Generator tripping on ground fault during load test

[knightsof3]

We have been testing a Cummins VTA-1710-G2 Diesel Generator 1800 RPM 480 Vac 671 kW with an Avtron K580 load bank that is hooked up to our three 480 Vac output cables and one cable to our facility ground. The generator output breaker keeps tripping when the load is around 350 kW. The test requires the load to be maintained at 560 kW for 30 minutes but it keeps tripping when the load bank switches higher than 300-350 kW. The generator output breaker is a TKM3F molded case breaker 1000 amps with a shunt trip coil wired to a Hi-Z model Z-2000 ground fault relay set at 200 amps at 6 cycles delay. Each time the breaker trips we get the ground fault light which indicates the ground fault relay actuated the shunt trip. The Hi-Z ground fault relay gets its signal from a Hi-Z model ZBT rectangular ground fault current sensor. This is a large rectangular laminated core that surrounds the three load output cables, which theoretically would sense an imbalance in the magnetic field if one of the cables had a path to ground, similar to a GFCI theory.

Heres what we know:

The resistors in the load bank and the load cables from the generator all meggered sat. No anomalies and all the resistances in the load bank were balanced 4.4 ohms for each bank. The generator tripped at various combinations of banks to attempt to get over 350 kW so the vendor said it wasn't one specific resistor bank that could be the problem child. Our megger confirmed this.

The generator neutral is ungrounded for reliability reasons since it is an emergency backup application. The generator meggered greater than 4200 megohms to ground. I'm not sure how there is a ground path for current with the generator neutral ungrounded and everything meggering sat as well.

Here's the strangest part I need help figuring out. The ZBT rectangular current sensor has the front side of the rectangle able to be removed to facilitate load wire installation. When the electricians removed the section of the current sensor to remove the normal load cables and install the vendor load bank cables they did not reinstall the front section of the rectangular sensor and this went unnoticed for a number of attempts at this test. I went out and noticed the front section missing from the CT and could not wrap my head around how this ground fault relay could be getting a signal from a CT with only 3 sides. I thought of an amp clamp that could not be closed around a conductor. Anyway we were grasping at straws and thought the detection circuit could have been acting erratically without a properly installed CT so we put the front piece on the CT and ran the test again and it tripped on ground fault again at the same exact point in the test.

1. How is an ungrounded generator getting a ground circuit and a 200 amp ground fault at that?
2. How does a ground fault relay circuit actuate consistently at the same point in the test with both an incorrectly and correctly installed current sensor?

Any input would be greatly appreciated.

 

[DaveE]

I definitely don't have an answer, but here are a couple of ideas:
1) Fortunately the problem sounds reproducible. So I would try to verify the ground fault current with another instrument, like maybe an oscilloscope and a current probe? This could help rule out faulty sensors. Also could provide a method for "tracking down" the source of the current.
2) I have often found with this kind of difficult troubleshooting problem that reexamining the basic assumptions that we are making can be helpful. Sometimes things we are sure are true actually aren't.

 

[knightsof3]

Where we would we put the amp clamp? On the ground cable from the load bank trailer to station ground?

 

[DaveE]

I would start by putting it on the wires that go through the GFCI sensor; to verify if there really is a ground fault, or if there is a problem with the sensor.

 

[berkeman]

Welcome to the PF.

Paging @anorlunda @jim hardy

 

[essenmein]

Probably highly not allowed, or safe, but could you disconnect the ground and just let the load float and see if the ground fault is still tripped?

If the ground fault looks for (I'm guessing) an imbalance in currents, could the load switching cause this if all contactors are not closing to the new resistance at the same time?

Could the current spike/noise from the load switch trip the breaker?

 

[anorlunda]

What are you able to measure for phase voltages and phase currents prior to the trip?

Edit: Also, what are the settings of the ground fault relay. When it supposed to trip.?

 

[knightsof3]

Probably highly not allowed, or safe, but could you disconnect the ground and just let the load float and see if the ground fault is still tripped?

If the ground fault looks for (I'm guessing) an imbalance in currents, could the load switching cause this if all contactors are not closing to the new resistance at the same time?

Could the current spike/noise from the load switch trip the breaker?

The vendor would not let us remove the loan bank trailer from ground for procedural and safety reasons and I don't blame him.
Its not really am imbalance the ct is sensing ao i phrases that wrong. In theory A and B could draw 5 amps and C draws 10 amps but if there's no ground current all the A B C phase vector sum still equals zero and the CT would sense nothing. The current out still equals the current in, inside the CT. With a ground fault there is current flow outside on the ground which is not inside the ct so you would not get get the return current. Thats a half ass explanation but I know that it won't work with a phase imbalance it has to be a ground fault. Also resistors don't have inrush current i don't know if that's what u meant by starting spikes but eithrt way the breaker wasnt tripping on magnetic instantaneous it was the ground relay and the generator would run all day at around 300/350 kw only when u went higher the ground relay tripped. U would think if a ground existed it would always exist not just when u switched to the next kW setting

 

[Averagesupernova]

Since the trip always occurs at the same load I would assume there truly is an imbalance in the load. A GFCI senses imbalance in the conductors by lumping together the current in each wire which of course involves the assumption that the current is going somewhere it should not be. Increasing the load simply increases the imbalance past the threshold. This is where I would start.

 

[knightsof3]

What are you able to measure for phase voltages and phase currents prior to the trip?

Edit: Also, what are the settings of the ground fault relay. When it supposed to trip.?

The voltage is steady at 480 Vac and I was told the current is at around 300 to 400 amps.
The output breaker trips at 1000 amps and the GF relay dial is set at 200 amps of ground fault current after a 6 cycle delay at 60 hertz. Theres no way its producing an actual 200 amp fault to ground it would need two grounds to produce that and the generator and output cables meggered 4200 megaohm to ground. The generator is ungrounded as I mentioned earlier so even with a true ground there would be no path for current to flow back to the generator unlike a traditional neutral bonded wye generator. While sacrificing personel safety for reliability this system would need a second fault introduced to get current flow thru ground for the current sensor to detect the phase vector sums of ABC not equalling zero. The very current sensor that wasnt even assembled correctly when it tripped the first handful of times LOL

 

[knightsof3]

Since the trip always occurs at the same load I would assume there truly is an imbalance in the load. A GFCI senses imbalance in the conductors by lumping together the current in each wire which of course involves the assumption that the current is going somewhere it should not be. Increasing the load simply increases the imbalance past the threshold. This is where I would start.

This current sensor system on three phases sees the sum of the current in vs current out. If there is no ground the vector sum is zero regardless of phase load imbalance. But keep in mind the CR was only installed correctly half the the time and result was always the same.

 

[DaveE]

Probably highly not allowed, or safe, but could you disconnect the ground

Nope. I wouldn't do this, especially if there is an indication of abnormal operation. This type of test that sacrifices safety is almost always avoidable. There is almost always a way to make the required measurements in a safe manner.

 

[DaveE]

One test I might try is to take all of the wires out of the GFCI sensor, but leave everything else unchanged. Leave the wires next to, but not through the sensor. Of course you want to close up the sensor magnetic path. Then see if the GFCI still trips.

 

[DaveE]

Also, I'm suspicious about the generator neutral connection. Of course, it should be ok as you described it, but I would double check your assumptions there. Is it really not connected to ground? If so, what happens if you do intentionally connect it to ground. Unbalanced phase currents could flow from the generator neutral through ground and trip the GFCI.
If you haven't already, measure the 6 phase-phase and phase-neutral voltages and the neutral to ground voltage when under a medium load. Asymmetry there may provide a clue.

 

[knightsof3]

One test I might try is to take all of the wires out of the GFCI sensor, but leave everything else unchanged. Leave the wires next to, but not through the sensor. Of course you want to close up the sensor magnetic path. Then see if the GFCI still trips.

The output of the 1000 amp breaker has 2) 500 mcm cables per phase going back to the vendors load bank trailer. These cables are not moving anywhere lol. The current sensor is built into place besdes the removable front section. This or determ the secondary CT leads from the the Hi-Z ground fault detection module.

 

[knightsof3]

Also, I'm suspicious about the generator neutral connection. Of course, it should be ok as you described it, but I would double check your assumptions there. Is it really not connected to ground? If so, what happens if you do intentionally connect it to ground. Unbalanced phase currents could flow from the generator neutral through ground and trip the GFCI.
If you haven't already, measure the 6 phase-phase and phase-neutral voltages and the neutral to ground voltage when under a medium load. Asymmetry there may provide a clue.

The generator neutral is bolted to an isolated piece of bus and has a reaistance if 4200 megaohms to ground. The load bank circuitry read 480 and the generator panel meter read 480 v but we did not take any readings with a volt meter on all the combinations u mentioned.

 

[knightsof3]

Does anyone know anything about capacitive coupling between an ungrounded generator and ground possibly causing nuisance trips on the ground fault relay?

 

[jim hardy]

If something expands with heat and shorts to ground there ought to be some smoke residue... I assume you've examined the generator .

i'd have a look at the CT's burden resistor first thing.
http://www.electricalpartmanuals.com/part_manuals/pdf/relaysAndMeters/HI-Z/Relays/925-5900.pdf

that manual doesn't have a schematic of the electronic part.

We once had a ground fault relay misoperate because of a defective capacitor.
It was a can type just like the motor run capacitor in your airconditioner.
One of its leadwires was fractured inside the case. When the capacitor got warm its metal case expanded and that wire ceased to make contact at the fracture rendering the capacitor open circuited..
A young relay engineer found it by warming the relay with a hair dryer. If i could remember his name i'd give credit.

Can you pass a wire through the CT, apply some current and measure voltage across your burden resistor ? I'd try ten amps because it's so readily available, but an AC "Buzz Box" welder could make your 200 amps.
Then check the calibration procedure for the relay and see if what you measure is about what should be expected ?
That'll tell you which direction to chase your trouble.
.

 

[Averagesupernova]

I agree with @jim hardy . Substitute a known controlled current through your current transformer. See what happens. Simplify simplify.

 

[knightsof3]

If something expands with heat and shorts to ground there ought to be some smoke residue... I assume you've examined the generator .

i'd have a look at the CT's burden resistor first thing.
http://www.electricalpartmanuals.com/part_manuals/pdf/relaysAndMeters/HI-Z/Relays/925-5900.pdf
View attachment 234040

that manual doesn't have a schematic of the electronic part.

We once had a ground fault relay misoperate because of a defective capacitor.
It was a can type just like the motor run capacitor in your airconditioner.
One of its leadwires was fractured inside the case. When the capacitor got warm its metal case expanded and that wire ceased to make contact at the fracture rendering the capacitor open circuited..
A young relay engineer found it by warming the relay with a hair dryer. If i could remember his name i'd give credit.

Can you pass a wire through the CT, apply some current and measure voltage across your burden resistor ? I'd try ten amps because it's so readily available, but an AC "Buzz Box" welder could make your 200 amps.
Then check the calibration procedure for the relay and see if what you measure is about what should be expected ?
That'll tell you which direction to chase your trouble.
.

Thats so funny you mention the burden resistor because I had notices a 200 ohm cement /ceramic looking resistor terminated on ine end if the CT secondary terminals but the other end of the resistor dangling in the air. I am not sure how it got that way or if that is the burden resistor because the ZBT vendor manual says the burden resistor is “permanently mounted” whatever that means. But i think we took a resistance at the 21 and 22 terminals at the gf relay the ct input terminals and get 200 something ohms if I remember correctly so idk what the deal is with that being needed or not. As far as injecting a know current thru the CT that's a great suggestion but we won't be able to try until more work can be rescheduled on this diesel when the vendor load bank returns.

 

[jim hardy]

Thats so funny you mention the burden resistor because I had notices a 200 ohm cement /ceramic looking resistor terminated on ine end if the CT secondary terminals but the other end of the resistor dangling in the air.

can you get a cellphone snapshot of that ?

It'd be worth a phonecall to the folks who built your relay to ask about it.
Ask for "their most gray haired engineer" and email him your snapshot.

One thing for sure
if you have a 200 ohm resistor and a transformer winding that are connected in parallel
and your ohm-meter read doesn't read less than 200 ohms
one or the other of them is open.

a dangling burden resistor would sure match your symptom.
I'd research that part of the relay so as to be armed with knowledge before the next round .

An old saying of mine
"In troubleshooting you got to always check the dumb stuff
when it doesn't pan out you simply don't admit you were foolish enough to look.
but when it does pan out everybody says "Wow ! How'd you ever think of that ?"

trace out the wires and see if you can lay your eyes on that burden resistor.

and inquire of that 'most gray haired engineer' what should be your ohmmeter reading where you mentioned.

good luck

old jim

 

[knightsof3]

Tan rectangle hanging off to left is 200 ohm resistor

I couldn't find that Hi-Z from linden NJ is still in business but here are some pics

 

[essenmein]

The vendor would not let us remove the loan bank trailer from ground for procedural and safety reasons and I don't blame him.
Its not really am imbalance the ct is sensing ao i phrases that wrong. In theory A and B could draw 5 amps and C draws 10 amps but if there's no ground current all the A B C phase vector sum still equals zero and the CT would sense nothing. The current out still equals the current in, inside the CT. With a ground fault there is current flow outside on the ground which is not inside the ct so you would not get get the return current. Thats a half ass explanation but I know that it won't work with a phase imbalance it has to be a ground fault. Also resistors don't have inrush current i don't know if that's what u meant by starting spikes but eithrt way the breaker wasnt tripping on magnetic instantaneous it was the ground relay and the generator would run all day at around 300/350 kw only when u went higher the ground relay tripped. U would think if a ground existed it would always exist not just when u switched to the next kW setting

Re resistors switching, not talking about inrush just the noise from a contactor closing, there's always some EM from an ark or bounce.

Seeing the picture of the current transformer around all 3 wires, you're right, a phase imbalance as long as all the current is going between the phases and not somewhere else, would (should?) see zero sum current.

But unless I'm missing something from the picture the load resistor for the CT is not connected (one end hanging in the air), that could cause current dependant nuisance tripping.

 

[dlgoff]

But unless I'm missing something from the picture the load resistor for the CT is not connected (one end hanging in the air), that could cause current dependant nuisance tripping.

Yep. Here's a picture from the HI-Z sensor manual. Note the red underlined text.

 

[Averagesupernova]

I have to admit that when I first read through this thread I missed the fact that the GFCI wasn't supposed to trip until many amperes (3 digits). I expected to find the cause being much more subtle. Now finding out that the current transformer is likely driving an open circuit, the cause probably is very subtle like leakage capacitance.
-
I have to wonder what kind of sparks would have flown if there would have truly been many amperes leaking off to where current should not be going. An open secondary on a current transformer is a bad thing.

 

[jim hardy]

I have to wonder what kind of sparks would have flown if there would have truly been many amperes leaking off to where current should not be going. An open secondary on a current transformer is a bad thing.

yep, agreed.

Now if there's no fault and no phase imbalance there should be no mmf(current) to magnetize the core. At least at 60 hz.

It is well known though that 3rd harmonic of all 3 phases is in phase and returns through the neutral.

I'm wondering if, with an open CT secondary, there's enough 3rd harmonic current circulating around the distributed capacitances to make a small amount of 180 hz flux in the core.

That your trip is repeatable suggests it's not a random transient .

Part of your preparation for next round might be provision to measure amplitude and frequency of whatever is at CT secondary. An oscilloscope would be best.
As your generator load increases so will its third harmonic content.You said you've ungrounded the generator neutral.
I would suggest changing from Ungrounded to High Impedance Grounded . That gives you the same ability to operate in the presence of a phase to ground fault but with additional protection against accidental ferroresonance which can pierce insulation.
Typically one grounds through a resistor with resistance about equal to Xc of the system's distributed capacitance. That assures low Q should something commence to resonate. A few hundred ohms at appropriate wattage would be a blind guess .
A 480 volt incandescent lamp would announce a ground fault.

Anyhow thanks for the pictures. Perhaps whoever assembled your generator package has a grey haired engineer ?

That dangling resistor looks suspicious to me. I'd answer that question before next test run.

old jim

 

[Averagesupernova]

That your trip is repeatable suggests it's not a random transient .

My exact thoughts from the beginning.

 

[jim hardy]

My exact thoughts from the beginning.

Thanks -

I went out and noticed the front section missing from the CT and could not wrap my head around how this ground fault relay could be getting a signal from a CT with only 3 sides. I thought of an amp clamp that could not be closed around a conductor. Anyway we were grasping at straws and thought the detection circuit could have been acting erratically without a properly installed CT so we put the front piece on the CT and ran the test again and it tripped on ground fault again at the same exact point in the test.

1. How is an ungrounded generator getting a ground circuit and a 200 amp ground fault at that?
2. How does a ground fault relay circuit actuate consistently at the same point in the test with both an incorrectly and correctly installed current sensor?

Question 2 is a stumper. A CT should be almost indifferent to current with that segment missing. It takes a lot of amps to push flux through that air gap.

 

[jim hardy]

with the front of the CT out it resembles an old fashioned horseshoe magnet

and i can imagine it producing some voltage into an open circuit
especially if one of the phase wires flopped outside the end of the horseshoe.

But it is quite a coincidence that it tripped at same current when put back together.

These mysteries usually become apparent after one roots out the underlying cause.

And nobody said there's only one thing wrong.

Sanity check -
that removable front bar of the CT is made of steel, isn't it ? Laminated like a transformer core ?
I'll be chagrined if we're applying CT troubleshooting technique to some sort of Rogowski current sensor...old jim

 

[jim hardy]

...tan rectangle hanging off to left is 200 ohm resistor

it sure looks like something that could've got knocked loose while manhandling those cables...
I am thinking it's likely either the burden resistor or a resistor they put there to protect the CT from somebody lifting a wire down at terminal 21 or 22 of the relay.
In either of those cases it ought to be connected.How big are those wires ? Do i see 350kcm ? Are there two per phase or just one ? If two, do all six go through the CT ?

grabbing at straws now
i don't know what is the purpose of this ground jumper
but some draftsman thought it important enough to put on the drawing. Good question for that 'gray haired engineer' ?

these phone numbers were in an installation manual online.
http://www.electricalpartmanuals.com/part_manuals/pdf/relaysAndMeters/HI-Z/Relays/925-5900.pdf

Worth a try ?

good luck

old jim

 

[knightsof3]

Those phone numbers arent to that company anymore and I could not find who took over this product online but good thought i might be abke to pursue this further. Also 350 mcm or kcmil sound right I originally thought 500 off the top of my head but that puc says 350 amd there's two per phase. We cannot change the design specs unfortunately and the gen neutral is ungrounded for its application so it must stay that way. I do find this style of ground fault current sensor ridiculous for this application and it will never even work as inteneded on a true ground fault because there will be no fault current returnining to the neutral. Even more sily is that if there was a second ground that caused a short circuit between phases it would trip the 1000 amp magnetic breaker but the ground fault current sensor would in theory not detect a “ground” because all the supply current was returned from the phase wires to the generator stator and it would not detect leakage back to the neutral

 

[knightsof3]

it sure looks like something that could've got knocked loose while manhandling those cables...
I am thinking it's likely either the burden resistor or a resistor they put there to protect the CT from somebody lifting a wire down at terminal 21 or 22 of the relay.
In either of those cases it ought to be connected.How big are those wires ? Do i see 350kcm ? Are there two per phase or just one ? If two, do all six go through the CT ?
View attachment 234095

grabbing at straws now
i don't know what is the purpose of this ground jumper
but some draftsman thought it important enough to put on the drawing. Good question for that 'gray haired engineer' ?
View attachment 234096these phone numbers were in an installation manual online.
http://www.electricalpartmanuals.com/part_manuals/pdf/relaysAndMeters/HI-Z/Relays/925-5900.pdf

Worth a try ?
View attachment 234094

good luck

old jim

Also good observation on the bond to ground on the secondary CT wiring. I was not able to verify if that existed in the field but next opportunity I will have them look for that

 

[Windadct]

I have tried to read each post - but do not see if the Burden Resistor has been re-connected. Pretty sure the Relay is V sensing and the burden resistor is NECESSARY -and IMO the one shown is not mechanically sufficient for this critical item- note the Mech design of the one in Don's Post 24). With the burden resistor disconnected, the CT output is open circuit and sending HV spikes to the relay - the relay sense circuit is (likely) quite crude.

Also - large window CTs are not perfect, the location of the cables will cause an apparent imbalance, not enough to normally look like 200A worth!~ but with the burden resistor missing...

Also - as a test, use a clamp on CT and measure the current in the ground/ neutral connection ( this can be done without disconnecting it) - not 200A, I am quite sure.

Another test - with a current source, like a MCB test set, you can wrap a cable around 10 x and pump 20 A through it to get to 200 A... and do this test with and without the resistor connected. I'll bet it trips at about 20-30A effective ( 2-3 A x 10 turns) without the resistor and --- 200A with the resistor.

 

[knightsof3]

I zoomed in on the vendors manual image and the resistor i saw hanging is definitely the burden resistor designed for the sensor. 200 ohm 10 watts it looks like it says just like I saw. Unfortunately i won't be able to have the resistor hooked back up and run the test again until the load bank trailer is rescheduled but this is a huge clue.

 

[jim hardy]

I zoomed in on the vendors manual image and the resistor i saw hanging is definitely the burden resistor designed for the sensor

well i think you're almost there...

with that resistor missing it'll be spectacularly oversensitive.

If you can hang a simple battery handheld 'scope across it it'd sure be interesting to see the waveform . 60 hz, 180 hz, or a mix ?

old jim