11kV Motor failing to start 2

[allan789]

We have a 11kV, GEC, 6.5MW motor which is connected via a gearbox and pinion to a ball mill. The motor is soft started by a LRS.

About a year and a half ago, April and again now yesterday the mill has gone for a start and has sat there not turning. It has sat there pulling 1800A until the protection eventually triped.

In all cases the motor and LRS has been inspected with nothing found. The mill was then inched by a seperate inching motor with no apparent mechanical windup or resistance. The mill was able to be restart after the inspection each time.
The LRS conductivity was measured at 14mS as compared to a design value of 12.7mS.
We have also had the motor reciently tested by a motor test company with no obvious abnormailties.

We are at a loss to what could be the cause of this intermitant fail to starts and any suggestions would be greatly appriciated.

 

[edison123]

For a slipring motor, you're pulling almost 6 times the FLA at the start, which is quite high. Suggest you recheck the connections from the rotor to the LRS (brushes, springs, cable terminations at both ends).

Muthu

http://www.edison.co.in

 

[rbulsara]

The mill was able to be restart after the inspection each time.

.

Are you saying that the motor started every time after "inching"? If so that is the normal procedure, especially when mill has sat idle for a while.

Rafiq Bulsara

http://www.srengineersct.com

 

[Drivesrock]

GEC Motors and Generators are nowadays Converteam, well since about 2003. You could try your local Converteam rep to get them to check out the motor and starter based on the original project. Where is the motor installed?

 

[electricpete]

Muthu’s suggestion makes sense to me. If you had intermittent open on one of 3 phases of rotor, I think you would develop no torque (like single phasing stator of a 3-phase induction motor).
Maybe rotating it is enough to reseat a brush.

Less likely, I have seen sleeve bearing machines intermittently bind up upon starting if they have been sitting stationary for awhile. But breaking them loose was not easy.


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

 

[waross]

I have had to roll 3000HP motors before starting. Sleeve bearings dry out in as little as 12 hours. If you have sleeve bearings I would roll the motor with the turning gear before each start. Leave it on the turning gear until the turning gear amperage drops and stabilizes.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[allan789]

The rotor connections have been tested with the results attached, there is some small variences but we believe they are within tolerance. We havent done an internal tank inspection as yet. I have attached the multilin waveform capture which captues the end of the last event, it shows stator phases and differential CT's. I do note that there is differential current and indeed the last trip was on the differential current not over current.

The mill was installed 2002, the motor was replaced around 2 years ago. From the history ive herd this problem has not occured before than.

There is 13 brushes per phase so to me it seems unlikley that all of the brush connections would go bad.

The mill has to be inched if is the mill has been stationary for longer than around an hour. This is mainly so that the ore and ball charge to not become 'frozen', if the mill is started with a frozen load and the frozen load sticks together until it is on the top of the mill then falls it will cause severe damage to the mill. Indeed in this case the mill was inched up until around 2 minutes before the mill was started.

I did forget to mention in my first post that this has occured after shutdowns or long downtimes (6 hours).
One of the theorys floating around is that the sodium carbonate is settling out of solution and so the motor tries to be started with a less conductive solution and thus does not develop sufficient torque. The circulation pump is run for an hour or so before starting under this assumption. Personally i dont think this is probably it as sodium carbonate is quite soulible and i have not herd of it coming out of solution before. has any one else herd of that type of senario?

 

[electricpete]

So, part of what you are looking at is: what changed.

One thing to think about is voltage. Does your incoming voltage vary?

It appears you have about a 20% droop in stator voltage during the start. Is that normal for this motor?

How many poles (what is synch speed)? It occurs to me if more than 2-pole, we could do a d-q transform and look at the oscillations. Line frequency oscillations would be expected/normal. Oscillations at line frequency times number of poles pairs would indicate unbalance in the rotor circuit. The calculation is worthless if 2-pole motor, so I'd like to know before I try.

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

 

[electricpete]

Also the traces show a stator current that decays away very slowly after the motor trips. That is odd. Is it a delta-wound motor with CT's monitoring inside the delta (that would seem unusual). Otherwise I can't immediately see why we expect slowly decaying stator current indication after the trip.

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

 

[electricpete]

If the problem seems to occur after long shutdown, another possibilty is that the required mechanical torque is higher during that time for some reason... either something about your driven equipment/process, or else less likely something about sleeeve bearings (Bill mentioned drying out in a matter of 12 hours... our problems came after months of sitting idle).

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

 

[electricpete]

Attached I have shown some traces in a more readable format.

Slides 2-4 are voltage and current for phases A, B, C. Current lags voltage as expected (I assume the data included was phase to neutral voltages... labeling was ambiguous but magnitudes and phases match expected phase to neutral voltages).

Apparently where the trace begins at t=-0.25 sec, the motor has already been energized for some period (the dc transient is gone).

At around t = +0.05, the motor trips as evidenced by decreas in current and increase in voltage (by about 20%). The peculiarity previously mentioned is highlighted in slide 1.... phases A and C don't stop immediately but decay to zero in equal/opposite fashion. I don't know what that's all about.

Slide 5 is the d-q variables. I set the frequency to 50.07. There is also a shift in the entire time axis of 0.0556 seconds for eveyrthing which lines up phase A with a cosine function and creates the expected pattern: Vq is positive dc while Vd is zero.

Id (the component in quadrature with applied voltage) is much higher than Iq (the component in phase with applied voltage) as expected since the load is very inductive. The traces are very bouncy ... I think in part because the sample rate is very coarse (600hz) and perhaps I have not recreated the exact frequency perfectly (although 50.07 is the best match).

What we see at the beginning of the waveform is line frequency (50hz) oscillation in Id and Iq. That is a normal feature after energization (it corresponds to the dc offset in the phase currents, but paradoxically disappears much lter than the phase currents in simulations). There is also a growing 100hz component that starts out barely perceptible at -0.15 seconds, and dominates the waveform just before trip at 0.05 seconds. I'm sure there can be lots of causes of that which I'm not aware of, but the one thing I was looking for was "pole pass frequency" which for a locked rotor condition would be poles/2 * line frequency. If this is a 4-pole motor, this could be indication of a rotor-side imbalance which is growing substantially during this period.

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

 

[electricpete]

Just to avoid confusion, I should mention that the time scale of slides 1,5 and 6 is shifted upward by 0.0556.

So the trip happens just after 0.05 seconds in slides 2-4, but it happens just after 0.010 seconds in slides 1, 5 and 6.


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

 

[electricpete]

Actually, 2*LF is pole pass frequency for a locked rotor condition regardless of number of poles.

Fp = 2*s*LF
For s=1, Fp = 2*LF

Another way to say it
Fp = p *(fsync - fmechanical)= p * (fsync-0) = p*fsync = p*(2*LF/p) = 2*LF

The frequency indicating rotor circuit unbalance in a locked-rotor machine is 2*LF (i.e. 100hz) for any number of poles


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

 

[electricpete]

Although I am certainly at a loss to explain why unbalance didn't seem severe at the beginning, and why it would be growing over time.

Take it for what it's worth... analysis for your consideration. You get what you pay for ;-)

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

 

[electricpete]

Then again there is the differential current and associated trip. I don't need to tell anyone that on the surface that suggests a stator fault.

Attached I have added the differential currents and ground currents plotted as slide 7. They don't appear anything near sinusoidal or periodic with line frequency.

We combine this with the observation of the decay of A and C phases after the trip, it builds a puzzle. Actual stator fault or some kind of malfunction in the CT circuit?

Perhaps you can describe the stator windings (wye or delta) and the location of the CT's.

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

 

[electricpete]

Whoops. Attached is the revised powerpoint with differential and ground currents added as slide 7. (Note there is a different scale for A phase than the others as indicated in the legend on right side of plot).

Also you said you had tested rotor and attached some results... but I didn't see that in the attachment (just the fault recording).

Seems like some electrical testing of stator and rotor and perhaps CT circuits are all in order.

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

 

[allan789]

Thanks for your posting electricpete, you have obioulsy put alot of time and effort into your above postings

its a 6-pole machine.

20% does seem exsessive to me also. I just did a quick simulation through PTW and at 5 times FLA it predicts 21% voltage drop for our system. For a normal start (2*FLAish) i certianly would'nt expect 20% voltage drop.

Its a star-wound motor with CT's on the line and star side of the stator winding. There is rotor CT's at the LRS but unfortunately they only feed into our DCS and thus their trend is only in minutes and does not help. If the fault were to happen more predictable we would set up our data recorder to capture this, but we use it so often its hard to leave it in the one spot for months.
I agree with your comment electric pete in resgards to the slow decay of the motor current, i am at a loss to explain this also.

As for the recording the multilin relay has only a limited amount of waveform capture. The motor is energised for approx 5 seconds before the current finally builds up enough to trip the motor. The multilin only records the final section of the fault.

The motor was tested by a test company in july with the results attached (including LRS tests which i failed to attached earlier!). The test company we use dont have any answers for us and dont seem too concerned, which does not help us!

 

[edison123]

5xFLA at start indicates that the motor is no longer a slipring motor => SS liquid has become fully conductive or the shorting contactor is closed.

Muthu

http://www.edison.co.in

 

[LionelHutz]

Is it possible the motor could have been energized with the LRS not in the start position? I'm thinking part way through the travel so the resistance is reduced a fair bit compared to where it should be on a start. If you decrease the rotor resistance too much then you will really decrease the locked rotor torque.

 

[electricpete]

For any scenario we come up with, we can come up with a counter question that seems to contradict it. If the problem is on the rotor side, then why did we get a stator differential trip? Perhaps we might say CT saturation from higher than expected currents... but why does the A phase differential current seem to grow over time? And why do decaying A and C currents remaining after the trip like that?

Or if we say the problem is on the stator, how do we end up with balanced currents? And still can't quite explain that A and C behavior after the trip.... seems to require a loop which includes A and C phase CT's. Can't even explain that with phase to phase fault... unless the CT's which are used to provide phase current are the ones on the neutral side (vs wye?).

Looking again at A-phase differential current, we can say it has a very definite component of current at 2*50hz = 100hz througout. What on stator side would cause that??? Nothing on the stator side that I can think of... Would seem to point toward rotor side I think but just another piece of the puzzle.

Was the electrical testing done after the differential trip or before?
Has another successful start occurred after the differential trip?
Is this a new installation, or has operated fine for awhile and then just started acting up (I assume the latter from some of your comments, but just wanted to verify).


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