Technical Guide
Testing self-powered relays with SVERKER 900
Page 20(33)
Figure 24. Graphical verification for the time curve for measurement errors 10% lower than the injected ones.
These measured points are below the
curve “ABOVE”.
As conclusion, the test results can be accepted, as the measured points are reasonably between the high and
low curves.
Obviously, the tests should be repeated for the other phases, or maybe they could be done just once by
injecting three-phase faults.
5.3. Testing the protection function without pre-fault (switch onto fault).
As mentioned in par. 3.2, there can be a difference in the relay performance when the power system fault
occurs after a situation with a certain pre-fault (load) level, or without it. As also mentioned, this condition can be
connected to the “switch onto fault” situation, but it may happen when the circuit breaker is closed on an
energized and unloaded feeder.
This test aims to compare the operate time of the relay in the two conditions:
1) Pre-fault of 250 mA present for 1 second, fault current of 7 A (above I>> threshold of 4 A)
2) No pre-fault, fault current of 7 A (above I>> threshold of 4 A)
Remember please that those current values are always referred to the “test” current. It is always good to have in
mind the primary values associated to the secondary test currents.
One rule easy to remember is that 1 A in the test circuit, is equivalent to 50 A primary, no matter
which “nominal
current“ (Is) has been chosen, or which CT has been used with the WIC-1 relay.
It is enough to run 5 tests
12
with the pre-fault current of 0,25 A (Figure 25) to get an idea of the operate time,
then set the pre-fault current to 0 A and run 5 tests again (Figure 26).
For repeating the tests, MTT in pre-fault and fault instrument has been used. Different approaches are of course
possible.
12
“One shot is not enough”. Repeating the tests 5 times, gives an idea of the spread of the values, and helps to be
comfortable with the test results. We are evaluating the operate time! Note that we have not repeated the tests for the
inverse characteristics, as many shots are anyway generated. However, it could have been easily done, and we would have
seen a sort of distribution of time measurements around the mean value, probably near the theoretical value. In case of
doubts, use the statistical approach, especially if you need to report some doubtful performances.
