Relay Tolerance and Setting Errors

[Modula2]

In Ieee 242-2001, "Recommended practice for protection and Coordination of Industrial and Commercial Power Systems", Table 15-2, a value of 0.12 sec. is given for relay tolerance and setting errors, static relays, coordination time interval, field calibrated relays.

I'm wondering what the true basis of that number is and what it includes. The relay I'm looking at has a timer accuracy of +/-0.25 cycles for Inst/DT elements, which is not even close to 0.12 Sec.

Does someone have some helpful pointers? I asked the oracle Google, but not much out there. Thanks.

 

[davidbeach]

Breaker time, CT errors, safety margin. Say your "faster" device has a CT error that produces less current at the relay than the CT ratio would suggest and the the "slower" device has a CT error that produces more current at the relay than the ratio would suggest; the faster relay will take longer to trip and the slower relay will trip is less time than suggested by the curves. Is 0.12s still long enough? Don't count on it.

 

[dpc]

With instantaneous elements, CT error can be more of a factor due to CT saturation, since the settings tend to be high. You also need to read the fine print in the relay specs regarding minimum response time. Digital relays can actually be somewhat slower than electromechanical instantaneous elements. The digital relay must sample enough points to compute the sine wave trajectory and to do dc offset filtering, if that is part of the algorithm. Then, once the relay has made a trip decision, it has to pickup the output relay on the relay, which is still generally an electromechanical device. Then there is the operating time for any lockout relays, interposing relays, or aux relays.

When you factor in all of the delays and error bands, I suspect 0.12 sec will be about right. When the breaker clearing time is added, 0.12 sec may not be enough.

The last time I went through a similar evaluation, we were trying to coordinate an upstream digital relay with a downstream electromechanical instantaneous trip. With a 5-cycle breaker, I ended up with a time delay of 0.2 sec on the upstream relay.

 

[Modula2]

David, the 0.12 sec I presented was only for the "relay tolerance and setting errors" portion of the coordination time interval (CTI). Additional items for the total CTI, such as breaker opening time, are not in that number, as I am not questioning them. In the reference I provided, total for a 5 cycle breaker total, calibrated static relay, is 0.20 sec. For the time overcurrent(O/C) portion, I can agree with you in principle on CT errors, etc. That portion is not usually an issue.

However, for the Instantaneous Definite Time region, if all such errors still keep us in that region, then all we have is relay timer accuracy, in my case +/- 0.25 cycle, for 0.50 cycle margin, or 0.01 sec at 50 Hz, or 0.0083 sec at 60 hz.

I don't intend to go below 0.20 sec, but I would like to understand where the numbers come from. I think the O/C region and the DT regions must have separate numbers and the 0.20 CTI is a generalization that needs to be fine-tuned in each app.

 

[dpc]

In my case, I went directly to the tolerances given in the relay specs, assumed worst-case errors in opposing directions and added in anything else I could think of.

But you have to read the relay specs closely and make sure you are adding in all sources of error and delay, not just timer accuracy.

 

[Modula2]

I think you are right dpc. I will do that and go into more details. For saturation, I found GET-8501 helpful, although my app is another mfr. Thanks

 

[dpc]

The CT saturation is a tough one because digital relays use different algorithms to deal with it (each one is the best of course!) and there is limited real-world test data.

But if you're dealing with metal-clad switchgear, those CTs are usually low accuracy (C100) and can easily saturate on close-in faults. Assumption has always been that the instantaneous unit should operate before saturation gets severe, but who really knows.

I don't think the IEEE guidelines are too useful when dealing with instantaneous and definite-time delays.

 

[slavag]

Hi.
David and Dpc covered all are well.
I would like only add some example of calculation, what I use ( BTW, this calculation are not important, in all cases we put time grading min 0.2sec)
Time grading = dT
Te = tolerance of the relay operation time
Tr = overshoot time (needed time for canceled trip ).
Tcb = CB operate time.
Tm = marginal time (delay on aux relay, by CT saturation, etc, actually about 20ms, in practice it's enough because all settings are delayed).
for digital relay.
dT= 2xTe ( 2x25ms)+ Tr(30ms) + Tcb ( 50ms )+Tm ( 20 ms )=
150ms
Regards.
Slava

 

[swz202]

0.25 cycle = 0.00417 sec, isn't it?

 

[jghrist]

The relay timer accuracy may be ±0.25 cycles, but is this after the filtered current magnitude exceeds the pickup? I don't think you can expect filtered current magnitude to reach full value until after one cycle of samples (usually 16 samples per cycle).

 

[davidbeach]

The filter takes a cycle to match the value following a step change, though some relays claim a half cycle filter. But it the step change was sufficient, the output of the filter can exceed the pickup value anywhere between a 1/4 cycle and a full cycle. Timer accuracy is based on the element being picked up, so it is somewhat removed from the actual value being input to the filters.

There is also the issue of how fast an element resets when current falls below the pickup, and how far below pickup it needs to go for the element to pickup.