Inexplicable motor failures 3

[eeprom]

Hi,
I have two pumps, both 500 Hp, running on VFDs or across the line in bypass mode. Both motors have a history of failing windings. Both motors are inverter rated. There are no RTDs on either. Both motors are within 150ft of their VFDs, and neither motor is using VFD cabling. Both motors have two breaker in series. Breaker 1 is in a GE MCC and has a GE microversatrip with long and ground protection. Both long time settings are about 1.15X with a trip time of 5 seconds at 6X, both ground trips are set to 100A. Breaker 2 is an ABB LI breaker, with the long set at 1.2X FLA and the I set to 8X plug rating, which comes out to 6400A. Downstream of the ABB breaker are contactors for VFD operation and across the line starting. The VFD also has fuses internal.

Recently, motor #2 tripped twice on overload while running at a very low mechanical load. The trip was occurring in breaker 1 in the GE breaker in the MCC. The relay indicated the overload element had tripped. We meggered the motor and everything appeared to be fine. I did not get the megger readings, but there were in the hundreds of megs.

We started the pump (motor #2) up again, in bypass mode, and kept it at low load while we measured the phase currents. They were about 105% of the FLA, and there was some imbalance, about 2%. The pump was running at about 1/5 or 1/6 of it's rated load, but the motor was drawing more than it's FLA.

I asked to switch pumps so we could see what that current reading on motor 1 would be for the same mechanical load. However, that motor (motor 1) tripped upon starting, also in bypass mode. Again it was the GE relay that tripped, not the ABB. I did not see this, but I was told it tripped immediately. According to the GE relay settings, to trip immediately had to be caused by the ground fault element. The GE relay has no short time element, and the instantaneous element has been turned off.

We meggered motor 1 and it came in around 500 megs on all phases. We meggered again including the cables from the output contactor down to the motor and got the same. Phase to phase measurements were all similar, around 0.6 ohms DC resistance, but the meter used is not very accurate at that low of a reading. This is a fluke 287, and I don't think it's accurate to 0.1 ohms at that load. The calculated impedance of the motor windings, operating at pf of 0.84, results in a per winding impedance of 0.492/_32. Therefore I believe the fluke readings are reasonably valid.

We then tried hand rotating the pump and it rotated easily. There is no binding.

This is as far as I got. I can detect no ground fault, but it appears we are tripping on a ground fault. Motor 1 won't start because it faults. Motor 2 is running in its service factor while operating at 20% of its load. What I believe is that the motor windings are being deteriorated by spiking from the VFDs. As the winding insulation breaks down, some of the windings become shorted, and it then takes more current to create the same torque. The increased current causes heating, further breaking down the insulation, and then eventual failure.

This is happening to both motors. It seems it has be to related to the VFDs. However I cannot find anything wrong. What I am hoping for from the forum is someone who can help me calculate how spiking and reflected waves can shorten motor life. Or, point me in another direction.

Thanks for your help.

 

[waross]

Also, what's an acceptable phase current imbalance?

An inexact rule of thumb;
Current imbalance equals the square of the voltage imbalance.
eg 2% Voltage imbalance = 4% current imbalance, 4% voltage imbalance = 16% current imbalance.
This rule of thumb is widely inaccurate but serves to illustrate that a small voltage imbalance may result in a large current imbalance.
A voltage imbalance will cause a circulating current between the grid and the back EMF. THis will be mostly limited by the impedance of the motor at slip frequency. that is, very close to locked rotor impedance.

What works the toaster test reveal? Just a higher resolution impedance test?

A shorted turn acts as an transformer with a shorted secondary or low voltage winding.
Consider the example of a winding with 100 turns and one turn shorted.
A DC resistance test will show a 1% difference compared to a healthy winding.
Use a ballast resistor to get the exciting current above the hysteresis level.
I would use an electric kettle or toaster to force about 10 Amps through the winding. (I once tested parallel cable connections with the electric kettle out of the plant lunch room.)
On a healthy winding, the current will be related to locked rotor current at the reduced voltage, ie: 10 V rather than 480 V.
On a six lead motor or a 9 lead motor, check the windings independently.
On a delta or wye connected winding, two phases will check bad. One phase will check clear.
Remember, the lower voltage may not cause the breakdown.
This test is good if it identifies a fault, but is inconclusive if it does not show a fault.
Unused leads should be insulated, not shorted.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[electricpete]

on a 500hp motor probably form wound, the likelihood that you have shorted turns which did not immediately progress to ground fault but instead somehow caused overload trip seems pretty darned small. i think you've gone in the wrong direction. you never answered the question how you determined motor #2 load to begin with.

condemning a motor based on relatively low current unbalance without measuring voltage balance is generally not advisable.
the expected current unbalance caused by voltage unbalance can be a heckuva lot higher than the voltage unbalance in a normal healthy motor
go to books.google.com and search for motor current unbalance and look for the results in "Energy-Efficient Electric Motors, Revised and Expanded" (excerpt below)

 

[eeprom]

The motor load was determined by flow and pressure. The voltage imbalance was 2 volts at 480V. The current imbalance percentage was the same loaded and unloaded. The voltage imbalance loaded was 2 volts at 480V. The motor was replaced because it was overloaded at far less than full load. It is still possible that there is more mechanical load than is determined by pressure and flow. The pump was decoupled and free turned by hand with no problem.

 

[eeprom]

ElectricPete,
Let me ask further...If I add the current vectors on the motor loaded, assuming the phase angles are 120 apart, the residual ground current was 60A. If that imbalance scales to starting current, the measured ground current would be 300A or more. There is a ground fault relay on the motor, with the setting at 200A. When I got there I was told that the motor tripped twice, and that it tripped immediately. The only other setting in the relay is the long time, which would have taken 5 seconds to trip at starting current.

Therefore, the breaker was tripped by a ground fault. Twice.

The motor was meggered, and passed easily. There is no short to ground. Then the motor ran for an hour without problems loaded, and then unloaded. Again, there is no short to ground. But yet it tripped twice on ground fault.

I cannot imagine what else it could be other than imbalance.

 

[waross]

Disable the ground fault trip during starting and see if it starts or trips.
Is there anywhere you can put a clamp meter to look for over 200 Amps to ground.
A window CT or Rogowski coil on the motor leads?

the likelihood that you have shorted turns which did not immediately progress to ground fault but instead somehow caused overload trip seems pretty darned small.

I mostly agree with Pete on this.
I have suggested an impedance check to rule out that small likelihood.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[eeprom]

The motor has already been replaced. If the new motor has the same issues with loading we'll know soon. If the new motor trips, then we have a relay problem, but given that this problem occurs on both pumps, I think the most likely problem is that the vfds haven been allowed to run too slowly for too long, causing the mirrors to overheat. This, I believe is breaking down the windings.

If we are losing turns due to insulation breakdown, we lose torque. That makes sense. The imbalance idea may prove to be wrong but it makes sense. We will know next week.

 

[electricpete]

I have traced this down to phase imbalance, I think. The loaded motor is drawing 634, 591, 587 A, and the voltages are within 2 volts.

ok i missed that part but I'm still skeptical.
How did you measure current and how accurate do you think those measurements are? were these read off a relay or a clampon? were they steady state or transient.
where did you measure the voltage at? if this was a transient did you measure it at the same time as the currents?

We will know next week.

ok, please let us know. sorry i didn't mean to be argumentative here. it was tough to follow the story.

 

[eeprom]

Electeicpete,
No worries. I very much appreciate the input. I care more about solving the problem than being right. In fact I would be more then happy to solve the problem and still be wrong.

The motor amps were measured at the motor with a clamp on ampeg. Reading rms. I've used it many times and I've never had issues. The voltages were also measured at the motor, with a fluke. I don't know how perfect they are, neither have been calibrated in a while. But I suspect they are close.

The measurements are steady state and I would be shocked if there was a dc component in them. I don't know though. I've never even considered that. I did zero the meters. That's just something more to consider in the future.

I will follow up next week when we get the motor report. Thank you again for your thoughts.

 

[3DDave]

As I suggested before, try to get oscilloscope traces to see what is getting to the motor/reacting from it.

I am also surprised there is no temperature sensor in the motor.

 

[Dopler]

Good day.

Have you tried running only the motor with load uncoupled?
To isolate any mechanical load and measure motor current.

 

[Marke]

When you say that the pump is unloaded, are you sure that mechanically the load reflected to the pump is actually reduced?
What type of pump is this and how are you controlling the load?
If it is a turbine pump, or a moving cavity pump, or some multi stage pumps, closing down the discharge side will actually increase the shaft load, not reduce it.
I have experienced many times, the differences between turbine pumps and centrifugal pumps, and all to often, I see people trying to throttle them and creating major problems!!

Mark Empson
Advanced Motor Control Ltd