Online Estimate Of Measurement Errors - GE L90 Instruction Manual

9.1 OVERVIEW
Depending on the 87L settings, channel asymmetry (the difference in the transmitting and receiving paths channel delay)
cannot be higher than 1 to 1.5 ms if channel asymmetry compensation is not used. However, if the relay detects asymmetry
higher than 1.5 ms, the
are enabled in the
CURRENT DIFFERENTIAL
GE's adaptive elliptical restraint characteristic is a good approximation to the cumulative effects of various sources of error
in determining phasors. Sources of error include power system noise, transients, inaccuracy in line charging current com-
putation, current sensor gain, phase and saturation error, clock error, and asynchronous sampling. Errors that can be con-
trolled are driven to zero by the system. For errors that cannot be controlled, all relays compute and sum the error for each
source of error for each phase. The relay computes the error caused by power system noise, CT saturation, harmonics,
and transients. These errors arise because power system currents are not always exactly sinusoidal. The intensity of these
errors varies with time; for example, growing during fault conditions, switching operations, or load variations. The system
treats these errors as a Gaussian distribution in the real and in the imaginary part of each phasor, with a standard deviation
that is estimated from the sum of the squares of the differences between the data samples and the sine function that is
used to fit them. This error has a spectrum of frequencies. Current transformer saturation is included with noise and tran-
sient error. The error for noise, harmonics, transients, and current transformer saturation is computed as follows. First, the
sum of the squares of the errors in the data samples is computed from the sum of squares information for the present
phaselet:
Then fundamental magnitude is computed as follows for the same phaselet:
Finally, the local adaptive restraint term is computed as follows, for each local current:
Another source of the measurement errors is clock synchronization error, resulting in a clock uncertainty term. The L90
algorithm accounts for two terms of synchronization error corresponding to:
• Raw clock deviation computed from time stamps. There are several effects that cause it to not track exactly. First, the
ping-pong algorithm inherently produces slightly different estimates of clock deviation at each terminal. Second,
because the transmission of time stamps is spread out over several packets, the clock deviation estimate is not up to
date with other information it is combined with. Channel asymmetry also contributes to this term. The clock deviation
computation is indicated in equation 8.15 as
and then average of absolute values is computed. If GPS compensation is used, then GPS clock compensation is sub-
tracted from the clock deviation.
• Startup error. This term is used to estimate the initial startup transient of PFLLs. During startup conditions, a decaying
exponential is computed to simulate envelope of the error during startup
The clock uncertainty is expressed as:
Eventually, the local clock error is computed as:
9
The local squared adaptive restraint is computed from all local current sources (1 to 4) and is obtained as follows:
2
I
LOC_ADA_A
9-12
FlexLogic™ operand is set high and the event and target are raised (if they
87L DIFF CH ASYM DET
menu) to provide an indication about potential danger.
SumSquares
1_A k
I
1_MAG_A
4
2
--- - SumSquares
I
1_ADA_A
N
. If 2 channels are used, clock deviation is computed for both channels
i
clock_unc clock_dev
clock_unc
--------------------------------- -
CLOCK
A
9
2 2
18 I I
1_ADA_A 2_ADA_A
L90 Line Current Differential System

9.1.16 ONLINE ESTIMATE OF MEASUREMENT ERRORS

N 2 1
2
4
i
--- -
1_f_A k p
N
p 0
2 2
I I
1_RE_A 1_IM_A
2
I
1_A k 1_MAG_A
start_up_error
2
2
I I
LOC_RE_A LOC_IM_A
2 2
I I
3_ADA_A 4_ADA_A
9 THEORY OF OPERATION
2
2
I CLOCK
q_ADA_A A
(EQ 9.28)
(EQ 9.29)
(EQ 9.30)
(EQ 9.31)
(EQ 9.32)
(EQ 9.33)
GE Multilin