Generator Ground Diff Trip

[jghrist]

One of our municipal utility clients had a ground differential trip on their peak shaving generating plant recently. The generator is diesen engine driven, 1825 kW, paralleled with two other identical generators and with the utility system. The trip occurred 25 minutes into the run and was restarted successfully. The generators were operating in Peak Shave/Base Load mode and were set to generate a constant 1825 kW.

Generators are 4.16 kV with 400A grounding resistors. Three generators are paralleled and connected to the utility 12.5 kV substation with a 5000/6250 kVA delta-grd Y transformer, delta on the generator side.

I believe that the ground differential is set too sensitively and that the residual current from the phase CTs has enough noise to trip the relay. The relay is a Beckwith M-3240. Phase CTs are 400:5. Neutral CT is 50:5. Pickup is 0.2 amps (minimum setting) with a time delay of 2 cycles. The CT ratio correction factor is 7.99 (maximum setting).

Oscillography of the event is on the attached Word document.

What really has me wondering is the current dip and oscillations starting at cycle 34. Current was about 270A (1960 kVA) before the dip and settled to 150A (1090 kVA)after the dip, before tripping. There was no dip or increase in voltage. Note that there seems to be a frequency deviation also. Phase C peaked at 32.2 cycles and at 33.2 cycles. After the current dip, Phase C peaked at 34.7 cycles and 35.7 cycles.

As far as we can tell from records, there was no trip of any distribution feeder at the time, and the substation is connected to a 100 kV transmission system that would absorb any excess generation.

Any ideas?

 

[slavag]

Hi Jghrist.
It seems as very low setting.
Very starnge dip at 35 cycles.
May be was some problem with AVR or prime mover?
Strange aslo some difference between Ia (red) and others phases.
But is not seems critical. ( before dip, was same with phase Ib (blue)).
Are you have recorder of voltage anf freq. in this time?
Regards.
Slava

 

[jghrist]

Voltage showed no deviation or distortion. Rock solid at 4.2 kV on all phases until the unit tripped tripped.

 

[slavag]

Next strange thing.
0.2A is 16A of primary ground differential current.
According to graf, it was about 7A.
Not so clear for me.
Normal noise according to graf about 7A.
Next : two boarding setting, not so good.

 

[slavag]

Sorry, I meant, 0.2A is min settinf and 7.99 correction factor is max setting.

 

[jghrist]

slavag,
The phase CT secondary current residual is multiplied by a CT ratio correction factor of 8 (actually 7.99 max setting) before it is compared to the neutral CT secondary current.

I don't know what you mean by "two boarding setting".

No frequency records.

I think your idea about a prime mover problem may be the reason for the dip. Note that the current phase shifts between cycle 32 and cycle 36. The current shifts approximately 150°. Basically a current reversal. The voltage doesn't shift. It looks like the engine fuel was cut off and the generator started motoring. The 150A current after the dip was enough to turn the engine. If the ground differential hadn't tripped, the reverse power element would have tripped the breaker.

The engine-generator has run without problem since this incident.

 

[ScottyUK]

The 'instantaneous' frequency during that event dropped to less than half nominal - I wonder if you are seeing some CT saturation effect causing the imbalance?

The system in your area appears to have shifted just over half a cycle in phase. Must have been a big disturbance system to cause that - I haven't seen one as severe as that before. How remote is the site - on the end of a really long transmission line?


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[slavag]

Jghrist,
" two boarding setting"="my bad English". Cancelled it.
As, I understand, you have also other format of recorder and evaluted it. Agree with you about 150 shift of current.
It seems as reverse power.
I don't know this relay, are this relay include also event list or fault report?

In any case, prime mover fault may be reason for the dip, but isn't reason of ground fault trip.
Regards.
Slava

 

[slavag]

Jghrist.
I check again ( of course only visualy) graf of neutral/residiual currents.
It's not seems good. Current defently was lower than pick-up of relay.
I don't know, may be logic of this protection include not only diffrential element, also some directional logic element or operated according to peak-to-peak current.

May be possible check one of genset with real reverse power?
or are tested in commissioning time?
Regards.
Slava

 

[davidbeach]

I believe that the generator began to motor with the change from generation to motoring happening around cycle 34. Interesting that we have currents from cycles 32-36 but the most interesting time for a ground differential was from cycles 82-84 or there 'bouts. When did the trip actually happen, cycle 112 when the residual currents cut off, or about 83 when the phase currents cut off? (Guess which one I was looking at when I started typing that.) If the breaker did open about cycle 83, where does the residual current come from? Do you have COMTRADE files you could post? Has Beckwith looked at this? What about phase angle between the residual and neutral currents? I am assuming that the 0.2A minimum setting is secondary amps. So many questions...

 

[ScottyUK]

Hi David,

Motoring sounds like an excellent explanation. I think I have become too focused on the last graph and made the assumption that the current recovered to roughly the same level, as it appears to, on that graph but plainly it does not on the first one. My only excuse is it was late!

Approximately how much power does a motoring diesel engine consume? Obviously a lot higher than a turbine, which is pretty much the only thing we ever see motoring at our site.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[jghrist]

slavag,

Relay is Beckwith M-3420 (not M-3240 as in OP).

Phase CT secondary residual (3I0) was less than 0.2A, but the operating quantity is |(RC·3I0-IN)| where RC is a CT ratio correction and is set at 7.99.

There is a directional element based on the product (-3I0·IN2·cos[&phi;]) but the directional element is inoperative if I0 is less than 0.2A. I'm going to have to check with Beckwith on this because I0 is < 0.2A and the operating quantity is > 0.2A for the entire record, so why didn't the relay trip right away? Maybe the directional element is only inoperative if RC·I0 < 0.2A.

David,

Comtrade file is attached. Trip asserted at sample 1369 (cycle 85.56). Looks like breaker tripped during cycle 83. There is a Woodward EGCP-2 generator control relay which also has a reverse power element. I'm having the client check the event log to see if this operated. I suspect that it did and tripped the breaker.

I'm going to check the phase angles of the residual and neutral currents. The residual current is mostly noise, so I don't know if the phase angle will have any meaning. The neutral current looks like it is all third harmonic, so the phase angle of the fundamental will probably also be useless.

I think that the spike in neutral current came from one phase of the breaker interrupting before the others. This happened early in cycle 83. Phse B interrupted when it reached current zero earlier than the other phases.

In the second sample in cycle 83, the residual current is +0.25A secondary and the neutral current is +1.36A. Operating quantity is (7.99·0.25 + 1.36) = 3.36. It appears that the direction of the residual and neutral currents are opposite which would indicate that the neutral CT has the wrong polarity, but the residual current is practically all noise. The noise in the residual current in general is as high as the residual current at the time of the neutral current spike, so I can't read much into it.

Scotty,

It looks like a motoring diesel engine requires 150A or about 900 kW. Seems like a lot to me considering that it starts with a battery.

 

[jghrist]

I notice that only the COMTRADE configuration file got attached. Here is the data file.

 

[slavag]

Jghrist, thanks for COMTRADE.
It's defently reverse power, you are right.
I'll download manual of relay, something not clear me with
RC*3I0-IN.

What is a time delay of reverse power?

Next point, possible trip from logic of Woodward in case of
some mechnical problem, not from reverse power.
I see it not one time.

Next point, for my pinion it's not reason: one phase open before others for the peak in neutral current.
It's must be zero-seq components and not unbalance.
Regards.
Slava

 

[jghrist]

Slavag,

I'm trying to get the reverse-power settings, both for the Woodward EGCP-2 and the M-3420.

There is relatively little zero-sequence current even during the breaker trip. The problem with comparing the residual and the neutral is that there is so much noise in the residual current. There are several samples before the trip where the residual current exceeds 20A primary.

 

[davidbeach]

I don't have time to look at any of it now, but if the relay issued the ground-diff trip AFTER the breaker was already opening then all bets are off. One pole open can foul up many protective functions that aren't expecting an open pole.

 

[slavag]

I have some problem with download manual.
But it's not imoprtant.
Jghrist, please ask to mnf. about mode of this function.
I know only one:
1. Calculation must be base on the fundumental zero-seq. components of residiual and neutral current.
Id=(3I0-IN) ( on this moment w/o RC).
2. second operation element must be directional
Id*cosF , in case of more than 90 deg fault in zone, less from 90deg fault out of zone.
3. From my expirience CT 50/5A is not good choice for the
sensetive eart fault protection ( or differ earth fault protection). I usually recommend to my customers use 100/1A
in this case may be 200/5A (also not so good, but better from 50/5A).
Regards.
Slava

 

[slavag]

O.K.
I downloaded this manual.
Jghrist, first thinking was right thinking.
Manual: For higer values of RC ( your customer have 8, is 400/50)noise may create substantial diff current
making higer pick-up setting desirable.

Next point, as you wrote, if 3I0<0.2A ( less approx)
directional element are disable.

According to graf 3I0 was less 0.2A, in this case
system used RC3I0-IN. Next Q is to know-how of mnf.
1. Are algorithm calculte according to fundamental companent?
2. How much cycles for decide: disable or enable directional elements, etc.
Conclusion.
Low pick-up setting, high RC setting--- unwanted trip.
For my pinion need change both of parameters.
1. Change CT from 50/5 to 200/5
2. Change setting to something like 0.5-1A.
Regards.
Slava

 

[jghrist]

Beckwith sent me a paper describing the 87GD algorithm. The operating quantity is based on a 16 sample average; consecutive samples, not a sliding window. The paper says the diffential quantity is (3I0 - IN). I have made a request for clarification. 16 sample averages of (RC·3I0 - IN) are > 0.2A for the entire record. 16 sample averages of (3I0 - IN) are > 0.2A only for the 16 samples where the breaker is opening.

Because of the sampling method, the time delay accuracy is -2 cycles to +1 cycle instead of the ±1 cycle in the manual. With our setting of a 2 cycle time delay, there may not be any actual delay.

I'm thinking we need to increase the pickup and the time delay, but how much depends on answers I get from Beckwith.
The EGCP-2 alarm log was reset when the generator was restarted, so we lost that data. The operator says that the only alarm was a "Remote Fault" indicating that something other than the EGCP-2 tripped the breaker. The M-3420 oscillography indicates that something other than the M-3420 tripped the breaker. The only trip signals to the breaker are from the EGCP-2 and the M-3420. I think this may remain a mystery.

 

[slavag]

Heh, "dead case".
It remaind me, one start of 36MW DG, after protection retrofit. After 48hours of continiusly work we give O.K. to customer, operator started this DG,synchroniezed it to network for on-load test and after several min, "something" send trip to CB. NO events in relays, NO events in Woodward, NO LOR operated. Of course customer saied me: It's your new relays. Several additional hours we tried understand what happend (with SER, TER, PI,event list, w/o any result), started DG again and two years - NO problem ( protection problem).
"Dead case", "something send trip".
Regards.
Slava.
Jghrist,please send us results, it's very intresting case.