An interesting event 1

[davidbeach]

Saw something the other day that I'm interested if anybody else has seen anything like it.

The subtransmission system has two lines that run from station A to station B, by different routes. Both lines connect to the same bus at each end. One of the lines serves a tapped distribution station; said distribution station has a Dyn1 transformer to serve the distribution load. One of the distribution feeders leaves the substation, travels a ways and then meets up with the subtransmission circuit a ways away from the distribution station and then shares the same poles for 2/3 of a mile or so. At the end of that 2/3 of a mile there is an underground feed to a customer. The underground cable faulted, single line to ground. We know where and what since that had to be repaired.

What I found very interesting is that we saw zero-sequence current on the subtransmission system during the fault, none before, none after. So it appears that when a loop exists with a sufficiently low zero-sequence impedance it is possible for a distribution fault to mutually couple to the subtransmission and drive a circulating zero-sequence current around the loop. Without mutual coupling, the lowside 1LG fault would only produce positive- and negative-sequence current on the highside.

Unfortunately, I don't think the relay manufacturers considered this type of event when developing their directional elements. One end of the line was a strong source, the other a very weak source, so the bulk of the negative-sequence current was in the opposite direction of the zero-sequence current. That's not what the classical fault models would suggest. The weak end, that saw both negative- and zero-sequence currents in the same direction was able to make a correct directionality decision. The other end, though, seems to have figuratively thrown up its hands and made no decision at all; no directionality determination was made at this end.

Has anybody else seen anything like that? If so, what, if anything, have you done to increase security of the subtransmission protection against potential misoperations?

What is the largest impact anybody has seen where faults at one voltage level impact the protection of different voltages (particularly much higher voltages) in a common right-of-way?

 

[cranky108]

Sorry we don't have that problem. But I can offer that we use only negitive-sequence polorizing, because we don't trust zero-sequence enough.

 

[collies99]

Is it possible that the ground fault was large enough and was seen by the line protection as a large enough unbalance and was thus registered as zero sequence current. It would thus be interesting if the line protection elements or any for that matter responded for a downstream fault through the transformer.

 

[cranky108]

Is there any chance this might have been caused because of CT error?

 

[waross]

This may be a word problem.
I see the terms ground and neutral often used interchangeably when referring to higher voltage transformers.
I see zero sequence often used to describe ground faults on a wye system. However when the wye is fed from a delta, the zero sequence current to ground is reflected as a zero sequence phase to phase current. Thus a true zero sequence relay/algorithm will correctly report a zero sequence phase to phase current.
A relay or algorithm that depends on a measured or inferred current to ground to detect zero sequence currents may not report a phase to phase zero sequence current.

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

 

[davidbeach]

The residual currents at the two ends are of the same magnitude and 180 degrees apart when referenced to the a-phase voltage. With the delta high-side on the transformer at the distribution station there would be no zero-sequence (residual) current on the high-side. With enough examples of low-side faults as seen from the high-side, I know that the relays don't show residual current if there isn't any mutual coupling.

We also use only negative sequence polarizing for lines away from generation sources, but the directional algorithm is based on the negative- and zero-sequence currents having the same cause. In this case, the negative-sequence current is entirely due to the low-side fault while the zero-sequence current is entirely due to mutual coupling.

If the directional logic had made a determination at the end where it didn't, it would have gone with forward based on the negative-sequence current toward the distribution station rather then reverse as it should have been for the zero-sequence current into the terminal. Things would have gotten ugly if we had gotten a forward declaration at that end.

 

[davidbeach]

I suppose I should add that there is additional information I have access to but don't feel I can share at the moment. Two purposes to posting; 1) to see if anybody else has seen something similar (this is in category off things that can't happen that I have examples) and 2) to warn others.