Residual voltage and earth current protections coordination 2

[vuan]

In a 33 kV isolated industrial installation, there had been a trip from busbar residual voltage protection which opens the utility circuit breaker, leading to total blackout.

After busbar inspection nothing was found, so after reconnection from utility one feeder circuit breaker because of earth current. After inspection of the feeder 33 kV cable it was found that a cable joint failed making an earth short circuit.

After joint repair and further reconnection, everything come to normal.

The residual voltage is supposed to be feeder's earth current protection backup, with coordinated delays.

Question is why the feeder's earth current protection did not trip in the first time when residual voltage tripped.

My theory is that in an isolated system, an earth fault puts one phase to earth, and the current to earth may be below the earth fault protection setting. But in any case, if one phase goes to earth level, there will be residual voltage.

In those cases I have read that residual voltage protection has to be only alarm but not trip. So after the alarm it is needed to start looking for the fault, may be opening feeder by feeder until the alarm disappears, and then start normal earth fault investigation. The only risk is that transformers will be with 73% more voltage to earth, that may affect isolation, but if failure is investigated immediately, this is not a risk. May be we have faster ageing of isolation. Second risk is that a second earth fault will become a short circuit, but in this case overcurrent protections should clear it.

This is my theory, but of course suggestions are welcome.

 

[alehman]

Let me see if I understand what happened. You had a trip due to residual voltage first. You inspected and found no problem so you closed the breaker again. At that point it tripped due to earth overcurrent? Is this system is high-resistance grounded?

It sounds as if the relays are not properly coordinated (or not properly connected or not functioning).

Residual voltage could be used as an alarm only for the case of a single earth fault. If so, there must be another relay for tripping (typically over-current). But you say the residual voltage is backup to the earth overcurrent relay, which is the reverse situation. In that case it should be a trip function, not alarm.

 

[AMBMI]

Vuan I think your "theory" is correct. Probably the fault resistance was big enough to have an earth fault current smaller than the feeder earth protection threshold. Anyway in such case the transformer winding is not really experiencing the L-L voltage as L-G voltage. The rule should be: more ground current you have less zero sequence voltage and L-G over voltage on the healthy phases you have, less ground current you have more zero sequence voltage and L-G over voltage on the healthy phases you have.
Regarding which protection functions you could use I would prefer to have also a zero seq overcurrent element acting as backup protection on the bus incoming(s). At the this point the 59N could act as alarm for high resistive faults.
One question: are you using directional earth fault elements (67N)?

 

[vuan]

Thanks for your answers,

alehman,
First it open the utility CB, because of 59N. After operator's reconnection a different CB open. It was a feeder CB because 67N, who downstream the CB afterwards we found a dammaged cable joint. The MV system is neutral insulated, and only distribution transformers and reactive compensators are connected there. We think that the relays worked ok, because in the second time, the feeder's relay saw the earth fault.

AMBMI,
Yes, in that feeder we have directional 67N. So you mean to set in the incomings the 51N coordinated with feeders' 67N? Am I right?

 

[AMBMI]

Yes Vuan, you are right.
I was just wondering if you were using 67N or 51N in the outgoing feeder. When many outgoing feeders are present it's possible to trip only the faulted feeder using not directional elements but obviously the trip threshold cannot be too small.
I was wondering if it was your case.

 

[vuan]

AMBMI,
The engineering decided to use 67N for the outgoing feeders with longer cables, and 51N for those with shorter ones.
I was thinking if at least in the 67N we have to try with lower Io setting. Anyway the designers say that two phase short circuit after second earth fault is very risky, so they wanted to trip with 59N. Anyway I think it is an issue of probability. When this 59N alarm is on, maintenance crew has to act as fast as possible.

 

[burnt2x]

vuan,

What's your 59N threshold setting?
Compare 59N setting with 67N polarizing voltage setting.

 

[slavag]

Hi.
Vuan you are right.
Several points:
If you use 67N, use only it, isn't important longer or shorter cable.
51N possible use, but you need calculate capacitive current in many situation. 67N is great function and operated w/o any problem.
Put on the infeed back up 67N function with some time grading. And 59N used as back-up of back-up protection.
Don't forot installed in parallel with broken-delta VT connection damping resistor against ferroresonance.
Coordination between 51N and 67N: unpossible build.
between 67N and 59N, not sure.
Regards.
Slava.
Designer: save money and result: blackout. What is a differnce today: one cent.

 

[DiscoP]

I agree with slava. The cable length shouldn't determine whether directional protection is applied. This should depdend on the arrangement of network.

Assuming 2 incomers from the utility to the one main board, I would have expected 67N on the incomers and 51N on the outgoing feeders.

Is the 59N trip using an IDMT or Definite Time. Is the 67N and 51N IDMT.

 

[vuan]

burnt2x,
59N is 10% 0.5s alarm and 25% 0.8s trip incoming.
67N Uo = 10%.

Thanks everybody for valuable comments

 

[alehman]

I agree with Disco and slava. This design seems rather strange to me. Can you respond to Disco's question?

59N, 67N are % of what ?

Grading might be 59N alarm, the 67N or 51N outgoing trip, then 59N trip.

Do you have 51N on the incommer?

 

[vuan]

% of Uon, voltage of open delta measurement 100 V, also VT secondary are 100 V.
48N, 67N and 51N are all DT.
There are generator and utility incomers. Generator has 51N and utility has 67N but direction utility side looking for incomer transformer.

 

[burnt2x]

vuan,

59N 10% alarm in 500ms and to trip at 25% Uon in 800ms when compared to the 67N 10% Uon setting is good!

Recommend to verify Io setting and angle pickup. Also, it is best to grade your 67N tripping time to about 300ms from 59N trip (800ms)= say 0.5s or earlier.

One more Q: Does the 59N and 67N get inputs from the same broken-delta PT? I have had a problem similar with the afore-mentioned setup.

 

[vuan]

burnt2x,

67N is 100 ms, 8A. Angle is also ok because when we had the problem operator tried to reconnect and the cable joint developed a full short circuit to earth and 67N tripped ok.

Yes, both 59N and 67N are connected to same open delta for Uo.

 

[slavag]

OK Vuan.
Now I have additional Q: Why gnerator with 51N and incomer
with 67N in trafo direction?
W/O scheme is dificult say, but I not like this design.
Once we do similar application and used some other solution:
on incomers and generators ( was two and two with coupler) we puted 67N in two direction with time grading according to direction. Work fine, belive me, each fault I have call from customer.
Burnt2x, what do you meant angle pickup?
Regards.
Slava

 

[vuan]

slavag,
I did not made the engineering, and the protection report did not say nothing about this. I will check with designer and come back.

Regarding angle pickup, I think this is the setting of current direction in 67N relay module.

 

[slavag]

What type of relay, if you can say of course.
I used only one type of relays and it have basic angle -90deg and operated according to I0sinf sector.
we used logik I0sinf> I setting and Uo>Uobasic. Tested by live short circuit and we forgot about relay, setting, etc.
I know that isn't your design, I only would help to you with this issue.
Regards.
Slava

 

[vuan]

slavag,
Relay is vamp 255,

http://www.vamp.fi

, very nice and friendly for setting, programming, comms, disturbance record and protecting of course.

 

[slavag]

OK, we on the same side. Guys from Finland know 67N for ungrounded system as well, I think lot of Finland's distribution is ungrounded system.
good luck
slava

 

[Runsor]

Vuan, you wrote "The MV system is neutral insulated, and only distribution transformers and reactive compensators are connected there." Does this mean that your network is compensated i.e. it is grounded with a Petersen Coil?

If yes, do you have any resistor parallel with the Waldemar's coil?

If there is no parallel resistor to give some resistive 3Io, both the amplitude 3Io and angle of 3Io v. Uo depends very much of the degree of compensation and the distance to the earth fault.

Slavag mentioned sin(phi) characteristics, which is used with isolated networks for 67N.
For compensated network cos(phi) characteristics is used instead.

In ABB SPACOM 67N terminology cos(phi) means that the relay is sensitive to the resistive part of 3Io and sin(phi) means that the relay is sensitive to the reactive (capacitive) part of the measured 3Io. In both cases the broken delta voltage -Uo is used for polarization. Corresponding terms in VAMP 255 are Res and Cap.

You could ask vampsupport(at)vamp.fi to help.