Fault-Line Identification of HVDC Transmission Lines by Frequency-Spectrum Correlation Based on Capacitive Coupling and Magnetic Field Sensing

Abstract

The proliferation of high-voltage direct current (HVDC) system is growing fast thanks to its distinct technical advantages such as long-distance bulkpower delivery at a lower cost, asynchronous interconnections, and reduced power loss [1]. The protection system of HVDC system must detect, identify and isolate the fault quickly to keep the system stable by isolating only the components that are under fault, whilst leaving the rest of the network in operation [2]. However, the criteria of under-voltage and voltage derivative protection systems may fail because there also arises a large voltage drop/variation on the healthy transmission line due to the electromagnetic coupling between HVDC transmission lines incurred by the sharp transient current from the faulty one [3], [4]. Thus the healthy line could be mistakenly identified and isolated, resulting in further disruption and delaying system restoring time. In order to distinguish the healthy HVDC transmission line from the faulted one, a fault line identification technique based on the frequency spectrum correlation is developed in this paper. Firstly, the equivalent circuit of a single-circuit HVDC system was established in Fig. 1(a)to analyze the electromagnetic coupling between HVDC transmission lines based on voltage and current of the positive $( \mathrm {U}_{p} , \mathrm {I}_{p})$and negative $( \mathrm {U}_{n} , \mathrm {I}_{n})$polarity, respectively (Fig. 1(a)). As such, the voltage of one transmission line can be correlated with the current of the other line as (taking the voltage of negative polarity as an example) [3]$\mathrm {U}_{n\_{} m}( \mathrm {t}_{2}) \quad = \mathrm {U}_{n}( \mathrm {t}_{2})- ( \mathrm {I}_{n}( \mathrm {t}_{2}) \times \mathrm {R}_{n} \quad + \mathrm {L}_{n}($$dI_{n}/ \mathrm {d}_{t})- \mathrm {M}_{pn}($$dI_{p}/ \mathrm {d}_{t}))(1)$where R n (L n ) is the line resistance (self-inductance), $\mathrm {M}_{pn}$is their mutual inductance, $dI_{n}/ \mathrm {d}_{t}(dI_{p}/ \mathrm {d}_{t})$is the current change rate within a time from $\mathrm {t}_{1}$to $\mathrm {t}_{2}$. By taking the Fourier Transform of Eq. (1), the frequency spectrum of voltage (£$( \mathrm {U}_{n\_{} m}( \mathrm {t}_{2}))$) must exhibit a similar pattern with the one of the varied current (£$(dI_{p}/ \mathrm {d}_{t})$) if this voltage variation is incurred by the current change. For example, the frequency spectrum of line current $( \mathrm {I}_{p} , \mathrm {I}_{n})$and voltage at the measurement point $( \mathrm {U}_{p\_{} m} , \mathrm {U}_{p\_{} n})$after a grounding short-circuit (lasting 0.03s) of the negative transmission line is analyzed in Fig. 1(b). Since the voltage variation of the positive pole is incurred by the current change of the negative pole, the frequency spectrum of the voltage for positive pole $( \mathrm {U}_{p\_{} m})$exhibits a high similarity with the current of the negative I n while the voltage of negative pole $( \mathrm {U}_{n\_{} m})$with the current of positive pole I p is not. Therefore, the similarity degree of frequency spectrum between voltage of a line and current of another can be used to identify reliably whether the voltage variation is incurred by electromagnetic coupling or fault, and thus avoiding false operation of protection system. In order to verify this technique, a single-circuit HVDC system was established in PSCAD/ EMTDC [5]as shown in Fig. 2(a). A grounding short-circuit fault was simulated with the system operating in the normal status under 2 kA rated current. The frequency spectrum correlation coefficient C 1 describes the spectrum similarity between the voltage of positive polarity and current of the negative polarity, and C 2 vice versa) after the fault was calculated (Fig. 2(b)and (c)). The result shows that the voltage of the positive polarity and the current of the negative polarity are highly correlated C 1 remains large and C 2 drops sharply) after the fault happens at 1.00 s, indicating this voltage is induced by the current variation of negative polarity rather than by the short-circuit fault itself. As such, despite the fact that a large voltage variation on the positive polarity occurred by electromagnetic coupling, this healthy line would not be identified as the faulty line. By comparing the sampling window of 0.01, 0.005, 0.001s, it can be observed that the shorter time window for frequency spectrum analysis can provide a faster response time based on a higher spectrum resolution [6]. A capacitive-coupling and magnetic-field-sensing assisted platform composing of two paralleled induction bars, and an array of magnetic sensors at both ends was previously developed in [7], [8], and it is capable of measuring overhead line voltage and current. This platform was adopted to implement this identification technique, and the details of simulation results will be reported in the full paper. This identification technique can further enhance the reliability of HVDC system by avoiding the unnecessary outages due to accidental shut down of healthy lines, fostering the interconnection of load centers with the use of HVDC system for increased power transmission capacity.