Operation of the System of Coupled Low-Voltage Feeders during Short-Circuit Faults ^†

Abstract

As a technique to control the voltage drop at network peak periods and voltage rise at middays when a high number of rooftop photovoltaic systems exist in a low-voltage feeder (LVF), two or more neighboring LVFs can be coupled. To add voltage controllability to the coupling point, a distribution static compensator (DSTATCOM) can be installed. An important issue for such a system is its operation under short-circuit conditions in one of the LVFs and relevant protection aspects. This paper investigates the performance of such a system under fault conditions and presents a protection scheme that can achieve the desired operation of the system, under short-circuit faults in either of the LVFs. The performance of the system of coupled LVFs is investigated by numerical analysis in MATLAB while the dynamic feasibility of the proposed technique is validated by simulation studies in PSCAD/EMTDC.

1. Introduction

Although the medium voltage (MV) lines supplying industrial customers usually have a loop configuration, the MV lines supplying residential low-voltage feeders (LVFs) traditionally have a radial or open-ring structure, especially in urban and semi-urban areas [1]. Thus, until recently they were generally classified as passive systems with largely predictable unidirectional power flow. To improve the maximum loadability index of the MV lines as well as to increase the system reliability indices in addition to providing self-healing capability for the MV lines and to reduce the amount of energy not supplied to the residential customers during faults in an MV line, it is suggested that these MV lines also to be in the form of a loop or mesh configuration [2,3]. In a looped MV line, sections of the line at the fault downstream can be reverse supplied if proper isolating switches are in place and active. Ref. [4] has conducted a feasibility study for upgrading the MV lines from radial and open-ring topologies into a loop arrangement. The proposed loop arrangement in [4] is based on the direct and permanent connection of the MV lines in either form of

• two MV lines fed from the same transformer make a loop arrangement;
• two MV lines fed from different transformers of a substation make a loop arrangement; and
• two MV lines fed from different transformers of two substations make a loop arrangement.

Ref. [2] has suggested installing sectionalizing switches and auto-reclosers in the MV lines and coordinating their operations. On the other hand, it is suggested in [5,6,7] that a loop arrangement can be achieved indirectly through a back-to-back connection of two MV lines via power electronics-based voltage source converters (VSCs). Ref. [7] has also presented the pole-mounted structure of a pilot 6.6 kV, 1 MVA back-to-back connection. Alternatively, it is suggested in [8,9,10,11] that a VSC-based unified power quality controller, interline dynamic voltage regulator, and interline power quality conditioner can be used for connecting two MV lines, which not only facilitates the power exchange between the two lines but also enables power flow control and/or power quality improvement in the lines. In the above-mentioned arrangements, different parameters including the size/length/loading of the lines and capacities and impedances of the transformers as well as the short-circuit capacities at the substations should be clearly considered during the planning stage.

Although the concept of looped MV lines is well investigated in detail in the literature and the benefits of such a scheme are known, the LVFs supplying residential customers have a radial structure, especially in urban and semi-urban areas (see Figure 1a). Voltage drop along these radial feeders and specifically at their end nodes has been always one of the main voltage quality problems in network peak periods. Utilities usually address this challenge by utilizing conductors with larger cross-sections (which have smaller impedances) or by selecting a high fixed tap for the distribution transformers. Thus, the voltage drop at the end of the LVFs can be kept above the minimum acceptable limit of 0.9, 0.92, 0.95, or 0.96 per-unit (pu), based on the standards in different countries [12,13,14,15]. Very recently, voltage rise at the end nodes of an LVF in middays has emerged as another voltage quality problem in the LVFs that have a high penetration of rooftop photovoltaic (PV) systems [16]. Nowadays, utilities around the world consider a maximum penetration level for PVs, such as the 25% or 30% limit in Australia [17,18], to keep the voltage rise below 1.05 or 1.10 pu, based on the different standards [12,13,14,15].

Different techniques are proposed in the literature to overcome these voltage quality issues. A new control algorithm for the inverters of PV systems is proposed in [19,20] that facilitates a reactive power exchange between the inverters and the grid in order to prevent voltage rise in LVFs. As an example, ref. [21] suggests that a 16% increase in the capacity of the inverters of PV systems is sufficient to accommodate the required amount of reactive power exchange. Alternatively, it is proposed in [22,23] to control and curtail the output active power of PV systems. In addition, it is suggested in [24,25] to install battery storage systems with each PV system to control their output active power. Reference [26] has put forward the idea of controlling the inverters of PV systems such that they exchange reactive power in both network peak and off-peak periods to prevent the non-standard voltage drop and voltage rise in both periods. References [27,28] have recommended to automatically step-up and down the voltage at the beginning of an LVF to prevent non-standard voltage drop and voltage rise at its end nodes by using distribution transformers with on-load tap changers (OLTC). As another option, refs. [29,30] have illustrated that the voltage rise and voltage unbalance can be effectively controlled all along an LVF if a power electronics-based distribution static compensator (DSTATCOM) [31] or a dynamic voltage regulator is installed at 2/3rd and 1/3rd of the feeder beginning. Ref. [32] has proposed to utilize a distribution transformer with OLTC as well as a DSTATCOM to limit the voltage quality issues all along an LVF. The main limitation of the proposals in [27,28,29,30,32] is that a DSTATCOM, a dynamic voltage regulator, or a transformer with OLTC is required for every LVF of the distribution network.

A new technique is proposed in [33] to improve the voltage profile of the LVFs during both network peak and off-peak periods. In [33], the authors have recalled the two well-known techniques of coupling MV lines via a non-automatic normally-open switch (see Figure 1b) and via a back-to-back power electronics-based connection of two lines (see Figure 1c). Based on the advantages and limitations of these two approaches, the authors have proposed in [33] to couple the neighboring LVFs directly from their end nodes while installing a DSTATCOM at the coupling point (see Figure 1d). In such a case, the voltage controllability at the coupling point of the LVFs will be provided. The first task of the DSTATCOM is to guarantee a desired balanced voltage at the coupling point of the LVFs, which is achieved by controlling the VSC of the DSTATCOM to exchange reactive power with the coupling point. The other task of the DSTATCOM is to facilitate active power circulation among the phases of each LVF. Ref. [33] has proposed the proper dynamic control for the DSTATCOM in this application and has evaluated the voltage quality in the system before and after applying this technique. The main advantages of this technique are:

• possibility of coupling more than two LVFs (which is not possible using the previous two approaches unless by adding more switches or converters),
• automatic power exchange capability among the coupled LVFs (which is not possible in the non-automatic normally-open switch approach)
• less imposed costs and losses to the system (compared to the back-to-back connection approach), and
• possibility of circulating the surplus generated power by the PV systems in one phase to the other phases(s) of any of the LVFs through the DSTATCOM.

By modifying the inherent radial nature of an MV line or an LVF into a loop configuration, the protection aspects of such a system need to be investigated and suitable protection schemes need to be developed and applied. The MV lines are usually equipped with overcurrent protection functions, which are coordinated for radial networks (with an assumption of unidirectional power flow) [1]. The first important fact about loop MV lines is the increased short-circuit current [4]; thus, it is required for the overcurrent protection functions in the distribution network, consisting of looped MV lines and distributed generation units to adopt new settings based on network conditions. In this regard, an algorithm is developed in [34,35]. In a similar trend, ref. [36] has developed an algorithm that can detect faults in a looped MV line using the trained neural network.

The second important fact about looped MV lines is that under short-circuit conditions, the current direction is from the MV lines towards the fault point, and thus, the fault current direction in an MV line can be the same or opposite of the direction of the post-fault current. From this perspective, refs. [37,38] have proposed to use directional protection function for looped MV lines while [39,40] have suggested utilizing communication-assisted protection functions such as permissive overreaching trip transfer (POTT) for accurate detection of the faulted MV line. References [41,42,43,44,45] have suggested using optimization techniques for the coordination of the directional protection functions in these applications.

Radial or open-ring MV lines (in Australia, Asia, and Europe) usually have a delta connection which results in no or a very small zero-sequence component of the current [46]. Thus, a zero-sequence component of current corresponds to a short-circuit fault in the MV lines. However, another important fact about looped MV lines is the presence of circulating currents in the system due to the unbalanced serial impedance of the lines or coupling with parallel lines [46], which, in return, generates zero-sequence components of current (up to a few amperes) in normal conditions. Ref. [38] has presented a modified protection scheme for an experimental looped MV line using a zero-sequence protection function.

Even though the protection schemes of looped MV lines are still under development, as the distribution system philosophy is being changed from passive networks into active networks (e.g., by the appearance of distributed generation units, microgrid formation in one part of the network, islanding features, etc.), the protection considerations of the system of coupled LVFs have not been addressed before in the literature and needs to be developed. This is the knowledge gap that is focused on in this research. In addition, in this research, a DSTATCOM is assumed to be present at the coupling point of the LVFs. It has been discussed in [47,48,49] that a DSTATCOM may change the equivalent impedance of the system during short-circuit faults and its impact needs to be considered and compensated by proper coordination of the settings of the protection systems. This issue increases the complexities of the performance of the considered system of coupled LVFs under short-circuit studies and its protection aspects.

The performance of the system of coupled LVFs during short-circuit faults in one of the LVFs is discussed in this paper and the desired operation for the system is proposed. Moreover, the protection aspects of the system of coupled LVFs are discussed in detail and an effective protection approach is proposed which facilitates the proper dynamic operation of the system when a short-circuit fault occurs in either of the coupled LVFs. MATLAB-based numerical analysis is used to introduce the system performance under short-circuit conditions while the dynamic feasibility of the proposed protection approach is evaluated with the help of PSCAD/EMTDC-based simulation studies.

The main contributions of this paper are:

• to investigate the desired operation of the system of coupled LVFs under fault conditions,
• to develop the proper protection approach for the system of coupled LVFs, and
• to evaluate the dynamic effectiveness of the proposed protection scheme.

The rest of the paper is organized as follows: The system under consideration is introduced in Section 2. Section 3 discusses the protection considerations of the considered system while the proposed and developed protection approach is introduced in detail in Section 4. Section 5 presents the results of some numerical analyses of the system of coupled LVFs under short-circuit faults and the results of the dynamic evaluation of the proposed protection scheme for this system. The general conclusions and findings of the research are summarized and highlighted in Section 6.

2. System under Consideration

Let us consider the MV line in Figure 1a that supplies two LVFs, namely LVF-1 and LVF-2, through two distribution transformers. The MV line is assumed to be a three-phase, three-wire system while each LVF is a three-phase, four-wire, mutual-earthed neutral (MEN)-type network [50]. Each distribution transformer is assumed to be a three-phase Dyn type. This is a common practice in Australia and the majority of Asian, European, and African countries but is not common in North America [1]. In Australia, the LVFs are usually supplied by a 25–315 kVA three-phase distribution transformer and are usually about 400 m long [50,51].

Figure 1d illustrates the considered system of coupled LVFs in this research. It is to be noted that although the coupled LVFs in Figure 1d are supplied by the same MV line, all provided discussions in this paper are general and thus valid and applicable even if they are supplied from different MV lines.

The DSTATCOM, installed at the coupling point of neighboring LVFs, is composed of a capacitor-based neutral-clamped dc link, a three-phase, three-leg VSC composed of insulated gate bipolar transistors (IGBTs) and anti-parallel diodes, and proper LC-type output passive filters [33]. The structure and control principle of the DSTATCOM is given in Appendix A.

Each radial LVF is protected against the short-circuit faults in the LVF by a switch-fuse (SF) at the secondary side of the distribution transformer, namely SF-1 and SF-2 [51]. Each SF has backup protection, which is a cut-out fuse at the primary side of the distribution transformer. The cut-out fuse also protects the distribution transformers in case of traveling surges (from switching or lightning) in the MV line.

Before discussing the protection schemes for the network in Figure 1d, let us consider the network in Figure 1b, which is the same as the system in Figure 1d if the DSTATCOM is ignored. The short-circuit fault conditions and the protection scheme will be discussed for this system first. Then, it will be modified for the system in Figure 1d to consider the presence of a DSTATCOM at the coupling point of the LVFs.

In the system in Figure 1b, depending on the PVs generation and load demand, a bidirectional current can flow in either of the LVFs in normal conditions; however, during short-circuit faults, a current will flow from the upstream side of all LVFs (i.e., the corresponding MV lines) to the fault point. In the rest of this section, first, the fault current contribution of each LVF for a short-circuit fault in one of the LVFs is investigated. Then, based on the findings of this investigation, the proper protection scheme is investigated for this system.

3. Desired Operation of the System of Coupled LVFs under Fault Conditions

As discussed above, a bidirectional power flow can be observed in either of the LVFs and through the switch in the system in Figure 1b and in normal conditions, as shown in Figure 2a. This figure also illustrates the protection zones of the system. Following the occurrence of a short-circuit fault in either of the LVFs, current flows from the upstream side of both LVFs towards the fault point, as shown in Figure 2b(i). It is desired to isolate the coupled LVFs immediately after fault occurrence with the help of the switch, which is located at the coupling point of the LVFs, as shown in Figure 2b(ii). It is to be noted that the switch can be a mechanical or a power electronics-based type (see Appendix B).

Appendix C discusses the analytical approach to calculate the short-circuit current under various fault types. If the switch isolates the LVFs, the fault current will be only supplied from the faulted LVF side (i.e., LVF-1) until SF-1 isolates the fault, as shown in Figure 2b(iii). However, if the switch fails to isolate the LVFs in time, both SF-1 and SF-2 will see the fault and both LVFs will be tripped, as shown in Figure 2c(ii,iii), which is an undesirable consequence of switch failure. A proper protection scheme for the switch, as well as the required settings, are discussed below:

Following a short-circuit fault in either of the LVFs, a portion of the short-circuit fault flows through the switch. Thus, the opening operation of this switch can be based on an overcurrent protection function, otherwise popularly known as the ‘phase 50′ function (see Figure 2d). In this case, the current pickup level (I_P) for the overcurrent function needs to be higher (i.e., 1.1–1.25 times larger) than the largest nominal current of both distribution transformers [52], i.e.,

$$I_{\mathrm{P}}=(1.1–1.25)\times \frac{\max (S_T^1,S_T^2)}{\sqrt{3}{V}_{\mathrm{LVF}}}$$

(1)

where V_LVF is the nominal line-line voltage of the LVFs (e.g., 380, 400, or 415 Vrms based on the different nominal voltages of different countries), $S_T^1$ and $S_T^2$ are assumed to be the ratings of the distribution transformers in LVF-1 and LVF-2, respectively, and max(.) represents the maximum function. Thus, the switch will not cause nuisance tripping when one of the distribution transformers supplies both feeders (under emergency or maintenance periods). On the other hand, the time multiplier setting (TMS) of the overcurrent protection function should be designed such that for a fault in either of the LVFs, the switch trips before SF-1 and SF-2. Otherwise, undesirable tripping will occur if SF-2 trips LVF-2 faster than the switch.

If the fault contribution of the healthy LVF (i.e., LVF-2) is smaller than I_P (which can happen due to the fault type/location/impedance or the impedance of the LVFs or distribution transformers), the overcurrent protection will fail to open the switch. To avoid such issues, another protection function can be used in parallel with the overcurrent protection function.

Since during a short-circuit fault in either of the LVFs, the voltage rms in the faulted phase(s) drops all along both LVFs, an undervoltage protection function, otherwise popularly known as the ‘phase 27′ function, can be utilized as a backup protection function to the overcurrent protection function. Let us define V_P (e.g., 0.6 pu) as the pickup level for the undervoltage protection function. The undervoltage protection function will be coordinated to open the switch if the voltage rms on both sides of the switch drops below V_P. Thereby, the switch will open as soon as either of the overcurrent or undervoltage functions flags a fault. Note that the protection approach has to be applied for each phase of the system and the switch needs to open after a fault flag is issued from either of the phases. This is illustrated in the ladder diagram in Figure 2d.

Based on the above discussion, the desired sequence of the operation of the protective devices should be defined as discussed below:

• Fault occurs at t = t₀;
• Action-1: The switch at the coupling point of the LVFs trips at t = t₁ (based on the command of the overcurrent and/or undervoltage functions);
• Action-2: The SF at the secondary side of the distribution transformer of the faulted LVF (i.e., SF-1) trips at t = t₂;
• Action-3: The SF at the secondary side of the distribution transformer of the non-faulted LVF (i.e., SF-2) trips at t = t_2′ as backup protection when action-1 fails;
• Action-4: The cut-out fuse at the primary side of the distribution transformer of the faulted LVF trips at t = t₃ as a backup protection when action-2 fails;
• Action-5: The cut-out fuse at the primary side of the distribution transformer of the non-faulted LVF trips at t = t_3′ as a backup protection when action-3 fails;
• Action-6: The first protective device in the MV line trips at t = t₄ when either action 4 or 5 fails;

in which t₀ < t₁ < t₂ < t₃ < t₄ and t₀ < t₁ < t_2′ < t_3′ < t₄. Note that t₂ may be equal, larger, or smaller than t_2′. Similarly, t₃ may be equal, larger, or smaller than t_3′; however, t_3′ > t₂ and t₃ > t_2′. The difference between t₂ and t_2′ as well as the difference between t₃ and t_3′ depends on the portion of the fault current passing through the SFs and cut-out fuses of the LVFs. Such coordination will be achieved by applying proper TMS for the overcurrent function and coordinating it with the time-current characteristics of SF-1 and SF-2 as well as by applying a proper delay-time for the undervoltage protection function. This is illustrated schematically in Figure 3.

After fault clearance, the SF will close and will energize LVF-1, as shown in Figure 2b(iv). Let us assume it happens at t = t₅ (where t₅ may be much larger than t₄ in case of non-transient faults). The switch is still at the open status. The voltage rms in all phases on both sides of the switch will be restored to the acceptable limits at t = t_5′ (where t_5′ > t₅). At this condition, the protection scheme can initiate a command for the switch to close, as shown in Figure 2b(v). This will occur at t = t₆, where t₆ = t₅ + ΔT₁ (i.e., after a fixed delay time of a few seconds or minutes). An overvoltage function, otherwise, popularly known as the ‘phase 59’ function, with a voltage pickup of V_P’ (e.g., 0.9 pu) can be used for this purpose. This is also illustrated in the ladder diagram in Figure 2d. In this figure, the switch will open when its tripping coil is energized and will close as soon as it is de-energized.

4. Operation of the System of Coupled LVFs in the Presence of a DSTATCOM

The discussions of Section 3 are focused on the system of coupled LVFs without a DSTATCOM (Figure 1b). The protection system will be more complex when a DSTATCOM is installed at the coupling point of the LVFs (Figure 1d) [47,48,49]. The main reason is that due to the presence of a DSTATCOM and its connection with multiple LVFs, a multi-switch scheme is required in which a switch is installed between the end node of each LVF and the DSTATCOM. Appendix D presents the analytical approach to calculate the short-circuit current under various fault types in the presence of a DSTATCOM.

Figure 4a illustrates a system of N-coupled LVFs where N ≥ 2. This system has N + 1 protection zones (i.e., one protection zone for each LVF and one protection zone for the coupling point), as illustrated in the figure. Similar to the discussions of Section 3, it is desired that following a short-circuit fault, only the faulted LVF gets isolated while the DSTATCOM is desired to remain connected to the non-faulted LVFs (see Figure 4b for desired performance of the system under a short-circuit fault in LVF-1).

Immediately after fault occurrence, current flows from the MV line(s) towards the fault point in all LVFs. In addition, current will flow from the dc link of the DSTATCOM toward the fault point. An important issue to be considered in such a case is that since the DSTATCOM is supplied by a dc capacitor, its stored energy is limited and it will discharge immediately after fault-occurrence (e.g., in less than 2 cycles); thereafter no current will be supplied by the DSTATCOM and the system will resemble the system in Figure 2a, with the only dissimilarity of higher number of coupled LVFs. It is to be noted that if the DSTATCOM has low-voltage ride-through capability or if it is supplied by an energy storage system such as a battery instead of a dc capacitor, it will continue to feed the fault for a longer period. However, the DSTATCOM proposed in [33] is composed of a dc capacitor to reduce the costs while providing the same system performance under normal conditions.

If the previously described overcurrent and undervoltage protection functions (in Figure 2d) are utilized for the switches, they will all sense the fault current and will sense the same voltage drop; thus, they will all trip (almost simultaneously or with a small difference in time) after a fault in either of the LVFs. Thereby, the DSTATCOM will also get isolated undesirably. To prevent this, a directional protection function is suggested.

Let us consider the assumed current directions for each switch as shown in Figure 4a, which is from the coupling point of the LVFs towards the distribution transformers. Once either of the overcurrent or undervoltage functions detects a short-circuit fault, it will also check the direction of the fault current. The switch which detects a forward fault will open but the switches that detect a reverse fault will remain closed. In Figure 4b(i), after a short-circuit fault in LVF-1, a forward fault is detected for SW-1 while a reverse fault is observed for the other switches (i.e., SW-2, 3, …, N). Thus, it is expected that only SW-1 will open while the other switches and the DSTATCOM will remain closed, as shown in Figure 4b(ii). Similar to the discussions of Section 3, this will be followed by the opening of SF-1 to trip LVF-1, as shown in Figure 4b(iii). In addition, after fault clearance, first SF-1 will close and will energize LVF-1, after which the overvoltage protection functions will detect acceptable voltage rms on both sides of SW-1 and it will close with a delay of ΔT₁. The ladder diagram of this scheme is shown in Figure 4c for Switch-1. A similar ladder diagram will be utilized for all switches.

Detecting the forward or reverse fault can be achieved through different techniques [52]. It can be detected by comparing the angle difference between the current and the voltage as well as the direction of active power or reactive power. Alternatively, the sequence components can be used for the detection [53].

Positive-sequence voltage and current are the only sequence quantities that are present for a TLG fault. These quantities can be used to produce a single three-phase directional element to detect the fault direction in case of a TLG fault (see Appendix E). For this, the positive-sequence directional element torque, otherwise popularly known as the ‘T32P’ function, can be defined as [54]

$${\tau }^{\mathrm{T}32\mathrm{P}}=\left|V^{\mathrm{po}}\right|\left|I^{\mathrm{po}}\right|\mathit{cos}\left(\angle V^{\mathrm{po}}-(\angle I^{\mathrm{po}}+\angle Z^{\mathrm{po}})\right)$$

(2)

where V^po and I^po, respectively, show the positive-sequence component of the voltage and current at the switch, and Z^po is the positive-sequence impedance of the line while |x| and ∠x, respectively, represent the absolute and angle of x. In (2), the positive torque represents a forward TLG fault while the negative torque represents a reverse TLG fault. In the ladder diagram in Figure 4c, for a forward TLG fault, the T32PF contact will close while for a reverse TLG fault, the T32PR will close. As can be seen from this figure, the T32P function is supervised by phase 50 and phase 27 protection functions.

For asymmetrical LG, LL, and DLG faults, (2) may lead to erroneous results in the fault direction discrimination. Thus, another approach will be used for the detection of the direction of such faults that is based on the negative-sequence impedance. For this, first, the negative-sequence component of voltage at the coupling point, denoted by $V_{\mathrm{CP}}^{-}$, and the negative-sequence component of current passing through switch-m, denoted by $I_{\mathrm{sw}-m}^{-}$, should be calculated. The negative-sequence impedance for switch-m, denoted by $Z_{\mathrm{sw}-m}^{-}$, can then be calculated as [55]

$$Z_{\mathrm{sw}-m}^{-}=\frac{V_{\mathrm{CP}}^{-}}{I_{\mathrm{sw}-m}^{-}}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\forall m\in \{\mathrm{1,2},\dots N\}$$

(3)

The following important points should be considered in this regard [53]:

• For both forward and reverse asymmetrical faults, $V_{\mathrm{CP}}^{-}$ is always negative;
• For a forward asymmetrical fault for switch-m, $I_{\mathrm{sw}-m}^{-}$ lags the driving voltage by the characteristic angle of the line and is considered positive;
• For a reverse asymmetrical fault for switch-m, $I_{\mathrm{sw}-m}^{-}$ is in the reverse direction (180° phase shift) of the forward fault; and
• For a forward asymmetrical fault, $Z_{\mathrm{sw}-m}^{-}$ is always negative and for a reverse asymmetrical fault, $Z_{\mathrm{sw}-m}^{-}$ is always positive.

In the ladder diagram in Figure 4c, for a forward asymmetrical fault, the T32QF contact will close while for a reverse asymmetrical fault, the T32QR will close. Similar to the T32P function, the T32Q function is also supervised by phase 50 and phase 27 protection functions. Alternatively, another suitable directional protection scheme such as those presented in [37,56,57,58,59] can be utilized.

The above discussions have focused on short-circuit faults in one of the coupled LVFs. On the other hand, a short-circuit fault may also occur on the coupling point of the LVFs (i.e., the connection point of the DSTATCOM). This fault will be detected as a reverse fault by all switches (i.e., either the T32PR or T32QR contact will close for all switches). This can be detected by a switch assuming a proper POTT is received from the T32PR and T32QR functions of all switches (as seen in Figure 4c). Under such conditions, it is desired that all LVFs isolate from each other in addition to the isolation of the DSTATCOM from the coupling point.

As discussed at the beginning of this section, the energy stored in the dc link of the DSTATCOM will discharge in a few cycles after a fault in either of the LVFs. This is valid for all fault types. Quicker fault detection by the protection scheme and operation of SW-1 will prevent a full discharge of the dc link. The DSTATCOM dynamic controller will facilitate the dc link voltage restoration after the opening of SW-1 (regardless of the level of discharge). Thus, a different protection scheme is developed for the switch located between the DSTATCOM and the coupling point of the LVFs. This switch is expected to isolate the DSTATCOM from the coupling point of the LVFs only based on a significant and sustained voltage deviation on the coupling point that can be detected by undervoltage/overvoltage protection functions with voltage pickups of 1.2 and 0.6 pu and after a time delay of ΔT₂ where $(T_2<T_1$). It is to be noted that this switch is not expected to respond to short-circuit faults in either of the LVFs unless a significant and sustained voltage deviation is observed due to a fault (which is possible in the case of high-impedance faults).

5. Case Studies and Simulation Results

As an example, assuming that the MV line in Figure 1b has a line-line voltage of 11 kV rms and a short-circuit fault capacity of 95 MVAsc, a number of contingency conditions are evaluated.

To evaluate the dynamic feasibility of the discussed protection approach for the system of coupled LVFs, several study cases are considered, a few of which are discussed below. The proposals are validated in PSCAD/EMTDC.

Let us consider an 11 kV MV line supplying two 11 kV/415 V Dyn-type distribution transformers, as illustrated schematically in Figure 1a. Only two radial three-phase, four-wire, MEN-type residential LVFs, namely LVF-1 and LVF-2, are considered in detail [60,61]. Each LVF is supplied by a distribution transformer with a rating of 200 kVA. The length of each LVF is taken as 400 m. The poles of each LVF are assumed to be located 40 m from each other. The feeders and their cross-sections are designed based on the nominal supplied power and the maximum acceptable voltage drop. The network data are for a typical LV urban residential network in Brisbane, Australia and are given in Appendix F. The residential loads are assumed to be in the range of 1–5 kW and are modeled as constant impedance type. The rooftop PV cells and inverters are modeled as discussed in [36]. The PV inverters are assumed to have a nominal rating of 1–5 kW and all are assumed to operate in unity.

First, let us consider the network in Figure 2a in which two LVFs are coupled through a single switch. The network is assumed to be initially at a steady-state condition. Two transient three-phase-to-ground faults, with a fault impedance of 1 mΩ, are later applied in the system. Fault-1 is applied in LVF-1 at t = 0.3 s and clears after 0.3 s while Fault-2 is applied in LVF-2 at t = 2.5 s and clears after 0.2 s. In both cases, the faults are detected within 0.01 s after fault-occurrence by the proposed approach; thus, the switch is opened in 0.01 s after both faults. The switch closes with a 0.5 s delay after the voltages on both sides of the switch return to the normal acceptable limit (i.e., after fault clearance). The results of this simulation are shown in Figure 5. This figure illustrates each fault and its duration, the open/close status of the switch at the period of study (i.e., 0 ≤ t ≤ 4), the three-phase voltages at either side of the switch, and the current flowing through the switch before fault-occurrence (i.e., t < 0.3 and 1.6 < t < 2.5), between fault-occurrence and opening of the switch (i.e., 0.3 ≤ t ≤ 0.31 and 2.5 ≤ t ≤ 2.51), after the switch is opened (i.e., 0.31 < t < 1.6 and 2.51 < t < 3.7) and at the switch closing time (i.e., t = 1.6 and t = 3.7). The results verify the appropriate operation of the switch (i.e., opening after detecting a short-circuit fault in either of the LVFs and closing after detecting the voltage restoration after fault clearance).

Now, let us consider the network in Figure 4a in which two LVFs are coupled through two switches. The network is assumed to be initially at steady-state condition. The same faults as discussed above are applied in this system. Similar to the previous case, the faults are detected within 0.01 s after fault-occurrence by the proposed approach. After Fault-1 (which occurs at t = 0.3 s on LVF-1), the negative-sequence impedance becomes negative and both switches detect a forward fault. Thus, only Switch-1 operates and disconnects the faulted LVF from the point of common coupling (PCC) of the DSTATCOM while Switch-2 remains closed. After Fault-2 (which occurs at t = 2.5 s on LVF-2), the negative-sequence impedance becomes positive and both switches detect a reverse fault. Thereby, only Switch-2 opens and the faulted LVF becomes isolated from the PCC of the DSTATCOM while Switch-1 remains closed. After fault clearance, the operated switch closes after the voltage restoration by a 0.5 s delay. The results of this simulation are shown in Figure 6. This figure illustrates each fault and its duration, the open/close status of each switch at the period of study, the three-phase voltages at the LVF-1 side of Switch-1, at the LVF-2 side of Switch-2 and at the PCC of the DSTATCOM before fault-occurrence, between fault-occurrence and opening of the appropriate switch, after the switch is opened, and at the switch closing time. The results verify the appropriate operation of the switches (i.e., the correct opening of the switch after detecting a short-circuit fault in the LVFs and its closing after detecting the voltage restoration after fault clearance).

It is to be noted that to focus more on the discussed protection approach and considering the fact that DSTATCOM operation and control is beyond the scope of this research, the DSTATCOM operation is not shown here.

6. Conclusions

The concept of coupling the neighboring LVFs is proposed in the literature as a technique for addressing the voltage drop and voltage rise problems of LVFs. This paper has focused on and proposed the protection for such a coupled LVF system when a short-circuit fault occurs in either of the LVFs. Through studies on an 11 kV network that supplies two neighboring LVFs via two separate distribution transformers and real network modeling, it is shown that overcurrent or undervoltage protection approaches are inadequate. Instead, the studies have demonstrated that a negative-sequence impedance-based directional overcurrent protection scheme can effectively and promptly discriminate the fault and properly protect the network by opening the corresponding switches. Through dynamic studies, the feasibility of the proposals for discriminating the fault is demonstrated for the considered system of coupled LVFs during short-circuit faults at various sections of the system. The studies have demonstrated that regardless of the fault location in the system of coupled LVFs, the developed proposal detected the faults within 0.01 s after fault occurrence, and thus, the corresponding switches open. Also, the studies illustrate that the corresponding switches close by the developed proposal within 0.5 s after the voltages on both sides of the switch return to the normal acceptable limit post fault clearance.