Background:
The Static VAR Compensator (SVC) has the capability of improving reliability,
operation and control of the transmission system thereby improving the dynamic performance of
power system. SVC is a widely used shunt FACTS device, which is an important tool for the reactive
power compensation in high voltage AC transmission systems. The transmission lines compensated
with the SVC may experience faults and hence need a protection system against the damage caused by
these faults as well as provide the uninterrupted supply of power.
Methods:
The research work reported in the paper is a successful attempt to reduce the time to detect
faults on a SVC-compensated transmission line to less than quarter of a cycle. The relay algorithm involves
two ANNs, one for detection and the other for classification of faults, including the identification
of the faulted phase/phases. RMS (Root Mean Square) values of line voltages and ratios of sequence
components of line currents are used as inputs to the ANNs. Extensive training and testing of
the two ANNs have been carried out using the data generated by simulating an SVC-compensated
transmission line in PSCAD at a signal sampling frequency of 1 kHz. Back-propagation method has
been used for the training and testing. Also the criticality analysis of the existing relay and the modified
relay has been done using three fault tree importance measures i.e., Fussell-Vesely (FV) Importance,
Risk Achievement Worth (RAW) and Risk Reduction Worth (RRW).
Results:
It is found that the relay detects any type of fault occurring anywhere on the line with 100%
accuracy within a short time of 4 ms. It also classifies the type of the fault and indicates the faulted
phase or phases, as the case may be, with 100% accuracy within 15 ms, that is well before a circuit
breaker can clear the fault. As demonstrated, fault detection and classification by the use of ANNs is
reliable and accurate when a large data set is available for training. The results from the criticality
analysis show that the criticality ranking varies in both the designs (existing relay and the existing
modified relay) and the ranking of the improved measurement system in the modified relay changes
from 2 to 4.
Conclusion:
A relaying algorithm is proposed for the protection of transmission line compensated
with Static Var Compensator (SVC) and criticality ranking of different failure modes of a digital relay
is carried out. The proposed scheme has significant advantages over more traditional relaying algorithms.
It is suitable for high resistance faults and is not affected by the inception angle nor by the location
of fault.