Coupling Seizure 3

[bhaskar5150]

In our DCU plant, we have 2 nos of FD fans (identical) for supplying combustion air to the burners with the following data :
Flow : 10.896 m3/sec, Static pressure :4700 pa (479 MMWC), Rated power : 61.25 KW
Temp : 35C, RPM : 980, Motor rating : 75KW. Impeller type : Backward curved.

Each fan has a IGV at suction for flow control & discharge damper is full close or full open & has a full close command on individual motor trip. Heater has a trip on low combustion air pressure in individual ducts (4 passes) to the burners.(25MMWC trip pressure against a normal of 34MMWC in individual passes).

Fan-A has a inherent problem of coupling failure (more than 3 times, last one in Jun'16).It is a practice here to run both fans in parallel since its commissioning. Present flow is around 6.6 M3/sec & static pressure is 410 MMWC. This is done by throttling the IGVs of both the fans.(Control is single command for both IGVs) In all previous cases, the coupling bolts & shims break getting uncoupled with the motor but motor still running (doesn't trip), therefore doesn't closes the discharge damper & flow from Fan-B goes to Fan-A & reverse rotates the fan. (This is obvious & only a after effect of coupling seizure). Then, discharge damper of Fan-A is closed manually.

Another important fact is that normally after every start-up, one fan (Fan-B in last 3 cases) is started first as the flow requirement is less initially & after a certain time interval the 2nd fan (Fan-A in last 3 cases) is started and IGVs of both the fans are adjusted accordingly to get the desired flow & static pressure.

From above, it can be concluded the Fan-A starts against a back pressure (created by the static pressure of Fan-B). Is it because of this that the Fan-A never recovers under parallel operation, operates on the left of the peak pressure of the fan curve and goes under stall. Due to this flow reversal in fan-A, coupling shims & bolts breaks as they are weakest components in that line. Also, the stalling zone is not defined by the manufacturer in the individual fan curve.

Also, there is no vibration issues reported & so the factors such as misalignment, bend shaft, eccentricity, unbalance may be ruled out.

If above is the cause, how to know & confirm it.

We are finding it very difficult to access the parallel performance of the fans and have no idea how the combined curve would behave under the present operating conditions.

Any advice/suggestion to get a permanent solution to this problem.

Thanks in advance.

 

[electricpete]

So from the evidence of a bulge in A but not B it suggests A is currently loaded higher or maybe has been overloaded in the past and taken a set. It doesn't directly suggest anything about flow/torque oscillation or reversal that I have seen. (will be interested to see those photos if you get time)

The main evidence you are seeing to support a theory of flow or torque oscillation/reversal I think is the mismatch between predicted and actual total flow? (along with theory that it's related to coupling failures based on starting seqeuenc). Is there any other evidence of flow or torque oscillation/reversal?

The scenario further you suggest is pump B carrying full flow and pump A carrying low flow due to stall. I was trying to visualize that from curves where identical fans can carry different flows at the same dp. It seems more likely for forward curved blade (with hump in the curve) than for backward curved based on the qualitative curve shapes posted here

http://lh5.ggpht.com/_1wtadqGaaPs/TGrQc7fTvcI/AAAAAAAART0/fb4Eakt3BYM/s1600-h/tmp1A999_thumb5.jpg

From that curve also I assume power is bhp (vs fluid hp) and would expect different BHP at the two operating points and therefore different motor currents. But you reported same current on both fans. Maybe for some machine it would be possible to operate at two vastly different operating points with exact same current but that would seem to be a lucky coincidence. The simpler and less coincidental interpretation of the currents would of course be that the two fans are sharing load equally (although that doesn't explain why coupling A is bulging more, unless it was permanently deformed from prior overload or maybe was assembled improperly with fewer shims)

But how to prove this point practically at this moment.

Compare the vibration patterns (and also listen). If one is stalling I'd expect some difference in pattern. IF on the other hand you find no difference in vibration pattern or noise, then based on that info (along with previouisly reported facts that there is no difference in the two current magnitudes and no current oscillation) then maybe start looking for other explanations.


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

 

[keyepitts]

What about axial thrust loads? If the one fan is running in reverse, wouldn't the axial thrust also be reversed? Is the fan thrust bearing designed to handle this? Or does the shaft maybe move a bit, deflecting the coupling shim pack excessively, then the motor starting up suddenly loads the deflected coupling until the axial load reverses and the fan shaft/ coupling returns to normal position. Just an idea, I'm not familiar with the construction of fans of this size.

 

[Strong]

El-Pete,

I can't accept most of your comments about effect of reverse rotation not affecting startup torque. I guess all of the anti-reverse rotation devices for fans and pumps should not be needed, if your comments were correct; or am I misinterpreting them? The peak torque transient depends on inrush current (electrical force) and the resistance to motion (inertia) of the motor and fan rotors. Reverse rotation (free-wheeling) of the fan adds more inertia, since the rotor must be decelerated before being accelerated in the correct direction. In addition to excessive peak starting torque, the coupling may see excessive torque during acceleration to full speed when passing through a torsional resonance. A torsional resonance at full speed can go undetected by relying solely on motor current measurement. I still say that reverse rotation, if occurring, is most likely cause of coupling torque overload during startup torque transient.

Walt

 

[electricpete]

I can't accept most of your comments about effect of reverse rotation not affecting startup torque.

Hi Walt. I respect your opinion. I don't want to be argumentative but I'd like to talk through the details point by point.

I guess all of the anti-reverse rotation devices for fans and pumps should not be needed, if your comments were correct; .

Reverse rotation causes a variety of problems. It takes the motor longer to get up to speed, drawing higher current the whole time which challenges the motor more and can lead to motor trip on thermal protection. I’ve heard it can unscrew certain impellers although I never understood that part. Reverse rotation with deenergized motor can in theory overspped the machine. There are a lot of things the device can do. Existence of reverse rotation device doesn't prove it's intended to prevent couplinlg failure.

peak torque transient depends on inrush current (electrical force) and the resistance to motion (inertia) of the motor and fan rotors. Reverse rotation (free-wheeling) of the fan adds more inertia, since the rotor must be decelerated before being accelerated in the correct direction.

Inertia depends on mass and geometry, it does not change. I can’t make much sense of your point.

Let me show you my way of thinking about it. Here my quasi-static analysis of starting. Consider it as a 2dof system:
Jmotor ==> Kcoupling ==> Jfan
Where J is rotating inertia, Kcoupling is coupling torsional spring constant.
Tmotor is em torque applied to motor
Tfan is fluid torque resulting from interaction with the fluid
Tcoupling is torque transmitted through the coupling
W = rotational speed in radians per second
quasi-static assumption: It includes two simplifying assumptions: 1- the entire train rotates at uniform w (neglects oscillatory behavior such as torsional resonance); 2 - assumes motor torque is a pure function of speed – neglects obscure oscillatory electric torque associated with dc component during motor start.

Acceleration equation at motor inertia and at fan inertia:
dw/dt = (Tmotor-Tcoupling)/Jmotor = (Tcoupling-Tfan)/Jfan
Solve Tcoupling = (Tmotor*Jfan+Tfan*Jmotor)/(Jfan+Jmotor)
where
Tmotor is electromagnetic torque. Under this quasistatic assumption it is the one given by the torque speed curve. We extrapolate it past the zero speed axis… most likely it is lower than locked rotor torque during reverse rotation.
Tfan is opposite polarity of Tmotor and so Tfan only serves to reduce the peak torque in the coupling during start when Tmotor>Tfan.
This quasistatic analsysis does not suggest any increase in peak torque from reverse rotation.
I previously tried to consider the limitations of the two quasistatic assumptions listed above, in the following bullets:

*I don’t' see that the potential for torsional oscillations is significantly affected by reverse rotation.
*The torque we talk about during starting represents an average over one or more power cycles… there is actually a large superimposed ac component of torque varying at line-frequency which tends to result from the dc offset in stator currents after start. I don't see a significant change in this.

These last two bullets are open to discussion…. the trickiest part of the analysis imo (the earlier part was easy).

In addition to excessive peak starting torque, the coupling may see excessive torque during acceleration to full speed when passing through a torsional resonance.

Yes. But no obvious difference for the start from reverse rotation or from standstill. If we look at speed vs time of motor accelerating from reverse rotation considering only the portion after it passes through zero speed, it looks identical as speed vs time curve of motor starting from rest. Both motor pass through any torsional resonance at the same rate. The only difference is some resonance the reverse rotating motor may have passed through before it got to zero speed and I have no reason to suspect such torsional resonance would be any worse than the torsional resonance between zero and full speed.

A torsional resonance at full speed can go undetected by relying solely on motor current measurement.

I agree. I don’t see that full speed torsional resonance has much to do with reverse rotation though.
LATE EDIT - I see now you are on a different point - you are suggesting there can be operating torsional oscillations placing stress on coupling without causing current oscillation? I'm not sure about that, have you seen it?

I still say that reverse rotation, if occurring, is most likely cause of coupling torque overload during startup torque transient.

I don’t rule it out. But I haven’t seen a compelling argument that this is expected in general during a start from reverse rotation.

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

 

[EdStainless]

As for checking speed you can use a vibration type tachometer (google it).
Hold it against the motor or fan housing and watch the vibrating reeds.

The other thing that happens with starting in reverse rotation is that the higher momentum of the fan requires max torque for a much longer time and it also causes more shaft wind-up and deflection. I have seen shafts twisted, bearings damaged, and couplings destroy from this. And this was in systems where each part had been tested to >2x locked motor torque.

I would start against a closed damper and then open once the fan pressure exceeded the existing system pressure.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[bhaskar5150]

Thanks EP and strong!!! Nice to see all your comments and analysis on the subject matter. Good lesions learned for me.
In simple terms, what I could understand is that as per strong's view, coupling failure might be due to torque transients during startup against reverse rotation and torsional oscillations at full speed not detectable through current measurements(current is steady at full speed). In contrary to that, as per EC's view, torsional ossilations are not affected by reverse rotation during start up and for torque reversals to happen at full speed (load), current should also vary, but in actual current is steady.
Here I would like to bring one point to your focus that the coupling is failing only after running a substantial period of time (min 4 months)after startup, this potentially signifies that something cyclic is happening (motor and load against each other) and eventually, coupling is the pray.

But if strong's point is correct, then strong may comment what other things we should consider/check (apart from current)at this stage to confirm occurrence of torque reversals or any possible reasons why torsional oscillation may not get detected through current at full speed (load). Possibly, please also think from the fan behaviour point of view under parallel operation.

 

[EdStainless]

El-Pete
I believe that the underlying assumption that most of us are making is that if the coupling material or alignment is damaged during startup then you will greatly shorten its running life. As much as we would like these couplings see significant load and stress when running, if they didn't we wouldn't need them we would hard couple machines.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[electricpete]

Thanks ed. I still don't understand what would be the mechanism, but if you have experience seeing couplings damaged from starting during reverse rotation, that counts for a lot to me.

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

 

[gruntguru]

It would be useful to see a flow map for these fans and the relevant operating point. Is it possible that the two fans are operating on either side of the pressure hill ie same speed and pressure but the lower flow case is in the surge region?

je suis charlie

 

[Tmoose]

Does your vibration group have a strobe light accessory in their analyzer/data collector kit?

http://www.monarchinstrument.com/product.php?ID=32

A strobe tach would -
1 - allow measuring the reverse rotation rpm
AND
2 - allow checking the coupling condition right after each start up
AND
3 - significantly boost your rep as a cool dancer

https://www.youtube.com/watch?v=me7Vr6Tzxa4

 

[desertfox]

Hi bhasker5150

I'd like to see some pictures of the bolts that failed in both side view and full on the fracture face, particularly as this is a cyclic failure.

What's not been mentioned as yet is how the coupling is replaced in terms of methodology for example:-

How often is the coupling inspected?
When tightening a new coupling, do you use a torque wrench and follow a star like pattern whilst tightening the nuts?

Now these failures may well be caused by the working torque however a bolt which is incorrectly tightened could quite
easily contribute or even be the root cause of these failures.

So from my point of view, I'd start with my maintenance procedures as these should be fairly easy to check.

desertfox

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

 

[byrdj]

I only noticed couple of mentions of alignment of the fan to the motor. the coupling is design to allow for MINOR misalignments and account for alignment changes that occur during operation. Normally the desired COLD aligment will have an offset to allow for expected thermal growth and become FAIR coupling during operation.

I have seen where the "alignment" taken with laser tool showed good, But a straigh edge and latter dial indicator showed it out over 0.2"

 

[Tmoose]

"I have seen where the "alignment" taken with laser tool showed good, But a straight edge and later dial indicator showed it out over 0.2" ."

One real possibility is one or both coupling hubs were not bored concentric to the OD.
The laser setups I used worked strictly off the shaft centerlines, with no real good method to even check for shaft straightness or hub eccentricity.

 

[CouplingGuru]

Gents,

The only reasons drive bolts break on a disc coupling is because of severe over torque, or bolt fretting from loss of clamping force on the joint. Misalignment will not cause a bolt failure, that will cause disc pack failures. Sounds like you have a rotating inertia to overcome, that could cause the motor starting torque to spike at start up. I generally don't recommend over sizing the coupling unless that is truly the weakest point. So this may be an easy fix if the coupling has a very low service factor like 1.25 or 1.5. Then just go to the next size up coupling. However, most of the time that is not a good idea, that will just transfer your problem to another area, like instead of breaking drive bolts you will be breaking drive shafts.

And the easiest way to measure total misalignment is to measure the gap between the center spool and the hub flange. Take this measure measurement all the way around the coupling, the total difference between the largest and smallest dimension will give you your total pack misalignment by doing some simple trigonometry. Tan^-1[(Total difference/2)/(Coupling Disc BCD/2)] = Angle per pack misalignment. That value will sum all parallel and angular misalignment into one value.

http://www.psccouplings.com

When it comes to couplings we are always here to help.

http://WWW.PSCCOUPLINGS.COM