SEL mirrored bits and relay failure

[pwrengrds]

I am involved in commissioning wind park in Calif.

They have two substations about 200 meters apart attached to a 5 mile line going to the 230kV utility grid connection. Sub 2 has been in operation for 2+ years, sub 2 is new. There are two line protection relays in sub 2, CT's in both substation are paralleled to the protection relays (SEL 311L and GE L90),and redundant SEL mirrored bit's (SEL 2506) between substation 2,3 to exchange line protection trips, breaker position, and breaker failure data.

They also use the mirrored bits to trip off substation 3 if the mirrored bits protection is lost (via loss of fiber or loss of DC to either set of the mirrored bits relay). So far substation 3 has tripped off 3 times with no indication of what the trip is in the substation, due to a missing packets going to the mirrored bits relay.

The SEL 2506 relay has DC supplied in sub 2, then the mirrored contact NC contact in sub 3 is wired to breaker trip coils of the main breaker in sub 3.

Where would I find a standard or paper on how relay failure should be dealt with. How should the relay failure be set up?

 

[dpc]

How do you know the cause is missing packets if there is no indication of cause? Have you examined the relay event data?

Based on your description, the cause seems more like comm problem than a relay failure.

You have to decide if you want a fail-safe system that trips on loss of communications or a system that does not trip if comms are lost. It can be done both ways.

There is some data stored in the SEL relays related to hardware faults that can be examined. You should contact SEL and discuss. I'd also suggest talking to one of their application engineers to discuss your particular installation. I think you will find them helpful. If it is an intermittent relay problem, you can send the relay back to SEL and they will test it at no charge. Even if they find no problem, they generally replace the motherboard before sending it back.

 

[slavag]

Sorry, Im not familar with SEL mirror bit systems, but have a simple Q. What is a time delay for communication fault?
Its a very not common solution, NC contact for trip, usually communication fault is used for block of protection ( line differntial for examole).

 

[pwrengrds]

I know at least one was due to that fault because we were at sub2 and were taking photos, one was near the fiber, and sub3 tripped off. The others I don't know, but that's the only one that does not have any type of indication.

So far I changed the SEL 2506 so it requires 2 packets in a row to trip, and I have been given permission to rewire it so the trip will go through two different SEL relays and have a 6 cycle time delay added.

I have suggested that the Engineer of record talk to SEL, I am the owners representative and question if that's the best way of doing it. I would like some typical examples of how this protection should be set up so I have some ideas of how it should be done.

The time delay for a communication fault was 1.5mSeconds, I have changed that to 3 mSeconds and if we run it through the SEL the time delay will be 6 cycles. The SEL 2506 relay alarms in 100mSeconds which is why I was given the 6 cycle time delay. I just don't think it's good enough yet.

 

[pwrengrds]

Yes these are communication faults, but they still trip off a 100MVA 230kV transformer and wind farm.

 

[cranky108]

Did you say you were taking pictures when one event happened? Were they flash pictures by any chance? Are you using ST connectors or the SEL less expencive connectors?

Why are you using the 2506, when most SEL relays will do mirrored bits internally?

As stated above, you can have securty or reliability, but not always both. you choose securty, and you can get some faulse trips.

 

[pwrengrds]

The mirrored bits is what I would have put in. I am not the designer, just the owners representative. I came in after the substation was energized, but before it was on line. The fiber uses ST connectors and are yellow colored fiber, SEL also advised that putting a dark covering would be better.

Yes, it was a flash photo, right next to the fiber. It has not tripped off since I changed the SEL 2506 setting to require the 2 packets to trip. It tripped 2-3 times before that. We will be putting in a 6 cycle delay (6 cycles) by running it through the SEL-351 relays, which is what should have been done in the first place.

I am concerned with long term fixes and was wondering if there was a standard way of doing this.

 

[cranky108]

One issue with mirrored bits is that there is a default setting so when there is a dropout of the communications, the default output will be pushed to the output contacts.
If you push a trip on loss of communications, you will get a trip for a loss communications (be careful what you ask for).

I have seen problems with sunlight and the other SEL connectors (That's why I asked).

It's better to set the delay to the maximum you can live with, for securty reasons.

 

[JG2828]

Cranky108,
You are referring to the V-Pin fiber connectors. I have also found those to be prone to loss of communication on flash photography.

 

[cranky108]

My experence was with sun light. Open the cabnet door and it would not work, close the cabnet door and it would. A simple tolet paper tube fixed the problem.

I've also seen simular problems with optical readers on EM meters.

 

[jghrist]

What line protection trips are being exchanged with mirrored bits? Direct communication between the line differential relays at each end would transmit and receive line current data used in the differential comparisons. With redundant line differential relays, and I assume, redundant communication channels, I fail to see the need to trip the line for momentary loss of one of the communications channels. This would seem to vastly reduce security for a neglible increase in reliability.

 

[zloy]

There is no relay failure, just an incorrect implementation of Mirrored Bits.

First, talking about SEL-2506. That is a simple device with no additional features like fallback values of MB, it merely follows data received (with an option of security counts).
If a data packet is lost it de-asserts ROK bit, that in turn de-enenergises output contacts. So the way it designed the whole scheme is supposed to trip the breaker on any packet loss - so it did on several occasions. The flash from your camera is also capable of momentary interruption of the data flow. Anyway, I would strongly recommend not using flash photo anywhere in industrial installations (unless you are 100% sure it’s safe) - some flame detectors are sensitive to it and there is a lot of embarrassment talking to a fire brigade.

If there is an intention to monitor the channel, you should use the alarm output - an alarm will be raised on remote / local unit failure or damage of FO cable. There is also a 100ms delay - that is more than adequate and I wouldn't try to use anything less than that anyway. If your requirements are stricter than that you need a second channel.

Also, if I got it right your substations are only 200 meters apart. When you mentioned that FO optic cables are yellow I would suspect they are single mode ones (and normally used for distances greater than 200 meters). I would check the nameplates of SEL units to see which FO cable type they made for and check all FO patch leads and FO cable all the way through to make sure there is no cable mismatch (single-mode and multimode). Usually SEL mirrored bits communication is pretty robust even if you do mix FO cables (happened on several occasions), but there will be occasional data packets loss with ROK bit dropping out.

We have quite a few MB schemes in operation and if they are properly commissioned they are trouble free.

 

[pwrengrds]

Bits that are exchanged are
Trip on line protection going to sub 3
Trip on Breaker failure going both ways
Trip on loss of DC or communication going to sub 3
Breaker position going to sub 2
Block close going to sub 3 (breaker failure 86 block)
The line protection relays are physically in sub2.

The alarm is monitored, but like you mentioned it's delayed for 100mSec and it trips in 1.5mSec (now modified to 3mSec), so we don't get indication which is why the utility is allowing the 6 cycle delay so we will have indication of what the trip was.

The utility is concerned that if they trip off the line sub 3 may not trip off unless they have this scheme. I have suggested that they modify it so that both sets of SEL-2506 relays have to loose communication for the 6 cycle delay before tripping instead of the current proposal of either SEL-2506 looses communication to cause a trip.