Externally summed CTs for bus protection 3

[Mbrooke]

Starting this not to steer off another thread, but I inquire 2 questions:

1. At what point or fault levels do engineers typically move away from externally summed CTs for (over-current) bus differential protection? C800 class CTs of concern here FWIW.

2. Can externally summed bus protection CTs be used in networked circuits?

 

[DTR2011]

Our local DisCo had adapted the fast bus trip scheme for new installations. Previously a High Z (GE PVD) scheme was used. SEL has a nice explanation of the scheme in the 251 Manual, along with a scheme that allows the bus relay to handle a failed feeder relay, or a feeder relay OOS for other reasons. There is a bit of wiring involved, so this is on new relay panels / switchgear.

I can confirm from commissioning these schemes, the 3 cycle time (between relays) is easily achieved.

 

[davidbeach]

After my adventures of the past few days I am so terribly glad that our dabbling in the 251/251C fast bus scheme was a brief episode and that we've long since moved into a more enlightened era of dual feeder relays, bus protection, breaker failure logic, and a maintenance mode switch on the transformer protection. It's all been SEL from then till now, but that fast bus trip era was truly a misguided, penny wise pound foolish, attempt at getting more that what you were willing to pay for. SEL had this great idea how you could get more out of fewer relays; they sold fewer relays and we got worse results over the long run. There's no free lunch.

That scheme has some obvious advantages and it was well sold; but that was 20 years ago. There are some very subtle ways for it to fail; at the moment I don't feel that sharing would be appropriate but it seems to have built in several traps. Don't (just) look at first cost; look at the costs (both tangible and non-tangible) of an extended outage simply because some corner case behaves radically differently from what everybody is more familiar with.

 

[Mbrooke]

You need to look at the burden and the system X/R ratio as well as the fault current to determine saturation.
Vs = (1+X/R)·If·Zb
I would recommend low impedance with restraint or high impedance. If you are concerned about cost, consider a fast-bus trip scheme instead of bus differential. If you have microprocessor relays on the main, bus tie, and feeders, you might achieve 3 cycle operation without any additional relays or CTs.

I know long runs in the CT circuit (higher ohms) decrease saturation, but, how does low burden play into this?

After my adventures of the past few days I am so terribly glad that our dabbling in the 251/251C fast bus scheme was a brief episode and that we've long since moved into a more enlightened era of dual feeder relays, bus protection, breaker failure logic, and a maintenance mode switch on the transformer protection. It's all been SEL from then till now, but that fast bus trip era was truly a misguided, penny wise pound foolish, attempt at getting more that what you were willing to pay for. SEL had this great idea how you could get more out of fewer relays; they sold fewer relays and we got worse results over the long run. There's no free lunch.

That scheme has some obvious advantages and it was well sold; but that was 20 years ago. There are some very subtle ways for it to fail; at the moment I don't feel that sharing would be appropriate but it seems to have built in several traps. Don't (just) look at first cost; look at the costs (both tangible and non-tangible) of an extended outage simply because some corner case behaves radically differently from what everybody is more familiar with.

Thanks, a wealth of wisdom as pure usual Considering your experience, practicality and advanced knowledge in the field of protective relaying I am going to take your word on fast-bus. Having decentralized bus bar protection for 12kv and above has never sat right with me, I actually prefer having more relays.

If you feel comfortable saying, what do you use for bus protection at the 12-34.5kv levels?

 

[davidbeach]

At that level we're using the 587Z plus optical arc flash detection plus a maintenance mode switch on the transformer protection. We go that route in part because some switchgear would require three 487Bs, one per phase, to cover the 8 or 9 breakers and there's no place to mount all of those. Anything transmission gets redundant 487Bs.

 

[Mbrooke]

Sounds like a well thought out practice. On the 487Bs, you've never had any inadvertent operation from saturated CTs, correct?

 

[cranky108]

David, nice comments.

The problem with that scheme with the 251 is that it keeps cropping up as a way to save money.

What I don't find is papers on how to best install the fibers for arc-flash protection.

 

[davidbeach]

I can't recall any misoperations of a 487B, neither over tripping nor under tripping, on our system

 

[stevenal]

I know long runs in the CT circuit (higher ohms) decrease saturation

Make that increase.

 

[Mbrooke]

Ok, guess I need a course on factors increasing CT saturation. Any good links or papers on the subject?

 

[davidbeach]

Stan Zochall's CT book available from SEL would be a good start.

 

[Mbrooke]

Do you have a link by chance? I did find these two PDFs, they look good:

https://cdn.selinc.com/assets/Liter...mize_JR-SZ_20110627_Web.pdf?v=20150812-085652

https://cdn.selinc.com/assets/Liter...ers/TP6038_19920817_Web.pdf?v=20150812-085648

 

[Mbrooke]

Also anyone know how to find the line angle on a distribution bus?

 

[davidbeach]

Third book on

https://selinc.com/publications/bookstore/

Appears to also be available through Amazon.

Why would you need a "line angle" for a bus? I imagine that you could calculate the impedance as though the bus were a line and use that. Pretty sure you'll get something over 80 degrees.

 

[Mbrooke]

The first paper makes not of line angles and CT saturation on page 3, which threw me off.

 

[Mbrooke]

And oh, thank you for the link. Much appreciated

 

[stevenal]

I also suggest IEEE C37.110.

 

[jghrist]

You can calculate X/R from the line angle. You need the fault current in complex form to get X/R or the line angle. Line angle = arctan (reactive part of fault current / real part of fault current). X/R = tan ( line angle). For a distribution bus on the load side of a power transformer, the X/R ratio is high because most of the fault impedance is the transformer which has a high X/R ratio.

 

[Mbrooke]

I take it 3 paralleled 30/40/50 MVA 115-34.5kv transformers will produce a very, very high line angle?

 

[davidbeach]

Yes. Hopefully you should be able to get the fault X/R from your system model by applying a 3LG fault on the bus.

 

[Mbrooke]

I have the transformer R and X for each unit, plus the source 115kv impedance. Will this be enough?