Fmag(x) = -Kmag*(x-xmagcenter)^0.75 so that all the forces act toward their equilibrium point instead of away from it. Use your imagination and flip the graphs horizontally.

What does all that prove? Nothing much other than illustrating that the functional form of the magnetic force given in the paper suggests there is an increase in the effective spring constant "near" the magnetic center. How near is near... would have to put some better numbers to it to characterize that better. Again, just trying to piece together a story that makes sense but I'm not sure I have the whole story.

Brainstorming what might be done, especially focusing on the mechanical side:
* stiffer coupling would help (if it's all about stiffness to resist aerodynamic forces as the articles seem to be suggesting)
* add an antifriction bearing in the housing on the outboard end if you have room (suggested in one of the articles)
* change the fan pitch was mentioned in attempt to reduce axial force to make the blades closer to a radial fan vs axial fan and reduce fan thrust. It might even be possible to go completely to a radial fan if the shroud shape (possibly redesigned) is capable or redirecting radial flow into axial flow. Or I can imagine using more powerful radial fan in attempt to compensate for reduced efficiency of the shroud/fan in delivering axial flow. Safe to say it requires some study by someone that knows what they're doing more than me.
* your practice of figuring out magnetic center the best way possible (by removing external fan) might help get it closer to magnetic center which should still help..
* your idea of offsetting from magnetic center when coupled in anticipation of what changes might happen would help in theory to the extent you can predict what thermal changes occur during operation. If you were lucky enough to be aware that there are remaining unbalanced fan force that cannot be removed when you scribe magnetic center, then you might attempt to compensate for those too when you establish coupled position, in attempt to land on true magnetic center during operation for strongest magnetic centering force.


=====================================
(2B)+(2B)' ?

 

[electricpete]

EDIT - I should have defined the forces with a negative sign: F = -Kcoupling*x and Fmag(x) = -Kmag*(x-xmagcenter)^0.75 so that all the forces act toward their equilibrium point instead of away from it. Use your imagination and flip the graphs horizontally.

Fixed in attachment here

=====================================
(2B)+(2B)' ?

 

[electricpete]

Below is linked an article which goes to considerable effort examining many variables involved in magnetic centering force.
"The Influence of Axial Magnetic Centering Forces on Sleeve Bearing Induction Motors" from IEEE/IAS/PCIC conference 2006

http://ieeexplore.ieee.org/document/4199057/

Safe to say, the IEEE article is nowhere as simple as the x^0.75 relationship given in the EASA-based document above.
I don't fully understand what's going on the IEEE document and I can't post it (it's copyrighted... although it's free to me as member of IAS).
The simplest expression that I could find for relative variation of magnetic centering force vs position was given by equation 3 of the IEEE article:
F(hu) ~ dLeffective/dhu = 1-2/pi*arccot(hu)
where
hu = h / g
h = offset from center position
g = gap distance accross the airgap
I think this equation applies to a situation of no ducts and identical length rotor and stator.

Attached I tried to compare this simplest IEEE expression to the previous one from EASA. By changing the relative scaling of the two forms I couldnt' make them match at large values of hu but I could make them match pretty closely for |hu| < 1. Maybe the EASA expression was a more approximate expression for use on small distances up to one airgap distance offset. Also when we take derivative of F to compute effective spring constant Keffective = dF/dx, the IEEE equation does not suffer from the problem of going to infinity at x=0 like the EASA equation did.

These aspects are plotted attached (IEEE red, EASA blue). I'd say it's nowhere near perfect, but it seems that the red IEEE curve in the 2nd plot is a better starting point to understand how the magnetic stifness falls off with distance (for this particular case no ducts, rotor and stator same length) than the blue EASA curve. Max stiffness at center, about half that at 1 airgap length offset, about quarter that at two airgap lengths offset. Obviously anyone reading this would be waaay better off going directly to the article than accepting my interpretation of it from my limited understanding. And there's a whole lot more there in that article.... particularly a lot of examination of effects of many variations of rotor and stator vent duct configurations.

=====================================
(2B)+(2B)' ?

 

[Sparweb]

I read the quoted material that EP found from Baldor and it made perfect sense to me. They attribute the hunting (of an unloaded rotor) to variable fan thrust. They also associated it with the centering force on the rotor. In my reading, when the rotor is centered, the stabilizing force is at a *minimum* and the shaft must be forced off-center to develop any magnetic force that will return it to center.

So (according to the Baldor document alone) you have an oscillating thrust from the cooling fan, pushing against a weak centering force from the rotor, which requires a significant displacement (your stated 1/8") before the force can arrest the fan thrust force. When the fan's thrust breaks down because the shaft has moved, the thrust drops and the shaft returns for another cycle.

I also agree that the fan thrust can be variable, because the TEFC motors I've used (admittedly, up to a piddly 10HP!) the effectiveness of the fan is completely dependent on its proximity to the stator housing. If that fan moved away from the housing, then it would lose effectiveness, static pressure on the outlet would drop, upstream the static low pressure on the inlet would drop, and the thrust force would be reduced. I never considered this a problem because small TEFC motors I've used do not have enough axial play relative to the gap between fan fins and the stator housing. It sounds like yours does. If the fans in your TEFC motors look anything like the small ones (not a safe assumption for me to make) then the fan blades may also be swaying all over the place, adding a *third* mode of displacement, driven by the shaft oscillation.

Take another look at your cooling fans.



STF

 

[electricpete]

In my reading, when the rotor is centered, the stabilizing force is at a *minimum* and the shaft must be forced off-center to develop any magnetic force that will return it to center.

At magnetic center, the magnetic centering force Fmag is at a minimum (0) but the associated effective stiffness is a maximum (stiffness is the slope of the Force vs position curve). I believe this variation in effective stiffness (as position changes) is the reason why the magnetic force provides the most effective stabilization when on-center where effective magnetic stiffness is maximum. That is the one-sentence punchline for all the rambling of my last few posts.

Sorry for monopolizing this thread. Good comments about external fan movement… makes good sense to me fwiw.


=====================================
(2B)+(2B)' ?

 

[JJPellin]

Thanks to all of you for the very interesting discussion. Much of the electrical details went right over my head. Much of the literature on the subject focusses on axial shutting when the motor is unloaded or uncoupled. Strangely, our motors are perfectly stable when uncoupled and only shuttle when running fully loaded, coupled up to a pump running at full rate. Our shuttling is generally about 1/16" rather than the 1/8" referenced by electricpete. I am going to add a pre-stretch to our alignment practices for these motor as a first attempt to resolve this. I am going to ask for documented magnetic center on new motors established with the cooling fan removed.

There has not been much discussion about vertical thermal growth on TEFC motors. I have always considered vertical growth on motors to be uniform on drive end and non-drive end. For these larger TEFC motor, I don't believe that is appropriate. I will get temperatures on the supports and calculate the thermal growth and make changes to our cold offset alignment targets.

Thanks again,

Johnny Pellin

 

[electricpete]

I have seen dozens of instances of axial shuttling in the past 25 years. At this moment, we probably have 3 or more that are chronically shuttling. We have solved a number of these by using improved practices for establishing magnetic center and careful adjustment of the coupling hub spacing. For some of these newer TEFC motors, we know the issue. When running the motor solo, the fan pulls the rotor off center and we mark that position as mag center. For these we have to remove the cooling fan and solo the motor to get a true center. I have considered just ignoring them if I could believe that this motion is not harmful to the motor, pump or coupling. But, it freaks out the operators and it seems like it should be easy to resolve.

As a point of reference, we probably have 2000 motors in operation. There are probably 200 or more with sleeve bearings. These motors were manufactured as long ago as 1955 and as recently as last month. In the past, our TEFC motors tended to be the smaller ones. Now, we are buying much larger TEFC motors and the problem is becoming more common.

More curiosity:
In your previous history (aside from the modified motors you descrbed in this post):
Is hunting more likely to occur while coupled/loaded or while uncoupled?
Have you seen it on any motors that don't have axial fans?


=====================================
(2B)+(2B)' ?

 

[JJPellin]

I am only aware of any axial shuttling on motors running coupled and loaded. We have seen this on WPII motors that I don't think have axial fans. On those motors, we tended to find problems with magnetic center that were more extreme. In one instance that I can think of, the mechanics set the motor on the center of mechanical float. On a later check, we found that mag center was about 1/4" different than mechanical center. When we properly positioned the motor rotor relative to mag center, the shuttling problem went away. In another motor, we changed from a sleeve bearing design to a ball bearing design because we could not get it to stop hunting.

Johnny Pellin