Cap bank tripping

[stevenal]

The image below is from a single phase cap bank relay. A prior event is nearly identical. Only the first three strings are in use, so please ignore currents 4 through 9. Strings 1 and 3 are perfectly matched as they should be, but string 2 current went high causing the bank to trip. String 2 remains high after tripping. Looks like all DC. Capacitor units were tested after the first event, and looked okay. The posted event occurred five days after re-energization. I believe it is measurement error, but wonder what situation would cause this. Any ideas? Thanks.

 

[waross]

I'm confused. Are those voltage traces or current traces?
I don't understand the current persisting after the trip.
Is this three phases or three capacitor banks on one phase?

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

 

[stevenal]

Bill,
The graph on the bottom is bus voltage, all others are current. Ignore the middle two graphs.

The relay is monitoring three strings of current on the C phase of a three phase cap bank and C phase bus voltage to provide overcurrent and impedance tripping. Similar relays monitor the other two phases.

I also don't understand current persisting. Since the cans test okay, I suspect a measurement error. I'm not sure what CT problem would cause this, though. Either with the external CTs, or the CTs within the relay. I'm leaning toward a relay problem, though. What think?

Steve

 

[waross]

Hello. Am I correct that the graph is showing about 12 cycles and the center is zero?
Is it possible that one relay has a problem with an offset zero?
That is, a failing relay?
Is it feasible to switch the relay with one of the other relays and see if the problem follows the relay?


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

 

[stevenal]

Autoscaling shifted the range up some. I1 and I3 combine to make the purplish line which ends up at zero. I2 starts at over 500 A positive ending up at around 100 A negative, so the DC offset is not constant.

Yes, that would be feasible. Also thinking of shifting center string to one of the unused inputs.