Open phase detection on distribution

[HamburgerHelper]

Our system has an issue with detecting an open phase or an open phase with a jumper that swings down into the phase below it in our distribution system. The arrangement is a step down delta-wye grounded transformer in the substation that feeds a bunch of wye-ground-wye-ground distribution transformers in the field. There is a programmable relay in the substation, the distribution transformers are fused, and there is a mix mash of automated and comm. controlled reclosers in the field.


When I first was told this problem, my first response was to use negative sequence or a I2/I1 ratio to detect an open phase. The problem I am seeing with this is that if your jumper fails near the end of the feeder, you will have a hard time detecting an open phase with I2 or an I2/I1 ratio without making them overly sensitive. All the other transformers further up the feeder drawing I1 will make it difficult to see the open phase from the substation. Someone brought up to have the relay alarm on large step changes of negative sequence current. I don't know how well that works. Nor do I like the idea of having to investigate each arrangement to determine what a "large" step change is.

How do you guys go about tackling this problem?

 

[davidbeach]

A voltage sensor at the end of the line and a communication path back to the substation may be a much better solution than trying to detect the issue at the substation. The whole point of using fuses on taps is to keep from needing to trip (or at least not lock out) at the substation. Try to detect a step change in I2/I1 at the substation and you may just find yourself tripping for blown tap fuses.

 

[cranky108]

I suggested to a coworker, not to long ago, that we can use spare PLC (power Line Carrier) equipment to do this same thing. Since the line we are concerned with is a single-phase on a delta system, the transmitter and reciever can be at the same location, with a capacitor at the far end.

In our case the line has only a handful of transformers along it's length, and very little load.

In your case, a transmitter at the far end, and one, two or three recievers at the substation.

May not be your best option, but it is an option.

 

[Mbrooke]

Recloser loop schemes automatically detect this. Any implemented or going to be?

 

[HamburgerHelper]

If I am understanding your loop scheme, you are talking about rolling load. In that case, there is a relay in the field that could do some of this but in a lot of places there is not.

 

[Mbrooke]

Yes, or automatic restoration of load for feeder outages. In those cases most schemes have 3 phase voltage sensing which will detect an open phase.

 

[bacon4life]

Is the issue detecting open phases? Or detecting phase to phase faults? Or detecting circuit to circuit faults on the same phase?

In the past, the plain old telephone system was the way to detect open circuits. In the future smart meters will fix it. In the present, we have not been able to widely apply negative sequence relaying on our distribution feeders due to a few factors:
-It is tough to keep feeders tightly balanced.
-Single phase switching between feeders or during cold load pickup appears as high negative sequence currents.
-Line to ground faults beyond delta-wye transformers look like negative sequence current.

 

[HamburgerHelper]

The issue is with the tanks overheating on zero sequence flux passing through the tanks. Would I be wrong in thinking that if an impedance was used to choke the ground source at the substation, not a lot but some, the amount of ground current flowing during a jumper failing and falling into another phase, which causes the transformer's tanks to overheat, would decrease significantly.

The sequence network for an open phase is all three networks in parallel. I don't know what the network is for a jumper falling into another phase but I suspect it is similar. SLG would change some with adding an impedance to the ground. If the networks are in parallel for a jumper falling into a phase just like with the open phase network, the amount of zero sequence current flowing during this condition would be dependent on how Z0 compares to Z1 and Z2.

Where can I find the network for this condition? The Clarke transformation I thought could do this double contingency but I can't find the paper.