1. Introduction
As an important part of linking the power system and customers, the Medium-Voltage (MV) distribution network directly supplies power to customers. Its operational reliability directly affects the continuity of the power supply. Faults are difficult to avoid due to the complex environment and wide distribution range of the MV distribution network. Statistically, about 90% of power outages are caused by faults in the distribution network. The fast and accurate location of the fault point is critical for reducing the outage time and loss as well as improving the power supply reliability of the grid.
Numerous studies have been conducted for fault location in MV distribution networks. According to the difference of location principle and measuring points, the location methods can be classified into the following three types: (1) the impedance method, which is based on the station-side voltage or current measurement; (2) the traveling wave method, which is based on the station-side or multi-point measurement; (3) the wide-area communication method. Among them, the impedance [
1,
2,
3,
4] and traveling wave methods [
5,
6,
7,
8] are traditional location methods and have been widely applied in fault location in the transmission network. However, many difficulties arise when these methods are applied to the distribution network. The impedance method fails to completely solve the problems caused by various line types and branches. The traveling wave method faces the challenge of wavefront identification due to short lines and the complex structure of distribution network. Moreover, there are practical application challenges, such as a high sampling frequency and significant investments in equipment.
Increasing information has recently been provided for the fault location of the distribution network with the diversification and generalization of the measurement device. To overcome the difficulties caused by the complex structure of the distribution network for fault location, the method based on wide area communication has become a popular research topic [
9,
10,
11,
12,
13,
14]. The measurement objects typically include feeder currents and node voltages. Meanwhile, a synchronous phasor measurement has also been applied in the distribution network’s fault location [
15,
16,
17,
18,
19]. The method of the feeder current measurement is based on the traditional distribution network automation and fault indicator, for which the location accuracy is proportional to the number of monitoring terminals. This type of method requires the installation of current measurement devices on feeders, which requires a significant amount of installation, operation, and maintenance. Node voltage measurement-based methods usually perform voltage measurements at switching stations, ring network cabinets, and distribution transformers to realize fault location based on a sparse voltage measurement on the MV side. Currently, for a multi-point quantitative location, it is more common to introduce a current injection source at the fault point [
20,
21,
22,
23]. The voltage of the sparse measurement and node impedance matrix are used to calculate the node injection current, which reflects the fault segment. In [
24], an impedance-based method is used to derive multiple fault points. The sparse measurement data are employed to establish LV zones to exclude the pseudo-fault point. In [
25], the relevant parameters are improved on this basis to make the location results more accurate. In addition, the direct voltage matching method has also been used to perform multi-point measurements [
26,
27]. However, fault cases must be simulated for all nodes, which requires high accuracy for grid modeling and a complicated process. The fault resistance must be estimated by an iterative method, for which the accuracy is difficult to guarantee. In [
28], the property of zero reactive power consumed by the resistive fault is used to establish the relevant equations to determine the fault distance. However, it may generate pseudo-fault points, which need to be further excluded. In [
29], the sparse voltage measurements are used to calculate the fault current when all nodes subsequently fail. Faults can be found in accordance with the principle of minimum current error, which could also generate pseudo-fault points.
The above sparse-measurement based location methods are mainly based on MV side voltage measurements for fault location, which share common limitations. Although MV side voltage information is a more intuitive and accurate reflection of fault information on MV lines, it requires the installation of additional voltage transformers, which increases costs and may also pose a risk of ferro-resonance to the system. In a complex and large distribution network, the MV side requires a large number of measurement points to achieve accurate positioning, which greatly increase the cost of positioning and are difficult to apply in practical engineering. For this reason, a practical solution for the distribution network fault location based on the LV side voltage measurement is presented in this paper. In [
30], the negative sequence voltage changes at each measuring point on the LV side before and after the fault is divided into different groups, and the group with the maximum mean value is used to determine the fault location. This method may yield incorrect results for network with long laterals, and it cannot be used for the location of the three-phase fault because negative sequence voltage is selected as the single analysis object for all fault types. Voltage similarity matching has been previously used for fault location based on LV side measurements [
31]. Moreover, a single negative sequence component was used for the analysis, which cannot be used for locating three-phase ground faults. In [
32], three-phase voltage sags caused by short-time connection of the auxiliary resistor are used for fault segment identification. However, the operation of the auxiliary resistor will increase the complexity of the overall location scheme and the increased fault current will pose a threat to line insulation. A location method has been proposed based on the ratio of positive and negative sequence voltages with a limited number of measuring points on the LV side [
33]. Most existing fault location methods based on the LV side measurement employ negative sequence components as their analysis object, which cannot be used for a three-phase fault. Some location schemes may receive multiple or incorrect results for a complex network with numerous and long laterals. In addition, the same characteristic quantity has a different sensitivity to different faults. Single negative sequence voltage does not give the best location results in some cases.
Compared to the MV side, it is relatively convenient to obtain electrical quantities on the LV side. LV side measurements can be combined with terminals such as intelligent fusion terminals, electricity consumption information acquisition systems, and smart meters to collect relevant data. The operation, maintenance costs, and hardware investments of LV side measurement devices are relatively low. In view of the problems of the above MV side location methods, fault segment location based on LV side characteristic quantities is considered. Based on analysis of the distribution characteristics of each voltage on MV side and the voltage transmission characteristics of distribution transformers, significant differences were found in the influence of the various faults on the voltage characteristics of the LV side. In a single-phase fault, the phase voltage and negative sequence voltage changes on the LV side are small, but the negative sequence voltage maximizes at the fault point. For a phase-to-phase fault, the LV side phase voltage changes significantly corresponding to the two faulty phases. Therefore, the characteristic voltages of different faults were determined based on the fault distribution characteristics of each voltage quantity on the LV side. The characteristic quantity was selected adaptively according to the fault type, which is conducive to the most sensitive reflection of the fault location. The characteristic voltage is used to determine the faulty path and the fault segment search algorithm to avoid the misjudgment. These two principles were combined to achieve the fault segment location of the MV distribution network based on the characteristic voltage measurement on the LV side. Compared with the previous method based on the MV measurement, this method does not need new measuring equipment but can use the existing low-voltage measuring equipment in the distribution network, such as the electricity information acquisition system and the intelligent fusion terminal, to collect the low-voltage side voltage data, which greatly reduces the positioning cost. Compared with the previous method based on the LV side, the proposed method in this study can realize the fault path and fault segment identification only by using the voltage value of the low voltage side measuring point without particularly complicated mathematical calculation and procedures, and the implementation is simple. According to the fault type, the characteristic voltage for the segment location can be determined adaptively so that all short-circuit faults can be located. The section search algorithm is used to avoid the generation of pseudo fault points. The contribution of this work can be summarized as below:
The fault distribution characteristics of the LV side are analyzed and used for fault location.
The transmission characteristics of the distribution transformer to the voltage on both sides of the MV and LV are analyzed, which is the basis of reflecting the fault location of the medium voltage line from the characteristic quantity on the low voltage side.
The sensitivity of different voltages to different faults is analyzed and the characteristic voltages of each fault type are determined accordingly.
A method is proposed to determine the fault path and fault section by using the characteristic voltage distribution characteristics of the LV side, which does not require additional measurement equipment, greatly reduces the positioning cost, and is simple to calculate.
This method can be combined with existing MV side-based location methods to achieve more economical and accurate fault location.
The remainder of this paper is organized as follows. In
Section 2, the distribution characteristics of the voltage on the MV and LV sides is analyzed. In
Section 3, the principle and steps of the proposed fault location method are described in detail, including the configuration of measuring points. In
Section 4, the performance of the fault location method is evaluated via the two distribution network models. Conclusions are provided in the last part.
2. Fault Voltage Distribution Characteristics Analysis
When a fault occurs in the MV distribution network, the voltage at the fault point is significantly reduced and the voltage amplitude of the entire network changes to a certain extent. Hence, the distribution characteristics of the system voltage can be used to locate the fault in the network. The LV side measurement information can indirectly reflect the MV side voltage information owing to the characteristics of transformer transmission. However, the fault information of the MV side cannot be fully transmitted to the LV side, owing to the limitation of the connection mode of the distribution transformer windings. In this section, the distribution law of the LV side voltage under different faults will be discussed in accordance with the transformer transmission characteristics based on the distribution law of the MV side voltage in the entire network.
2.1. Study of Fault Voltage Distribution Law on MV Side
Figure 1 shows the fault sequence network diagram when a single-phase fault occurs in the effective grounding system, where
Zs1,
Zs2, and
Zs0 denote the positive, negative, and zero sequence impedance of the system;
V1f,
V2f, and
V0f denote the positive, negative, and zero sequence voltages at the fault point; and F indicates the fault point. The shade of the red line in the figure indicates the magnitude of the sequence voltage, which can be used to better understand the distribution law of positive, negative, and zero sequence voltages. The amplitude of the positive sequence voltage gradually decreases on the fault path due to the large fault current flowing from the source side of the system to the fault point. The voltage variation of branches is small because only the load current flows through it. The amplitude of the positive sequence voltage downstream of the fault point is close to the fault point because the load current is small. The distribution law of negative and zero sequence voltages on the faulty path is opposite to that of the positive sequence voltage. Its magnitude gradually increases from the source end to the fault point with a maximum value at the fault point, and the voltage downstream is close to the fault point with a larger magnitude. Notably, the difference in the distribution of the voltage in the network is essentially the voltage drop generated by the current on the line. The overall distribution law of the sequence voltage under a different fault is consistent, although the level sequence current significantly varies under the same law of the sequence voltage distribution. Hence, the difference in the distribution of each voltage quantity in the network under different faults is also different.
Unlike the positive, negative, and zero sequence network, which can be completely decoupled, the distribution law of the phase voltage and line voltage is more complex. For simplification, only the fault phase voltage and the line voltage associated with the fault phase are analyzed in this study. The distribution law of the fault phase voltage is similar to that of positive sequence voltage. Its amplitude decreases gradually from the source side of the system to the end of the fault downstream, and the voltage variation per unit line length downstream of the fault is significantly small. It should be noted that the variation speed of the voltage on the line is positively related to the current amplitude. The distribution law of line voltage is directly related to the system neutral grounding mode and fault type. For a single-phase fault in the small current grounding system, the LV side line voltage is almost unaffected. For the same fault in the small resistance grounding system, the line voltage reduces gradually from the source end to the fault point; for phase-to-phase fault, the line voltage reduces gradually to zero from the source end to the fault point, and the line voltage on the downstream side of the fault is close to the fault point.
Theoretical calculations and simulation verification were conducted to further compare the variation characteristics and distribution law of voltages under different fault types and neutral grounding modes. The possible variations of each voltage at the fault point before and after the fault are presented in
Table 1, where
V1,
V2, and
V0 denote positive, negative, and zero sequence voltages, and
VP and
VL denote phase and line voltages, respectively.
Table 1 demonstrates that the variation characteristics of each voltage under different fault types vary greatly. When selecting the voltage distribution for the fault location, it is necessary to combine the neutral grounding mode of the system and the fault type to adaptively select the characteristics voltage with the largest change in magnitude and the largest difference in distribution for location. The neutral point ungrounded system and the neutral point directly grounded system are two extreme cases, where the voltage variations can be directly calculated. The results are shown in
Table 1, where LG, LL, LLG, and 3L represent, respectively, the single-phase, phase to phase, two-phase to ground, and three-phase fault.
Due to the influence of system parameters and neutral grounding impedance (arc suppression coil, grounding resistance), it is difficult to accurately calculate the voltage variation in small resistance grounding and resonate grounding systems. Assuming that the arc suppression coil in the resonant grounding system is fully compensated, its typical voltage variation is consistent with the ungrounded system. In the small resistance (10 Ω) grounding system, the voltage change is between the direct grounding and the ungrounded system due to the influence of the fault distance. It should be noted that the above analysis only considers the voltage variation law on the 10 kV side of the distribution network and the MV side sequence voltage information cannot be fully transmitted to the LV side due to the influence of the distribution transformer. The voltage variation on the LV side needs to be analyzed in combination with the distribution transformer.
2.2. Effect of the Distribution Transformer on Voltage LV Side Characteristics
To further obtain the changing characteristics and distribution laws of the LV side voltage under different fault conditions, the voltage characteristics of the LV side are analyzed with the operational characteristics of the distribution transformers in this section. The 10/0.4 kV distribution transformer in China primarily adopts the Dyn11 and Yyn0 coupling methods, and the typical coupling method Dyn11 is used as an example to reveal the transmission law of each voltage quantity. Based on the transfer characteristics of the transformer, the phase voltage on the LV side can be expressed as
where
Ua,
Ub and
Uc denote the phase voltage at primary side (MV side);
,
and
denote the phase voltage at secondary side (LV side); and
K is the transformer ratio. Based on the relationship formula, the LV side phase voltage corresponds to the MV side line voltage. Further, the LV side line voltage can be expressed as follows:
where
,
and
indicate the line voltage of the secondary side. Compared to (1), the line voltage of the LV side is still a linear transformation of the MV side phase voltage, but the relationship is more complex.
To further analyze the variation law of the sequence voltage, the MV side positive, negative, and zero sequence are indicated by
U1,
U2, and
U0, and the corresponding sequence components of the LV side are indicated by
,
, and
. The corresponding relationship between these sequence components of the medium and LV side is as follows:
As shown in (3), the zero-sequence voltage cannot be transferred to the LV side. Both positive and negative sequence voltages are reduced in magnitude by the ratio when they are transferred, and their phase will be simultaneously shifted.
According to the transfer matrix of each voltage, the line voltage on the MV side corresponds to the phase voltage on the LV side, and, thus, the two are synchronized in terms of variation. For the sequence component, zero sequence cannot pass through the transformer, and positive and negative sequence components pass through the transformer and have corresponding changes in magnitude and phase, but the magnitude on both sides also stays synchronized in variation. The voltage changes on the LV side are shown in
Table 2, where the meaning of each letter is the same as
Table 1.