An integrated scheme for a directional relay in the presence of a series-compensated line

Highlights

• We present an integrated scheme for a directional relay in series-compensated lines.

• Case studies of various fault and system conditions have been addressed.

• Many simulation experiments conducted to explore operability of the proposed scheme.

• The scheme proposed does not require voltage inversion detection.

• Effectiveness and robustness of the proposed protection scheme are validated.

Abstract

This paper presents an integrated method to identify the fault direction in the presence of a series compensator via also identifying the power flow direction. The presented method relies on monitoring both measured angle of the positive-sequence (PS) currents and magnitude of PS voltage. The change of PS voltage magnitude is used to detect the current inversion case. The presented criteria depend on the power flow direction before the fault occurrence. Therefore, a power flow direction identification method is presented using only the change of the phase angle of the PS current. Different system configurations are investigated to test the reliability of the proposed protection scheme (PPS). The power and protection systems’ configurations contain different capacitor ratios leading to current and voltage inversion scenarios, reversing the power flow direction and single-pole tripping. Simulation experiments conducted via the software ATP-EMTP has shown the proper operability of the PPS.

Introduction

In present scenarios, to meet the energy demand under the existing right-of-way, utilities are sometimes forced to use the transmission capability of the transmission line (TLs) up to their thermal limits. Flexible AC transmission system devices are one of the alternatives for those problems for existing and upgraded lines. The series capacitor is a most encouraging flexible AC transmission system device which is extensively utilized for improving the transmission capability. The protection of the series compensator is ensured by a metal oxide varistor (MOV) and a parallel gap arrangement. It takes part in increasing active power flow, mitigating sub-synchronous resonance, enhancing transient stability, damping of power oscillations, and limiting short-circuit currents. However, the insertion of a series capacitor results in the abrupt change in the apparent line impedance at the point of compensation of the TL, voltage inversion, current inversion, high-frequency oscillation, and also MOVs and spark gap arrangement makes the system highly nonlinear [1]. For a series-compensated TL (SCTL), distance, direction, fault location, and differential protection schemes are affected and lead to the mal-operation of protection schemes at different conditions [2], [3], [4], [5].

Different fault direction identification methods are presented using either single-end measurements [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14] or double-end measurements [15], [16], [17], [18], [19], [20]. In [1], an algorithm relied on the PS component (PSC) current, and PSC voltage for a SCTL is presented. In [2], an algorithm relied on the combination of both the underlying three-phase currents and voltages, and an artificial neural network (ANN) is presented. However, this scheme needs large samples and pieces of training for knowledge representation, and also it cannot manage the uncertainty factors of the TL. In [3], [4], [5], other schemes have been presented based on four classifiers using negative- and PS components. The four classifiers are integrated using a voting algorithm to obtain the direction of the fault. In [7], the fault direction is identified via monitoring the apparent power, and diverse fault types are addressed. In [8], an algorithm relied on current is presented. The S-transform was employed to calculate the phase-angle difference to determine the fault direction. In [9], the direction of a fault for transmission lines with a thyristor-controlled series capacitor is identified using a voting technique (VT), including three PSC-based classifiers. The algorithms reported in [1], [2], [3], [4], [5], [6], [7], [8], [9], need to initially determine the power flow direction before any action because the miss detection of power flow direction may lead to mal-operation.

In [10], an algorithm relied on the phase-angle difference between the proposed reference signal and the post-fault current is presented. In [11], a method relied on superimposed PSC currents is presented. This scheme is mainly used for single-pole tripping conditions. In [12], a scheme relied on only the current signal for distribution networks. In [13], [14], an algorithm relied on the fault component energy, where under reverse faults, the change in fault component energy is positive, and vice versa. In [15], the sub-cycle frequency-based fault classification algorithm and fault direction discrimination are demonstrated. This algorithm has relied on the initial change of the voltage and current waveforms. However, this algorithm requires specially-designed transducers and needs a high sampling frequency. In [16], an algorithm relied on the superimposed impedance is introduced. The superimposed impedance is expressed from the instantaneous current and voltage samples, and then it decides the fault direction. In [17], an algorithm relied on the angle difference between fault and pre-fault PS current phasor is introduced. This algorithm is applicable under the single-pole tripping (SPT) condition. In [18], the performance of a directional relay based on Clarke transformation (CLT) applied to the SCTL is investigated. Also, three state factors influence the performance of the directional relays: MOV operation, sub-harmonics, and effective source impedance. In [19], a fault direction detection scheme for a double-circuit line with a thyristor-controlled series capacitor based on principal component analysis is presented. In this scheme, the principal component analysis of angles between negative- and PSCs has been implemented in order to extract the pattern and to estimate the direction of fault accurately. In [20], fault detection, classification, and direction estimation are based on the superimposed energy components. Under a reverse fault, the superimposed energy is positive, and vice versa. In [21], an algorithm that has relied on the fault, and pre-fault current samples is introduced, and the summation and direction of power flow at the normal conditions determine the fault direction. In [22], an algorithm relied on the differential impedance phase angle is presented. The ratio between voltage phasors and current phasors are used to obtain the differential impedance. In [23], [24], a scheme based on fault component integrated impedance (FCII) is introduced. This scheme depends on the source impedance, in which FCII is very small for an internal fault. In [25], a scheme based on apparent powers is presented. However, this scheme needs information about the source. Moreover, the schemes shown previously in [23], [25] have used a communication channel, which has the disadvantage of reliability lack in communication systems in practice. In [26], a protection scheme for a double-circuit TL is based on a combination of the discrete wavelet transform (DWT) and support vector machine (SVM), in which the SVM is trained using detailed coefficients of the WT of voltage and current signals. Despite time–frequency‐based schemes are faster; they are sensitive to noise. In addition to that, their need for high sampling frequency is the main problem. In [27], a scheme relied on the superimposed PS components of current is introduced for distribution and transmission systems. In [28], a fault direction identification method depending on the fault inception angle is introduced. In [29], an algorithm relied on both discrete Fourier transform (DFT), and the fuzzy logic system is presented, in which there is no need for any communication links. But most schemes, such as those reported in [5], [9], [29], are relying on large samples and pieces of training for the representation of knowledge, which excessively complicates their job. Also, they are not able to deal with TL uncertainty factors that may affect the reliability of fault direction. In this study, the PPS does not need large samples and pieces of training, and this enables speedy and precise measures for TLs.

In [30], the presented protection scheme has relied on the superimposed current component only, and there is no need for the voltage signal. In [31], the presented protection schemes are based on superimposed PSCs (SPSCs) by comparison of the phase angle between current and voltage measured at the relay point. But any schemes such as [30], [31] that use a fixed threshold value for the fault direction detection have limitations. Firstly, a new threshold should be considered with a change in the system configuration. Secondly, many factors may hinder the effectiveness of the conventional schemes, such as atmospheric conditions, and line construction, different causes and locations of faults, and parameter uncertainty. In this work, the PPS does not use a threshold value but uses reliability coefficients to alleviate this problem.

Different fault identification methods are presented using signal processing techniques [32], [33], [34], [35], [36]. In [32], an algorithm relied on both the alienation coefficient and the Wigner distribution function has been proposed. Also, statistical relations are proposed for estimation of fault location using peak values of the proposed fault index. In [33], [34], introduces an algorithm for the recognition of faults in the grid to which a solar photovoltaic system is integrated. The fault index calculated by multiplying the Wigner distribution index and alienation index. In [35], addressed the capability of the wind energy conversion system to remain integrated into the utility network and the effect of fault ride through in methods of wind plants.

It is found that a conventional protection scheme is inadequate to provide the direction of fault for all situations: current inversion (CI), voltage inversion (VI), and power flow change (PFC) for the SCTL. To redress this gap, in this study, a proposed protection scheme (PPS) relying on superimposed components is used for detecting the fault direction in a SCTL. The pre- and post-fault PSC currents and pre- and post-fault PSC voltages are investigated.

In this study, the problem of the PFC will be solved relying on the rule that the phase angle changes by 180° when the pre-fault direction changes. Then, an integrated protection scheme of a directional relay based on pre- and post-fault PSC current and pre- and post-fault PSC voltage is presented. This protection scheme does not require any VI detection. Case studies of various fault and system conditions have been addressed to verify the protection characteristics, such as SPT, CI, and VI, with and without MOV operations, high fault resistance (HFR), inception angles, distances, fault types, different sampling frequency, variation in operating voltage, frequency, source impedance, and short-circuit capacity (strength), and different TL length. Other conditions of a system are implemented like PFC, and they are found to be accurate.

The significant features of the PPS can be summarized as follows:

• A conventional directional relaying algorithm uses fault voltage and current phasors to derive the decisions and, thus, finds its limitation at a voltage or current inversion. This paper proposes a fault direction estimation technique for a SCTL using phase change in positive-sequence current and magnitude change in the positive-sequence voltage at fault.

• The performance of the PPS is not affected by CI and VI phenomena in SCTLs.

• The PPS is dependent on the measurements captured from one end.

• The PPS does not require high sampling frequency, unlike the traveling wave protection scheme.

• Artificial intelligence techniques are not needed in the PPS. This indicates that the PPS scheme is a simplified scheme compared with other schemes in the literature.

The rest of the paper is prepared in eight sections as follows: Section 2 points out the different features and challenges presented in the works introduced in Section 1. Section 3 presents fault analysis and direction detection during VI and CI. In Section 4, a description of the PPS is presented. Cases studied, results and their discussion are given in Section 5 in detail. In Section 6, the generality of the PPS is explored for an adapted three-machine system. In Section 7, the effectiveness of the PPS is compared with the few previously available algorithms, and different aspects of these algorithms are presented. Section 8 presents the conclusions drawn out from this study.