Analysis of power system faults (transformers, rotating machines, overhead and cable lines)

Identification of faults for protection

Both, protective relays and fault locators for overhead lines are dependent on the results of the auxiliary algorithms, which are applied for identifying fault features, such as: fault detection, fault direction discrimination and phase selection.

Fault detection is required to activate the measurement process of protective relays. In relation to specification of time intervals of the signals, fault detection can be treated as distinguishing the fault interval from the pre-fault interval.

Fault direction is understood as the calculations aimed at answering a question whether a considered fault is forward or backward with respect to the direction at which the protective relay is design to respond.

On the other hand, it is expected not to respond for backward faults, i.e. faults occurring back to the relaying point.

Phase selection is aimed at identifying the fault type, i.e. which phases are involved in a considered fault and whether it is an earthed fault or isolated one.

Fault detection

A number of approaches to fault detection are proposed in the literature. The abnormal conditions (not necessarily faults) are detected by watching the phase impedances and/or phase-current amplitudes and/or phase-voltage amplitudes and/or zero-sequence current amplitude.

Depending on a particular application, different activation criteria are combined in a different way. To speed up the fault detection, one may also apply derivatives of the relevant signals.

It is quite easy to introduce the adaptivity to such approaches. Knowing the breaker positions, monitoring the average load of a line, etc., the thresholds may be self-adjusted to improve the sensitivity and reliability of fault detection.

Much easier methods refer directly to samples of current and/or voltage waveforms. Disregarding a particular solution, two approaches (Figure 1) are commonly applied in contemporary digital protective relays:

Solution #1 – Sample-by-sample method

A sample-by-sample method computing numerically the first derivative of a watched signal. If this derivative overruns a pre-set value, an auxiliary counter starts to count up.

This counter is incremented by the absolute value of the derivative. When it reaches another pre-set threshold, a fault is confirmed. Certainly, the first threshold must be set above the maximum value of the scaled derivative under normal conditions.

Solution #1 – Cycle-by-cycle algorithm

A cycle-by-cycle algorithm compares a present sample with the sample one cycle back. The threshold for such a difference may be set much lower than in the sample-by-sample method. An auxiliary counter may be used to initiate when the absolute value of the defined difference overruns its threshold.

The detection is when the counter increased by the successive differences overreaches the second threshold.