L60 Signal Processing - GE L60 Instruction Manual

8 THEORY OF OPERATION
With respect to shunt reactors, keep in mind that the inductance (of the reactor) and capacitance (of the line) cancel each
other for the fundamental frequency only. When considering transients, an inductor is not a 'negative capacitor'. Therefore,
it is prudent to exclude the reactors from the measuring zone by subtracting the reactor current from the line CT current and
configuring the charging current compensation for the entire amount of the line capacitive current (not for the net between
the line and installed reactors). This approach is not only technically correct, but also simplifies the application by not
requiring monitoring of the status (on/off) of the reactors
Charging current is calculated and subtracted from the line current individually per phase. Depending on the number of ter-
minals on the line (two or three as configured by the 87PC function), half or one-third (in case of three-terminal line) of the
net line charging current is subtracted at each terminal. For breaker-and-a-half configurations, if the
setting value is "Two Sources Current", the charging current is subtracted per each breaker current individually and propor-
tionally to the current flowing through each breaker.
As a protection method, phase comparison is a time domain principle. It can be logically analyzed if implemented as a set
of operations on instantaneous signals, starting at the local AC currents and received DC voltages encoding the phase
information for the remote currents, and culminating at the trip integrators to measure the coincidence time between the
operating currents. Early (and still prevailing) implementations of microprocessor-based relays are generally based on fre-
quency domain processing. This means instantaneous currents and voltages are first filtered and represented by phasors,
(that is, magnitudes and angles), then trip/no-trip decisions are based upon phasors or similar aggregated values. Suc-
cessful implementations of the L60 phase comparison principle is based on instantaneous values, not phasors. There are
several reasons for using instantaneous value, the main one being the analog nature of the remote information. The trans-
mitted/received analog signal is an on/off binary signal that encodes the information not on signal magnitude, but rather on
timing with respect to actual continuous time. In addition, this signal is a subject to impairments that cannot be alleviated by
means of filtering, but by manipulations on its shape. Therefore, it is logical to process the communication signals in the
phase comparison relay in the time domain, and adjust the reminder of the algorithms to follow the instantaneous
approach, not vice versa. The time domain approach follows the methods of the last generation of analog phase compari-
son relays, giving a chance for equally good performance.
The L60 samples currents and voltage inputs at a rate of 64 samples per cycle. Current samples are pre-filtered using
band-pass Finite Response Filters (FIR), with a weighted average of signal samples in a selected data window, to remove
the decaying DC component and low-frequency distortions. The L60 implementation uses a data window of 1/3rd of a
cycle, resulting in an extra signal (phase) delay of approximately 1 ms.
The pre-filtered instantaneous currents can be used directly in phase-segregated implementations. In mixed-mode applica-
tions, they must be converted into a single composite current. This operation uses symmetrical components and may seem
at odds with the time-domain approach.
However, the conversion can be done without introducing unnecessary delay by applying a pair of orthogonal filters.
Orthogonal filters are two filters that yield phase responses shifted by 90°, and preferably have similar magnitude
responses (that is, filtering capabilities). The two filters are often labeled as direct (D) and quadrature (Q). Their outputs are
instantaneous values, but can be treated similarly to the real and imaginary parts of a phasor in the frequency domain. The
L60 implementation of a phase comparison relay uses a pair of short-window FIR filters to derive the D-Q components
while providing for extra transient filtering. Once the D-Q components are obtained, the instantaneous negative-sequence
based composite signal (I_2 – K × I_1) is created as follows (ACB phase rotation, phase A as a reference).
i
=
MIX
Note that the equation above is a linear combinations of current samples, as long as the operations of pre-filtering and
deriving the orthogonal components are linear. This guarantees security on external faults regardless of any transients as
long as the hardware/algorithms are the same at all line terminals. With both terminals applying the same linear processing,
the two mixed currents will always be out-of-phase as waveforms, regardless of their possible distortions and transients.
The quadrature component of the signal is needed to estimate magnitude of the input current for fault detectors:
i
=
MIX_Q
GE Multilin
1 1

-- - i ( ) -- - K 1 ( ) i (
1 K
– + –

D_A D_B
3 2
1 1

-- - i ( 1 K ) -- - K 1 ( ) i (
– + –

Q_A Q_B
3 2
L60 Line Phase Comparison System

8.1.10 L60 SIGNAL PROCESSING

3
) ------ - K ( ) i (
i 1 i
+ + + –
D_C Q_B Q_C
2
3
i ) ------ - K ( 1 ) i ( i
+ + + –
Q_C D_C D_B
2
8.1 OVERVIEW
87PC SIGNAL SOURCE

)
(EQ 8.2)


)
(EQ 8.3)

8-25
8