Abstract
The increase in the number of sensitive loads in power systems has made power quality, particularly voltage sag, a prominent problem due to its effects on consumers from both the utility and customer perspectives. Thus, to evaluate the effects of voltage sag caused by short circuits, it is necessary to determine the areas of vulnerability (AOVs). In this paper, a new method is proposed for the AOV determination that is applicable to large-scale networks. The false position method (FPM) is proposed for the precise calculation of the critical points of the system lines. Furthermore, a new method is proposed for the voltage sag monitor (VSM) placement to detect the fault locations. A systematic placement scheme is used to provide the highest fault location detection (FLD) index at buses and lines for various short-circuit fault types. To assess the efficiency of the proposed methods for AOV determination and VSM placement, simulations are conducted in IEEE standard systems. The results demonstrate the accuracy of the proposed method for AOV determination. In addition, through VSM placement, the fault locations at buses and lines are detected.
POWER quality is one of the most critical factors that affect the reliability and operational efficiency of electric systems [
Various methods have been proposed to determine the AOVs. In [
It is essential to determine the fault location for the normal operation of power system and the improvement of power quality [
Numerous studies have focused on monitor placement based on the monitor reach area (MRA) concept when considering AOVs.
With the increase in nonlinear loads and harmonic emissions in power systems, the PQM placement has been investigated by considering the AOV for voltage sag and harmonic resonance mode analysis for parallel resonance [
In this paper, a new method is proposed to determine AOVs. This method has a low computational burden and is thus particularly suitable for large networks. Accordingly, to determine the AOV for a specific bus, the evaluation commences from this bus itself. Based on the connected lines and the adjacent buses, the AOV is expanded. The AOV determination is typically conducted offline, and the real-time calculation is also feasible while considering the effects of system characteristics on the AOV. In addition, a precise method is proposed for calculating the CPs. CPs are computed for lines that are partially included in the AOV of the bus. This method ensures a high accuracy in AOV determination and guarantees convergence with fewer iterations. Furthermore, given the importance of detecting the voltage sag source locations in the system, the proposed method for VSM placement is designed to accurately detect fault locations. Thus, the VSM placement is conducted based on the AOV determination to detect fault locations at both buses and lines. Considering different short-circuit fault types in the system, we perform VSM placement separately for each fault type. As the VSM placement schemes may not be unique, the optimal one is chosen to provide the highest bus FLD and line FLD indices under all short-circuit fault types.
The main contributions of this study are as follows.
1) A false position method (FPM) is introduced for the accurate AOV determination and precise CP calculation in the system.
2) A method for VSM placement is proposed for accurate FLD based on the AOV determination and system zoning.
3) Considering the fault location on the system lines, we conduct the line segmentation and an evaluation of VSM failure scenarios to obtain the optimal placement scheme for the system.
The remainder of this paper is organized as follows. Section II introduces the fault voltage equations for the AOV determination and FLD based on the MRA matrix. Section III introduces the proposed method for AOV determination and CP calculation, and Section IV introduces the proposed method for VSM placement. Simulation results are presented in Section V, and the conclusion is given in Section VI.
In this section, the fault voltage equations are derived based on the sequence impedance matrix to calculate the AOV for each bus in the system. The MRA matrix is established based on the effects of each bus on the others by determining the AOV of all buses. Thus, it becomes possible to distinguish the affected area of buses through the VSM placement, thereby enabling the FLD.
First, the sequence impedance matrix of the system is obtained. As shown in

Fig. 1 Fault occurrence at point K on line F-T.
The transfer impedance between bus j and point K and driving impedance at point K can be defined as [
(1) |
(2) |
where is the sequence impedance of line l; is the sequence transfer impedance between bus j and fault point K; is the sequence driving impedance at fault point K; is the sequence driving impedance at bus X, and X=F or T; is the sequence transfer impedance between bus X and bus Y, and ; p is the proportion of length between bus F and fault point K to length of the entire fault line; and the superscripts 0, 1, and 2 denote the zero-, positive-, negative-sequence values, respectively.
Also, the pre-fault voltage magnitude at point K is given according to the fault position and the pre-fault voltages at buses F and T as:
(3) |
Therefore, the fault voltage magnitude at bus j can be determined under different types of short-circuit faults at point K on line l. For example, when a single-line-to-ground fault (SLGF) in phase A is considered, the fault voltage at bus j for phase A is given as:
(4) |
By substituting (1)-(3) into (4), the voltage equation is derived in terms of p. The length of line segment in the AOV is then determined by solving (4), and p is thus calculated. This process is extended to all the lines in the power system to calculate the AOV. The bus vulnerability index (BVI) is obtained by considering the fault points at the buses. Subsequently, the line vulnerability index (LVI) is determined based on the BVIs of the beginning and end buses of the lines, and then the AOV is determined [
The BMRA and LMRA matrices are established by determining the AOVs for all buses based on a specified voltage threshold. In a system with N buses, the BMRA matrix has a dimension of N×N, and if bus j is in the AOV of bus i, then the i

Fig. 2 Determination of line segments in an AOV. (a) IEEE 9-bus system. (b) Voltage magnitude curves at buses 1, 7, and 9.
Notably, the voltage magnitude curves of buses 2, 3, and 8 are higher than the specified threshold. Thus, a fault on line L2 does not result in a voltage sag at these three buses. In addition, the voltage magnitude curves for buses 4-6 are always below the threshold, and the entire line L2 is included in the AOV of buses 4-6. In other words, if a fault occurs on line L2, the voltages of buses 4-6 will decrease below the threshold. Consequently, the LMRA matrix for line L2 is represented by:
(5) |
Taking as an example, it indicates that a fault in s5 of line L2 results in a voltage sag at bus 9. By forming the LMRA matrix for all the lines, the LMRA matrix in the system is a 9×S matrix, where S represents the total number of segments across all lines.
Through the calculation of BMRA matrix, each bus has an AOV covered by installing a VSM at that bus. However, it cannot identify that at which bus in the AOV the fault occurs, and only the voltage sag is recorded. To determine the fault location using the BMRA matrix, the VSM placement must ensure that no identical columns exist in the fault location detection matrix (FLDM) based on the coverage area provided by each VSM. In other words, the VSM placement should result in as many distinct areas as the total number of system buses, where each bus must be located in its own separate area. In this case, it is possible to distinguish between faults occurring at different system buses.
For example, (6) presents the BMRA matrix for the SLGF in phase A in the IEEE 9-bus system. Considering the VSM locations at buses 2-4, the corresponding rows are selected from the BMRA matrix, resulting in the FLDM given in (7).
(6) |
(7) |
It can be observed that columns 1, 4, 5, and 6 in the FLDM are identical. Consequently, if a voltage sag is recorded by the VSM installed at bus 4, it indicates that a fault has occurred at one of these four buses. Therefore, by installing VSMs at buses 2-4, six unique columns are formed in the FLDM. Consequently, six distinct areas are obtained, and buses 1, 4, 5, and 6 are located in one of these areas. Thus, it is not possible to differentiate between faults at these buses, whereas fault locations at other buses can be recognized. For example, if the VSMs at buses 2 and 3 detect a voltage sag event and the VSM at bus 4 does not record a voltage sag, it signifies that the fault has occurred at bus 8. Therefore, to determine fault locations accurately, additional VSMs must be placed until no two identical columns are formed in the FLDM.
As the number of buses and lines in a power system increases, the evaluation of all the buses and lines for AOV determination becomes more time-consuming. To address this issue, a method to simplify this process is proposed. The computational burden is significantly reduced by excluding buses and lines outside the AOV.
The AOV determination through the conventional method, relying on the BVI and LVI for large systems, is quite time-consuming. Formulating the BVI vector requires the assessment of all system buses [
In the proposed method for AOV determination, the assessment commences from a specific bus with a sensitive load, for which the AOV must be determined. Initially, the lines connected to the sensitive bus are evaluated using the system incidence matrix. If the end buses of these lines are in the AOV, the AOV is expanded.

Fig. 3 Flow chart of proposed method for AOV determination of sensitive bus.
Firstly, the system sequence impedance matrix is calculated. The sensitive bus j and voltage sag threshold is considered to determine the AOV, and the lines connected to bus j are extracted from the incidence matrix. Then, the fault voltage equation is formulated for sensitive bus j and fault point K on the line l. The voltage magnitude is then calculated at the end bus of line l by substituting into the fault voltage equation ().
indicates that the end bus of line l is outside the AOV of bus j and line l has only one CP. The length of the line segment in the AOV of bus j is obtained by calculating the CP, and the process of AOV determination does not continue from the end bus.
indicates that the end bus of line l is in the AOV of bus j. Based on the fault voltage curve, the extreme point , which indicates the fault point on the line with the highest fault voltage, is determined using the golden section search (GSS). If , the line is completely in the AOV; otherwise, the line has two CPs and .
After the CPs of line l are computed, the end bus is added to set B, which denotes the expansion of the AOV. These steps are repeated for the buses added to set B and end until all the buses in B have been assessed. Finally, the AOV of bus j is determined. The buses and lines not evaluated previously are not in the AOV of bus j. Consequently, the sensitive bus does not experience voltage sag due to faults occurring on these lines and buses.
As an illustrative example, the proposed method for AOV determination is demonstrated for part of the hypothetical system shown in

Fig. 4 Illustration of proposed method for AOV determination.
In Stage 1, the fault voltage equations for lines 17-16 and 17-10 are formulated, and the voltages at the end buses 10 and 16 are calculated. The two buses are assumed to be located in the AOV of bus 17. By computing and assuming , we can conclude that these two lines are completely in the AOV of bus 17. Consequently, buses 10 and 16 are added to set B.
In Stage 2, the evaluations are performed for buses 10 and 16. Accordingly, the lines connected to buses 10 and 16 are considered, and the fault voltage equations for bus 17 are derived for the fault point on each line. For example, in the evaluation of line 16-12, the voltage magnitude at bus 17 is obtained for a fault at bus 12, and this value is assumed to exceed the defined threshold. In this case, the line has one CP, and the AOV does not expand to bus 12. For the lines connected to bus 10, the entire lines 10-22 and 10-6 are assumed to be in the AOV of bus 17. In addition, buses 9, 20, and 21 are outside the AOV, and thus the assessment does not continue for these buses. At the end of Stage 2, buses 6 and 22 are added to set B and the calculation continues for considered lines connected to these buses until the AOV for bus 17 is determined.
In previous studies, the secant [
If the root exists in the interval , the steps of the FPM for calculating the CPs are as follows [
Step 1: the lower and upper intervals are and , respectively, and the voltage sag threshold is selected.
Step 2: the fault voltage magnitudes and are calculated, and a value x is determined according to (8).
(8) |
Step 3: if the stop criterion is satisfied, then x is the CP, where is a small positive value as the convergence tolerance; otherwise, go to Step 4.
Step 4: if , the root is in the lower sub-interval, set , and return to Step 2; otherwise, go to Step 5.
Step 5: set and return to Step 2.
For lines with two CPs, the process involves calculating , after which Steps 1-5 are repeated for the intervals and to obtain the two CPs.
An additional advantage of the FPM is its independence from quadratic interpolation and approximated quadratic functions. Because p in the fault voltage equation is determined by substituting and , the order of fault voltage equations is inconsequential, and precise CPs are obtained using the main equation.
After the AOV for the system buses is determined and the BMRA matrix is formed, the VSM placement is performed to detect the fault locations in the system. The proposed method for VSM placement is presented in three steps.
The main objective of VSM placement is to detect the fault locations in the system using the minimum number of VSMs to cover all voltage sag events. Because the VSM placement is limited to system buses, the initial VSM placement is performed based on the BMRA matrix. For a system with n buses, the objective function is defined as:
(9) |
(10) |
(11) |
where is a binary variable representing the VSM placement at bus i, and indicates that the VSM is placed at bus i.
Constraint (10) ensures that each voltage sag event is recorded by at least one VSM. In other words, each column of the FLDM must contain at least one value of 1. In addition, to distinguish between the columns in the FLDM based on the BMRA matrix discussed in Section II, (11) must be satisfied to detect the fault locations at the system buses.
The optimal location of VSMs is determined by solving the mixed-integer linear programming problem defined in (9)-(11). Two situations may arise based on the BMRA matrix. The first situation occurs when no identical columns exist in the BMRA matrix. In this case, each bus has a unique effect on the others, and the fault locations can be determined across all the buses in the system. The second situation occurs when two or more columns are identical in the BMRA matrix. This implies that these buses have the same effect, and it is not possible to distinguish the fault locations at these buses. If there are n identical columns in the BMRA matrix, it will be possible to distinguish the fault locations based on the voltage magnitude by installing VSMs at these buses. In this case, the bus with the lowest voltage magnitude has a fault. Consequently, the combination is calculated for a set of identical columns. VSMs are then installed at the selected buses for each combination, and the placement of the remaining VSMs is determined by solving the optimization problem in (9)-(11).
For example, suppose that three columns {1,5,7} in the BMRA matrix are identical. In this case, VSMs are installed at two buses in the set {1,5,7} to detect the fault locations. Consequently, three VSM placement combinations are derived as: {1,5}, {1,7}, and {5,7}. For each combination, the variable is set to be 1 for each bus i, and the optimization problem is solved.
The BMRA matrix is formed for four types of short-circuit faults: SLGF, LLF, LLGF, and 3PF. For each type of fault, the optimal VSM placement is determined by solving the optimization problem given in (9)-(11). Notably, for LLF and LLGF, where voltage sag occurs in two phases, the VSM placement scheme is obtained separately for each fault phase. For example, in the case of the LLF between phases B and C, the BMRA matrices for phases B and C are calculated, and the VSM placement scheme is derived for the voltage sag in phase B (LLF_B) and phase C (LLF_C). Similarly, for the LLGF between phases B and C, two BMRA matrices are established. Consequently, there exist six BMRA matrices corresponding to different types of short-circuit faults (SLGF, LLF_B, LLF_C, LLGF_B, LLGF_C, and 3PF), and the optimization problem is then solved for each matrix.
The VSM placement schemes may not be unique for each type of short-circuit fault. In addition, because the measurement instrument may fail, it is essential to consider the effects of VSM failures on FLD. When a VSM fails, it can no longer record voltage sag events in the system. Thus, the chosen VSM placement scheme should exhibit the desired performance even in the event of a single VSM failure. Only the scenario with a single VSM failure is considered due to the low probability of simultaneous VSM failures. To determine the unique VSM placement scheme among different possible schemes for each type of short-circuit fault, the one exhibiting the highest FLD index is selected.
Based on the BMRA matrix, the bus FLD index is calculated for each VSM placement scheme. Various VSM placement schemes are obtained for each type of short-circuit fault. Then, a single VSM failure is then considered for each placement scheme based on the number of VSMs. Thus, the bus FLD index is determined for different types of short-circuit faults in each failure scenario.
Let us define SC as a set of different types of short-circuit faults, i.e., SC={SLGF, LLF_B, LLF_C, LLGF_B, LLGF_C, 3PF}. Accordingly, for the placement scheme i related with fault type and the VSM failure at bus n, the bus FLD index for the fault type is given as:
(12) |
where and represent the
For example, in the IEEE 9-bus system illustrated in
No. | Scheme | Bus No. with VSM failure | Bus FLD index (%) | Weighted bus FLD index (%) | Sum of weighted bus FLD index (%) | |||
---|---|---|---|---|---|---|---|---|
SLGF | LLF | LLGF | 3PF | |||||
1 | 1-2-3-7-8 | 1 | 88.89 | 61.11 | 72.22 | 66.67 | 84.17 | 396.67 |
2 | 88.89 | 77.78 | 66.67 | 55.56 | 85.00 | |||
3 | 66.67 | 66.67 | 77.78 | 66.67 | 67.22 | |||
7 | 77.78 | 77.78 | 61.11 | 55.56 | 75.83 | |||
8 | 88.89 | 72.22 | 66.67 | 55.56 | 84.44 | |||
2 | 1-2-3-7-9 | 1 | 88.89 | 72.22 | 77.78 | 66.67 | 85.56 | 427.78 |
2 | 100.00 | 77.78 | 77.78 | 55.56 | 94.44 | |||
3 | 100.00 | 77.78 | 88.89 | 66.67 | 95.56 | |||
7 | 66.67 | 83.33 | 66.67 | 55.56 | 67.78 | |||
9 | 88.89 | 72.22 | 66.67 | 55.56 | 84.44 | |||
3 | 1-2-3-8-9 | 1 | 88.89 | 61.11 | 72.22 | 66.67 | 84.17 | 397.78 |
2 | 88.89 | 66.67 | 66.67 | 55.56 | 83.89 | |||
3 | 88.89 | 77.78 | 77.78 | 66.67 | 86.11 | |||
8 | 66.67 | 83.33 | 66.67 | 55.56 | 67.78 | |||
9 | 77.78 | 77.78 | 61.11 | 55.56 | 75.83 |
In the bus FLD assessment, it is crucial to highlight that if m identical columns exist in the FLDM, the fault location at the buses, where the VSM is installed based on the voltage amplitude, will be detected. Considering the FLDM in (7) with , it is observed that columns 1, 4, 5, and 6 are identical. If the VSM is installed at bus 4, it is possible to recognize the fault location at this specific bus based on the voltage amplitude. However, the fault locations at buses 1, 5, and 6 remain undetected, resulting in . The bus FLD index is then computed as: .
After the bus FLD index is calculated for all fault types in each scenario, the weighted bus FLD index is calculated considering the occurrence probabilities of SLGF, LLF, LLGF, and 3PF as 80%, 10%, 5%, and 5%, respectively.
The sum of the weighted bus FLD index is then computed for each VSM placement scheme while considering the VSM failure scenario. Among the three VSM placement schemes related to LLGF_B presented in
To extend the evaluation of FLD for faults occurring on system lines, the process is performed in a manner similar to that for buses. Accordingly, the LMRA matrix is formed based on the line segments based on the AOV for each short-circuit type. Thus, similar to (12), the line FLD index is expressed as:
(13) |
where is the total number of line segments related to the
Thus, the line FLD index is calculated for the VSM failure at bus n and placement scheme i (scenario ) for different fault types in . A weighted line FLD index is then obtained for each placement scheme.
No. | Scheme | Sum of weighted bus FLD index (%) | Sum of weighted line FLD index (%) | Total sum of weighted bus FLD and line FLD indices (%) |
---|---|---|---|---|
1 | 1-2-3-7-8 | 396.67 | 97.31 | 493.98 |
2 | 1-2-3-7-9 | 427.78 | 118.57 | 546.35 |
3 | 1-2-3-8-9 | 397.78 | 97.63 | 495.41 |
As indicated in
In Step 2, a unique placement scheme is selected related to each fault type, resulting in a total of six unique placement schemes.
In Step 3, the bus FLD and line FLD indices are evaluated for each optimal placement scheme. The final optimal VSM placement scheme for the system is then determined based on the highest bus FLD and line FLD indices.

Fig. 5 Proposed method for optimal VSM placement.
The effectiveness of the proposed methods for AOV determination and VSM placement is assessed in IEEE stardand systems. In both systems, the phase winding connection of the transformers is assumed to be Yg/Yg, and the positive-, negative-, and zero-sequence impedances of generators are specified as j0.3, j0.2, and j0.05 , respectively. The zero-sequence impedances of lines are set to 2.5 times their positive-sequence impedances.
The effectiveness of the proposed method for AOV determination is assessed in the IEEE 118-bus system, which is a representation of large-scale system. The sensitive load is located at bus 70, and two assumed voltage sag thresholds of 0.7 and 0.8 p.u. are considered.

Fig. 6 AOV for bus 70 under an SLGF with voltage sag thresholds of 0.7 and 0.8 p.u..
As shown in
Method | Computational time (s) | |
---|---|---|
Threshold is 0.7 p.u. | Threshold is 0.8 p.u. | |
[ | 28.96 | 31.87 |
Proposed | 19.60 | 21.56 |
The method in [
The FPM for CP calculation is evaluated for bus 29 in the IEEE 30-bus system, and the results are compared with [

Fig. 7 Fault voltage curve of bus 29.
It is evident that the proposed method is more accurate in terms of CP calculation and has exactly detected the intersection point of the fault voltage curve and the defined voltage sag threshold.
Method | Voltage sag threshold (p.u.) | The first CP | The second CP | Total error (%) | ||||
---|---|---|---|---|---|---|---|---|
Voltage (p.u.) | Error (%) | Voltage (p.u.) | Error (%) | |||||
Secant method [ | 0.74 | 0.30700 | 0.740250 | 0.034 | 0.52400 | 0.7397 | 0.041 | 0.075 |
Bisection method [ | 0.30220 | 0.739800 | 0.027 | 0.52118 | 0.7399 | 0.014 | 0.041 | |
Proposed method | 0.30365 | 0.740008 | 0.001 | 0.52028 | 0.7400 | 0 | 0.001 |
The VSM placement steps are described in Section IV. The results of Step 3 for the selection of the final optimal VSM placement scheme are evaluated. The final optimal VSM placement scheme must achieve the highest bus FLD and line FLD indices among all possible ones. Because of the effects of voltage sag on customers’ equipment, it is crucial to assess the sensitivity of the equipment. Thus, the voltage tolerance curves are used to determine the voltage sag threshold, which is based on the preferences of the system operator. Because the VSM placement is proposed in the transmission system, the voltage sag threshold is set to be 0.7 p.u. while considering factors such as the fault clearance duration in the transmission system and the voltage tolerance curve.
To evaluate the proposed method for VSM placement, the implementation is carried out in the IEEE 9-bus system.
By assuming a zero-fault impedance ( ) and evaluating Steps 1 and 2, we obtain six unique placement schemes related to all fault types in the IEEE 9-bus system, as given in
No. | Scheme | Fault type | Fault location | FLD index (%) | Weighted FLD index (%) | Sum of weighted FLD index (%) | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
SLGF | LLF_B | LLF_C | LLGF_B | LLGF_C | 3PF | ||||||
1 | 2-3-4-7-9 | SLGF | Bus | 100.00 | 100.00 | 66.67 | 100.00 | 100.00 | 77.78 | 97.22 | 135.65 |
Line | 40.00 | 44.00 | 32.26 | 26.67 | 47.37 | 15.38 | 38.43 | ||||
2 | 3-4-6-7-9 | LLF_B | Bus | 100.00 | 100.00 | 77.78 | 55.56 | 77.78 | 66.67 | 95.56 | 128.67 |
Line | 34.29 | 48.00 | 35.48 | 13.33 | 31.58 | 7.69 | 33.11 | ||||
3 | 1-2-5-7-9 | LLF_C | Bus | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 77.78 | 98.89 | 130.36 |
Line | 31.43 | 40.00 | 29.03 | 26.67 | 42.11 | 23.08 | 31.47 | ||||
4 | 1-2-3-7-9 | LLGF_B | Bus | 100.00 | 100.00 | 77.78 | 100.00 | 100.00 | 66.67 | 97.22 | 129.49 |
Line | 31.43 | 44.00 | 25.81 | 46.67 | 52.63 | 23.08 | 32.27 | ||||
5 | 1-3-4-7-9 | LLGF_C | Bus | 100.00 | 100.00 | 77.78 | 77.78 | 100.00 | 66.67 | 96.67 | 134.94 |
Line | 40.00 | 44.00 | 29.03 | 26.67 | 47.39 | 15.38 | 38.27 | ||||
6 | 1-2-3-4-5-7-9 | 3PF | Bus | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 152.25 |
Line | 54.29 | 56.00 | 45.16 | 46.67 | 57.89 | 23.08 | 52.25 |
In addition, the placement scheme
The proposed method for VSM placement is also evaluated in the IEEE 30-bus system [
No. | Scheme | Fault type | Fault location | FLD index (%) | Weighted FLD index (%) | Sum of weighted FLD index (%) | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
SLGF | LLF_B | LLF_C | LLGF_B | LLGF_C | 3PF | ||||||
1 |
1-3-5-6-9-12-17-22- 24-25-29 | SLGF | Bus | 100.00 | 83.33 | 80.00 | 73.33 | 80.00 | 56.67 | 94.83 | 111.7 |
Line | 16.80 | 16.81 | 17.39 | 19.10 | 18.50 | 15.08 | 16.84 | ||||
2 |
3-5-9-14-15-17-20-22- 26-27-29 | LLF_B | Bus | 76.67 | 100.00 | 80.00 | 53.33 | 50.00 | 53.33 | 75.58 | 91.7 |
Line | 16.02 | 17.95 | 18.01 | 15.97 | 17.32 | 13.07 | 16.10 | ||||
3 |
2-6-9-11-12-19-21- 24-25-29 | LLF_C | Bus | 93.33 | 80.00 | 100.00 | 63.33 | 66.67 | 53.33 | 89.58 | 101.1 |
Line | 11.37 | 11.68 | 13.04 | 12.15 | 12.99 | 12.06 | 11.56 | ||||
4 |
2-5-6-8-9-11-12-13-16-19-21- 22-23-24-25-29 | LLGF_B | Bus | 100.00 | 86.67 | 100.00 | 100.00 | 93.33 | 83.33 | 98.33 | 125.4 |
Line | 27.30 | 26.21 | 26.71 | 27.43 | 27.17 | 22.61 | 27.05 | ||||
5 |
1-2-3-6-7-12-13-17-19-21- 23-24-26-29 | LLGF_C | Bus | 93.33 | 80.00 | 90.00 | 73.33 | 100.00 | 73.33 | 91.17 | 114.2 |
Line | 23.77 | 23.08 | 18.63 | 18.75 | 19.29 | 18.59 | 22.98 | ||||
6 |
1-2-4-6-7-9-11-12-13-16- 20-21-23-26-29 | 3PF | Bus | 93.33 | 83.33 | 93.33 | 73.33 | 73.33 | 100.00 | 92.17 | 115.6 |
Line | 24.03 | 22.22 | 18.94 | 20.49 | 20.87 | 22.11 | 23.42 |
The placement scheme related to LLGF_B exhibits the highest weighted FLD index for system buses and lines. However, this placement scheme requires 16 VSMs, which incurs higher costs. The scheme with the fewest VSMs is scheme
Consequently, choosing the optimal VSM placement scheme depends on the budget constraints and FLD for the system.
Given that faults in power systems may have impedances, simulations are conducted under different fault impedance in the IEEE 30-bus system. The increase in fault impedances leads to a smaller AOV of buses. To assess the effects of fault impedance on VSM placement, a unique placement scheme is obtained for each fault impedance under various faults types. The efficiency of each scheme is then evaluated under different fault impedances to identify the optimal placement scheme. The scheme with the highest bus FLD and line FLD indices is selected as the optimal placement scheme for the system, which demonstrates higher performance in FLD.
The optimal VSM placement schemes under various fault impedances in the IEEE 30-bus system are shown in
No. | Scheme | Zf (Ω) | Fault location | FLD index (%) | Weighted FLD index (%) | Sum of weighted bus FLD index (%) | Sum of weighted line FLD index (%) | |||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
SLGF | LLF_B | LLF_C | LLGF_B | LLGF_C | 3PF | |||||||
1 | 1-3-5-6-9-12-17-22-24-25-29 | 0 | Bus | 100.00 | 83.33 | 80.00 | 73.33 | 80.00 | 56.67 | 94.83 | 258.83 | 49.43 |
Line | 16.80 | 16.81 | 17.39 | 19.10 | 18.50 | 15.08 | 16.84 | |||||
5 | Bus | 90.00 | 73.33 | 83.33 | 73.33 | 80.00 | 76.67 | 87.50 | ||||
Line | 15.74 | 13.14 | 18.24 | 19.23 | 17.88 | 16.09 | 15.89 | |||||
10 | Bus | 76.67 | 50.00 | 90.00 | 76.67 | 76.67 | 86.67 | 76.50 | ||||
Line | 16.46 | 17.68 | 16.68 | 17.65 | 17.34 | 17.79 | 16.70 | |||||
2 | 2-7-9-14-20-22-24-25-28-29 | 0 | Bus | 70.00 | 80.00 | 86.67 | 53.33 | 53.33 | 50.00 | 69.50 | 234.25 | 43.24 |
Line | 14.47 | 14.81 | 13.04 | 11.81 | 11.42 | 9.55 | 14.03 | |||||
5 | Bus | 100.00 | 66.67 | 80.00 | 46.67 | 66.67 | 56.67 | 93.00 | ||||
Line | 16.05 | 12.04 | 13.21 | 11.89 | 10.93 | 8.81 | 15.11 | |||||
10 | Bus | 73.33 | 50.00 | 76.67 | 56.67 | 66.67 | 73.33 | 71.75 | ||||
Line | 14.40 | 13.11 | 12.11 | 11.46 | 10.84 | 14.23 | 14.10 | |||||
3 | 1-4-6-7-9-12-16-17-18-22-24-27-30 | 0 | Bus | 90.00 | 86.67 | 96.67 | 86.67 | 86.67 | 66.67 | 88.83 | 263.5 | 60.11 |
Line | 19.64 | 20.23 | 17.39 | 21.53 | 19.69 | 19.10 | 19.58 | |||||
5 | Bus | 90.00 | 86.67 | 90.00 | 86.67 | 76.67 | 76.67 | 88.75 | ||||
Line | 20.37 | 20.07 | 18.87 | 22.03 | 18.87 | 18.01 | 20.17 | |||||
10 | Bus | 86.67 | 46.67 | 96.67 | 90.00 | 86.67 | 100.00 | 85.92 | ||||
Line | 20.58 | 18.29 | 18.20 | 20.43 | 17.33 | 22.42 | 20.36 |
A comparison is conducted between the proposed method for VSM placement and other existing methods [
No. | Method | Scheme | Number of VSMs | Fault location | FLD index (%) | Weighted FLD index (%) | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
SLGF | LLF_B | LLF_C | LLGF_B | LLGF_C | 3PF | ||||||
1 |
[ | 1-3-5-7-18-24-28-30 | 8 | Bus | 43.33 | 46.67 | 33.33 | 43.33 | 40.00 | 26.67 | 42.08 |
Line | 8.79 | 10.26 | 7.14 | 9.03 | 7.87 | 6.53 | 8.65 | ||||
2 |
[ | 2-4-14-15-16-19-21-22-23-25-29 | 11 | Bus | 80.00 | 90.00 | 76.67 | 56.67 | 53.33 | 43.33 | 77.25 |
Line | 13.12 | 20.23 | 17.70 | 15.97 | 15.35 | 11.55 | 13.75 | ||||
3 |
[ | 2-5-6-10-17-24-26-27 | 8 | Bus | 70.00 | 50.00 | 56.67 | 43.33 | 46.67 | 30.00 | 65.08 |
Line | 11.11 | 11.96 | 11.49 | 10.76 | 11.42 | 7.54 | 10.99 | ||||
4 | Proposed | 1-3-5-6-9-12-17-22-24-25-29 | 11 | Bus | 100.00 | 83.33 | 80.00 | 73.33 | 80.00 | 56.67 | 94.83 |
Line | 16.80 | 16.81 | 17.39 | 19.10 | 18.50 | 15.08 | 16.84 |
This study presents a new method for determining AOVs and proposes an FPM for accurately calculating CPs, particularly for lines that are partially included in the AOV. Moreover, a VSM placement method is proposed to identify fault locations at the buses and lines in the system. These methods are implemented in the IEEE 9- and 30-bus systems. Simulation results reveal that the FPM improves the accuracy of the CP calculation as compared with other numerical methods, enabling a more precise AOV identification.
In addition, by accurately determining the AOVs and line segments based on voltage sag occurrence, the optimal location of VSMs can be determined to detect fault locations. The VSM placement scheme with the highest bus FLD and line FLD indices is selected as the optimal placement scheme for the system. The proposed method for VSM placement considers fault locations at both buses and lines to achieve a comprehensive placement scheme. Furthermore, when the placement schemes with various numbers of VSMs are obtained, the best placement plan can be chosen based on budget constraints and the FLD indices. A comparison with other placement methods demonstrates the efficiency of the proposed method in detecting fault locations.
References
A. Jalilian and S. Samadinasab, “Detection of short-term voltage disturbances and harmonics using μPMU-based variational mode extraction method,” IEEE Transactions on Instrumentation and Measurement, vol. 70, p. 9002917, Apr. 2021. [Baidu Scholar]
J. Y. Chan and J. V. Milanović, “Assessment of the economic value of voltage sag mitigation devices to sensitive industrial plants,” IEEE Transactions on Power Delivery, vol. 30, no. 6, pp. 2374-2382, Dec. 2015. [Baidu Scholar]
G. Wu, Q. Zhong, Q. He et al., “Evaluation method and probabilistic index of voltage sag severity considering point-on-wave,” Journal of Modern Power Systems and Clean Energy, vol. 9, no. 3, pp. 633-638, May 2021. [Baidu Scholar]
IEEE Recommended Practice for Monitoring Electric Power Quality, IEEE Std. 1159-2019, 2019. [Baidu Scholar]
C. H. Park and G. Jang, “Systematic method to identify an area of vulnerability to voltage sags,” IEEE Transactions on Power Delivery, vol. 32, no. 3, pp. 1583-1591, Jun. 2017. [Baidu Scholar]
M. H. J. Bollen, “Method of critical distances for stochastic assessment of voltage sags,” IEE Proceedings: Generation, Transmission and Distribution, vol. 145, no. 1, p. 70, Jul. 1998. [Baidu Scholar]
G. Carpinelli, P. Caramia, C. D. Perna et al., “Complete matrix formulation of fault-position method for voltage-dip characterisation,” IET Generation, Transmission & Distribution, vol. 1, no. 1, p. 56, Jan. 2007. [Baidu Scholar]
Y. Wang, H. Luo, and X. Xiao, “Voltage sag frequency kernel density estimation method considering protection characteristics and fault distribution,” Electric Power Systems Research, vol. 170, pp. 128-137, May 2019. [Baidu Scholar]
S. Shakeri, S. Esmaeili, and M. H. R. Koochi, “Determining accurate area of vulnerability for reliable voltage sag assessment considering wind turbine ride-through capability,” International Journal of Electrical Power & Energy Systems, vol. 119, p. 105875, Jul. 2020. [Baidu Scholar]
Y. Zhou, H. Wu, B. Lou et al., “Identification of the area of vulnerability to voltage sags based on Galerkin method,” Electric Power Components and Systems, vol. 47, no. 4-5, pp. 345-356, Jul. 2019. [Baidu Scholar]
C. H. Park and G. Jang, “Voltage quality assessment considering low voltage ride-through requirement for wind turbines,” IET Generation, Transmission & Distribution, vol. 10, no. 16, pp. 4205-4212, Dec. 2016. [Baidu Scholar]
N. Khatri, V. Kumar, R. C. Bansal et al., “Stochastic evaluation of voltage sag in power system network considering effect of photovoltaic generation,” International Transactions on Electrical Energy Systems, vol. 29, no. 4, p. 2773, Apr. 2019. [Baidu Scholar]
J. Hu, W. Hu, J. Chen et al., “Fault location and classification for distribution systems based on deep graph learning methods,” Journal of Modern Power Systems and Clean Energy, vol. 11, no. 1, pp. 35-51, Jan. 2023. [Baidu Scholar]
L. F. Ugarte, M. C. de Almeida, and L. H. T. Bandória, “Fault location approach to distribution networks based on custom state estimator,” Journal of Modern Power Systems and Clean Energy, vol. 11, no. 6, pp. 1878-1889, Nov. 2023. [Baidu Scholar]
F. B. Bottura and M. Oleskovicz, “Optimal allocation of power quality monitors considering short-duration voltage variations and parallel harmonic resonance conditions in power distribution systems,” International Journal of Electrical Power & Energy Systems, vol. 144, p. 108580, Jan. 2023. [Baidu Scholar]
S. Shakeri, M. H. R. Koochi, H. Ansari et al., “Optimal power quality monitor placement to ensure reliable monitoring of sensitive loads in the presence of voltage sags and harmonic resonances conditions,” Electric Power Systems Research, vol. 212, p. 108623, Nov. 2022. [Baidu Scholar]
P. E. T. Martins, W. G. Zvietcovich, T. A. de O. Silva et al., “Multi-objective approach for power quality monitor allocation with symmetry in short-duration voltage variations,” IEEE Transactions on Power Delivery, vol. 34, no. 2, pp. 430-437, Apr. 2019. [Baidu Scholar]
P. E. T. Martins and M. Oleskovicz, “Multi-objective optimization aiming to minimize the number of power quality monitors and multiple fault estimations in unbalanced power distribution systems,” IEEE Transactions on Power Delivery, vol. 37, no. 2, pp. 1315-1323, Apr. 2022. [Baidu Scholar]
H. Jalalat, S. Liasi, M. T. Bina et al., “Optimal placement of STATCOM using a reduced computational burden by minimum number of monitoring units based on area of vulnerability,” IET Generation, Transmission & Distribution, vol. 17, no. 10, pp. 2260-2271, May 2023. [Baidu Scholar]
G. Lv, C. Chu, Y. Zang et al., “Voltage sag source location estimation based on optimized configuration of monitoring points,” CPSS Transactions on Power Electronics and Applications, vol. 6, no. 3, pp. 242-250, Sept. 2021. [Baidu Scholar]
M.G. Tan, C. Zhang, and B. Chen, “Configuration of power quality monitor considering voltage sag location,” Electric Power Systems Research, vol. 223, p. 109681, Oct. 2023. [Baidu Scholar]
A. Saadat, R. A. Hooshmand, A. Kiyoumarsi et al., “Optimal location of voltage sag monitors in distribution networks with DGs using network zoning,” IEEE Transactions on Power Delivery, vol. 38, no. 6, pp. 4157-4165, Dec. 2023. [Baidu Scholar]
Y. Deng, X. Liu, R. Jia et al., “Sag source location and type recognition via attention-based independently recurrent neural network,” Journal of Modern Power Systems and Clean Energy, vol. 9, no. 5, pp. 1018-1031, Sept. 2021. [Baidu Scholar]
S. C. Chapra and R. P. Canale, Numerical Methods for Engineers. New York: McGraw-Hill, 2015. [Baidu Scholar]
IEEE. (2023, Dec.). IEEE 30-bus test case. [Online]. Available: http://www.ee.washington.edu/research/pstca/pf30/pg_tca30bus.htm [Baidu Scholar]