Unit Differential Protection (87U)

[jliu1]

Our new design has a generator(296MVA), with GSU transformer (305MVA) and unit auxiliary transformer (24MVA) directly connected to the generator terminal. With the plant unit differential protection(87U), I understand that one set of CT should be at the HV side of the GSU and the second set should be at the neutral side of the Generator. Can anybody please tell me where the third set of CT should be placed on the Auxiliary transformer? HV side or LV side? Why? Thank you very much.

 

[electricpete]

I think you can take your pick. (But you'd better get someone else's advice).

If on the hi-side then you'll still have to provide redundant overlapping-zone protection for the aux transformer

If on the lo-side, then you'll have to reduce your sensitivity very slightly to account for aux transformer exciting current.

If you have a load-tap-changer on the aux transformer, then you'd probably want to keep your 87U ct's on the hi-side.

 

[peterb]

I would suggest that the CTs should be on the HV side of the UAT, to maximize sensitivity of the protection. You will probably also be providing separate UAT and generator differential protections, which will provide sensitive and selective protection - selectivity here refers to identifying fault location, as all these protections will trip the same breakers.

 

[jliu1]

Thanks very you two's quick response. I certainly prefer that the CTs are placed in the HV side.

But in ANSI C37.102-1995, Pages 29-30, it is specially recommended that this set of CT be located in the LV winding of the Aux transformer to avoid the possible CT saturation problem if the CTs are placed in the HV winding. Please provide your thought.

BTW, we do have the separate Auxiliary differential protection (87UAT).

 

[peterb]

Oh - did I forget to mention that the 87U should be a 3-winding type relay, with a restraint input for the UAT CT? Other than that, the HV CT would need to be specified to handle the full generator fault current (to ensure selective operation). Shouldn't be a problem with today's low-burden relays, but the calculation needs to be done.