ieee journal on emerging and selected topics in …static var compensator (svc) and static...

IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS, VOL. 7, NO. 3, SEPTEMBER 2017 459

Power Swing Detection in UPFC-CompensatedLine by Phase Angle of Current

Jalal Khodaparast, Mojtaba Khederzadeh, Senior Member, IEEE, Filipe Faria da Silva, Member, IEEE,and Claus Leth Bak, Senior Member, IEEE

Abstract— Power swing blocker (PSB) is a complementarypart of distance relay protection, that detects power swing,in order to prevent unintended operation of a distance relay.Unified power flow controller (UPFC) is used in power system tocontrol both active and reactive powers and its operation duringpower swing influences the performance of PSB. In this paper,two traditional methods for detecting power swing are analyzedin UPFC-compensated condition to show how UPFC influencesthe performance of these methods. The indices proposed forthe above-mentioned methods are examined in compensatedcondition. The results show that these indices may no longerwork in systems with UPFC. In addition, this paper proposes anew method for detecting power swing based on the phase angleof current at relay point and compares it with two other methods.The new method distinguishes power swing from a fault throughthe trend of the phase angle of current. Finally, intensive studiesare performed and simulations demonstrate the merits of theproposed method.

Index Terms— Distance relay, power swing, PSB, UPFC.

I. INTRODUCTION

GENERALLY it is necessary to test the impact of FlexibleAC Transmission Systems (FACTS) on distance relays

in different conditions (steady state, fault, transient and powerswing). Since FACTS devices lead to variations of line currentand bus voltage, they influence the performance of relaysunder different conditions [1]–[3].

Unified Power Flow Controller (UPFC) is a FACTS deviceused to control the power flow of a transmission lineand to improve system stability [4]. However, the use ofFACTS devices may lead to distance relays being over-reaching or under-reaching during a fault [5]. Many publi-cations are devoted to evaluate the performance of distancerelays in lines compensated by FACTS. In [6], the impactof UPFC on distance relay is discussed, and an adaptivescheme using the neural network is proposed. Reference [7]presents a study of the performance of distance relay whenStatic VAR Compensator (SVC) and Static SynchronousCompensator (STATCOM) are used and it is shown that these

Manuscript received October 15, 2016; revised January 3, 2017 andFebruary 26, 2017; accepted April 17, 2017. Date of publication June 6, 2017;date of current version September 1, 2017. This paper was recommended byGuest Editor H. H.-C. Iu. (Corresponding author: Mojtaba Khederzadeh.)

J. Khodaparast and M. Khederzadeh are with the Electrical EngineeringDepartment, Shahid Beheshti University, A.C., Tehran 1658953571, Iran(e-mail: [email protected]; [email protected]).

F. F. da Silva and C. Leth Bak are with the Department of Energy Tech-nology, Aalborg University, 9220 Aalborg, Denmark (e-mail: [email protected];[email protected]).

Color versions of one or more of the figures in this paper are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JETCAS.2017.2705684

compensation devices cause mal-operation of the distancerelay by influencing the estimation of impedance. In [8],the impacts of different modes of Thyristor Controlled SeriesCompensation (TCSC) on distance relay are analyzed, whereasthe effect of Static Synchronous Series Compensator (SSSC)on the tripping characteristics of distance relays isexamined in [9].

Despite extensive research on the impact of FACTS deviceson the performance of distance relays during a fault, limitedwork exist on their impact during a power swing. For example,three methods for power swing detection, impedance decrease,swing center voltage (SCV) and power derivative are studiedin [10] for fixed capacitor compensated lines. However, thisreference does not analyze the impact of fixed capacitor math-ematically; it only gets results by simulation. Moreover, fixedcapacitor is a simple compensator device and it is necessaryto analyze the above methods for more sophisticated FACTSdevices. A method based on negative sequence component isproposed in [11] to discriminate the power swing from a faultin a fixed capacitor compensated line. However, this method isonly applicable in compensated lines, with the index value inuncompensated lines being negligible; therefore, the reliabilityof this method in uncompensated condition is low. In [12],impact of UPFC on swing characteristics (variations of radiusand center point) is studied, and it is shown that parametersof transmission line (ABCD) would change in the presenceof UPFC. Based on the definitions of these parameters andthe steady-state model of UPFC, new parameters (A’B’C’D’)are extracted. The reference also examines the impact ofUPFC on the rate of change of impedance by simulating acompensated power system during unstable power swing. Theresults showed that the time recorded by CPSB is reducedsignificantly because of the UPFC. These results motivated theauthors to research how UPFC affect different PSB methods.

The aim of this paper is to analyze the impact of UPFC ondifferent power swing blocking methods using both analyticaland simulation methodologies. The main contributions of thispaper are as follows:

• Comprehensive investigation of two common PSB meth-ods (Method 1, Method 2) in UPFC-compensatedline. Each method is mathematically analyzed inUPFC-compensated condition and it is demonstrated thathow each perform in this condition.

• A new index is proposed (Proposed method) based onthe phase angle of current for discriminating power swingfrom fault. It is shown that this proposed method is validin both uncompensated and UPFC-compensated lines.

2156-3357 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

460 IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS, VOL. 7, NO. 3, SEPTEMBER 2017

Fig. 1. Two-machine equivalent system with UPFC-compensated line.

The paper is organized as follows: Section II examines theperformance of two PSB methods in UPFC-compensated lines,and analytically shows UPFC interferences in the formulationof these methods. A novel index based on the use of thephase angle of current for discriminating fault from powerswing is presented in Section III. Finally, Section IV presentssimulation results to verify the analytical achievements.

II. IMPACT OF UPFC ON TWO DIFFERENT

POWER SWING BLOCKING METHODS

This section performs mathematical analysis of differ-ent methods for PSB and their index is reformulated inUPFC-compensated condition. In order to examine the UPFCoperation in a wide range of variations, it is assumed thatUPFC compensates all changes during power swing.

A. Method Based on Center of ImpedanceTrajectory (Method 1)

The main idea of this method is presented in [13] andit consists in using circular locus of impedance trajectorytogether with its center position to detect power swing. It isshown in [13] that the center of impedance trajectory seenby a distance relay will never come to Zone 1 (first zone ofdistance relay) during power swing, but it will come to thatzone during a fault.

1) Performance of Method 1 in Compensated Line:The index presented in [13] was not considered forUPFC-compensated lines, but as a result of UPFC operation,the impedance trajectory in a compensated line is differentfrom an uncompensated one. According to the topology of thecompensated line (shown in Fig.1), the active power at Bus 1associated with compensated condition can be written as:

PC O M1 = PC O M

2 + Pse + PLoss − Psh (1)

where, PC O M2 is the active power at Bus 2 in compensated

condition, Pse is the injected series active power by the seriesbranch of UPFC, Psh is the drawn shunt active power bythe shunt branch of UPFC, Ploss is active power loss in theconverters. Moreover, the reactive power at Bus 1 associatedwith compensated condition can be written as:

QC O M1 = QC O M

2 + Qse + QLoss (2)

where, QC O M2 is the reactive power at Bus 2 in compensated

condition, Qse is the injected series reactive power by seriesbranch and Qsh is the produced shunt reactive power by shuntbranch to maintain amplitude of V1 at its reference value

(this parameter is considered zero (Qsh = 0), because thispaper focuses on the power control property of UPFC). Basedon (1) and (2), the active and reactive powers at Bus 1 are:

PC O M1 = Pre f

2 , QC O M1 = Qre f

2 + Qse (3)

where, Pre f2 and Qre f

2 are the reference values of active andreactive powers for UPFC, respectively, and Psh = Pse+ Ploss .As the energy storing capacity of DC capacitor is small, theactive power drawn by the shunt converter should be equalto the active power generated by the series converter. Further-more, the shunt converter should support active power loss inconverters. Hence, according to Fig. 1 and (3), the impedanceseen during power swing is:

ZC O Mrelay = |V1|2

PC O M1 − j QC O M

1

= |V1|2Pre f

2 − j (Qref2 + Qse)

(4)

where, V1 = |V1|� δ1 is voltage at Bus1. In order to under-stand the center of impedance trajectory in this condition,the denominator of (4) is analysed first. Since Pre f

2 and Qre f2

are constant values and Qse is a time-varying quantity duringpower swing, the denominator is a vertical line in the complexplane R, X that crosses the horizontal axis at R = Pre f

2 .As this vertical line is in the denominator of (4), its reciprocalshould be examined.

The reciprocal of any complex point (R + j X) on verticalline is considered new complex point (RR + j X X) given bythe following relation:

RR = R

R2 + X2 , X X = −X

R2 + X2 (5)

Therefore, based on (5), the reciprocal of every point indenominator (4) is:

RR = Pre f2

(Pre f2 )

2 + (Qre f1 + Qse)

2 (6)

X X = −(Qre f1 + Qse)

(Pre f2 )

2 + (Qre f1 + Qse)

2

By summing the square of XX and RR, (7) is obtained:

RR2 + X X2 = 1

(pre f2 )

2 + X2(7)

By adding 1

4(pre f2 )

2 to both sides of (7) and after some

simplification procedures, the connection between RR and XXcan be formulated as:

(RR − 1

2.Pre f2

)2

+ (X X − 0)2 = (1

2.Pre f2

)2

(8)

Equation (8) represents a circle with center C1 = ( 12.Pre f

2

, 0)

and radius r1 = 12.Pre f

2

. Based on (4) and (8), the impedance

trajectory during power swing in a compensated line is a circlewith the center (CC O M ) and radius (rC O M ) as:

CC O M = k21

2.Pre f2

+ j0, rC O M = k21

2.Pre f2

(9)

KHODAPARAST et al.: POWER SWING DETECTION IN UPFC-COMPENSATED LINE BY PHASE ANGLE OF CURRENT 461

where, k1 = |V1|/|V3| and V3 = |V3|� 0 is voltage at Bus3.From (9), it can be concluded that UPFC changes the centerof impedance trajectory during the power swing. Althoughthe center of impedance trajectory during power swing inuncompensated condition has to be in limited region (out ofZone 1), there is no limitation for the center position incompensated condition and so the center can be anywherein the complex plain based on the power system (associatedwith k1) and UPFC states (associated with Pre f

2 ). Accordingly,the presented method in [13] for detecting power swing is notapplicable in the UPFC-compensated transmission lines.

B. Method Based on the Maximum Rate of Changeof Active and Reactive Powers (Method 2)

The main idea of this method is presented in [14]. In thismethod, the rate of change of power is utilized to detect powerswing. During power swing, the maximum value between therate of change of both active and reactive powers is higherthan a threshold value, decreasing to a value lower than thethreshold value during a fault.

1) Performance of Method 2 in Compensated Line:The index presented in [14] was not considered forUPFC-compensated lines. According to Fig.1, apparent powerat distance relay point is:

SC O M1 = PC O M

1 + j QC O M1 = V1 (Ise + Ish)∗ (10)

where, Ish is the shunt current drawn by shunt branch of UPFCand Ise is the series current in series branch of UPFC, whichcan be formulated as:

Ise = V2 − V3

j X Line= V2 + Vse − V3

j X Line(11)

where, Xline is the reactance of transmission line. By sep-arating active and reactive powers of (11), the active andreactive powers at the relay point in compensated conditionare (12) and (13) respectively.

PU N1 = |V1|.|V3|

X Linesinδ1

− |V1|.|Vse|X Line

sin(δ1 − ρ′) + |V1|.|Ish |cos(δ1 − θsh)

(12)

QU N1 = |V1|

X Line(|V1| − |V3|cosδ1)

+ |V1|.|Vse|X Line

cos(δ1 − ρ′) + |V1|.|Ish |sin(δ1 − θsh)

(13)

where, ρ′ is the phase angle of series voltage and θsh is thephase angle of shunt current. According to (12) and (13),active and reactive powers in compensated condition includethree parts: the first parts for both the active and reactivepowers) are similar to those in the uncompensated condition,but the second and third parts result from the operations ofseries and shunt branches of UPFC. Although it is possibleto conclude that this index value during power swing inuncompensated condition is inside a limited range (0.7 − 1);this conclusion cannot be drawn in compensated condition

because of the second and third items in (12) and (13).Therefore, the method presented in [14] for detecting powerswing is not applicable in UPFC-compensated transmissionlines.

C. Proposed PSB Based on Phase Angle of Current

A new method for PSB is proposed in this paper, whichis based on the phase angle of current and can be used inboth uncompensated and compensated conditions. In order topresent the proposed method, it is necessary to understand thevariation of the currents angle.

1) Angle of Current at Relay Point: The proposed methodconsiders the trend of the angle of current at the relay as anevent detection index. Therefore, in this section, the behaviorof this parameter is examined for power swing and faultconditions in uncompensated lines.

a) Power swing: PSB should detect power swing con-dition and send blocking command (BLOCK = 1) for dis-tance relay to prevent unintended operation. Suppose that theuncompensated system is in power swing condition, and thecurrent at relay point is given by (14).

I1 = |V1|� δ1(t) − |V3|� 0

Z Line= |V |

Z Line(1 � δ1(t) − 1)

= |V |Z Line

(cosδ1(t) − 1 + j sinδ1(t)) (14)

where, |V1| = |V3| = |V |. Based on (14), the angle of currentduring power swing is:

� I1 = Arctan(sinδ1(t)

cosδ1(t) − 1) − � Zline

= π

2+ δ1(t)

2− � Zline (15)

where, � Zline = Arctan(Xline/Rline), which is constantduring the power swing. Equation (15) is obtained based onsinx/(1 − cosx) = cot (x/2) and Arctancotx = (π/2) −Arccotcotx = (π/2)−x . It is worthy to note that power swingis actually the modulation of both amplitude and phase of thesignal, so considering variable parameter for k1 = |V1|/|V3|is more reasonable hypothesis. However, according to theproposed method, k1 being constant or variable does not affecton variation of phase angle of current. Hence, in order toprovide straightforward formulation for phase angle of currentduring power swing (equation (14) and (15)), k1 is consideredconstant value. In addition, this hypothesis is examined insimulation result section where the performance of proposedmethod is investigated in real power system (IEEE 39-Buspower system) which k1 is not constant. The simulation resultsshows that the proposed method operates accurately even invarying k1 condition.

According to (15), variation of the angle of the current (� I1)during power swing is as (16), which shows that it relatesdirectly to the variation of δ1. If �δ1 is positive, �( � I1) wouldbe positive and vice versa.

�( � I1) ∝ �δ1 (16)

When a power swing occurs in a transmission line,the impedance trajectory moves towards the relay zones due


to the positive trend of phase angle of current (�δ1 > 0).Therefore, we can conclude that the variation of the angle ofcurrent is positive during power swing when the impedancetrajectory goes towards the relay zones; the trend of currentsangle is negative (�δ1 < 0) during the time interval theimpedance trajectory goes backward, but this time interval isnot considered by the proposed detection process.

b) Fault Condition: PSB should detect fault and provideunblocking command (B L OC K = 0). Assuming that thepower system is in steady state condition and a fault happensat t = t f , the currents at the relay before (I S

1 ) and after (I F1 )

fault are:

I S1 = |V1|� δ1(t) − V3

Zline

I F1 = |V1|� δ1(t) − VF

m Zline(17)

where, VF is the voltage of fault point, and m Zline isimpedance from the relay point to fault point (0 < m < 1).According to (17), the angles of current before and after faultare (18) and (19) respectively.

� I S1 = Arctan(

|V1|sinδ1(t)

|V1|cosδ1(t) − |V3|) − � Zline (18)

� I F1 = Arctan(

|V1|sinδ1(t)

|V1|cosδ1(t) − |VF |) − � (m Zline) (19)

where, � Zline = � (m Zline) because the transmission lineis considered homogeneous. In order to analyze the trendof angle of current at fault initiation, it is necessary toanalyze (18) and (19) separately. In (18), |V1|cosδ1 < |V3|,and the argument of Arctan is a negative value and so,its Arctan is an angle bigger than 900. However, in (19),|V1|cosδ1 > |VF |, and the argument of Arctan is a positivevalue and so, its Arctan is an angle inferior to 900. Therefore,� I F

1 − � I S1 is negative and (20) is valid:

�( � I1) = �( � I F1 − � I S

1 ) (20)

According to (20), it can be concluded that when a faultoccurs, there is a downfall in the angle of current, whichdemonstrates the negative trend. In order to make the analysiseasier, it is assumed that the angle of V1 is not affected bythe fault significantly. Moreover, in a real power system, thereis transient period, which is solved by cumulative summationoperator.

2) Proposed Algorithm for Detecting Power Swing andFault: When a power swing occurs on a power system,the impedance trajectory moves toward the distance pro-tection zones. Based on this, it is necessary to program adetector (Flag1) able to detect the presence of impedancetrajectory inside the farthest protection zone (for exampleZone 3; third zone of distance relay). This detector helps toprevent false operation caused by other transients (e.g. loadchanging) far from the distance relay zones. This detector canbe programmed as:

D(n�t) = |Zrelay(n�t) − Zcenter |i f D(n�t) < r3 → Flag1(n�t) = 1 (21)

where, Zrelay is the impedance seen by the distance relay, n�tis the current sample time, Zcenter is the center of Zone 3,D is the difference between the impedance and the centerof Zone 3, r3 is radius of zone3 and Flag1 is the output ofthe first detector, which is 1 when the impedance measuredby the distance relay is inside Zone 3. Moreover, anotherdetector (Flag2) is necessary to detect that the impedancetrajectory is entering to Zone 3 or leaving it. This detectioncan be obtained by comparing two consecutive samples of D.It is worthy to mention that Flag2 is activated when Flag1is up. This detector can be programmed (22).

i f (Flag1(n�) = 1 & D(n�t) < D((n − 1)�t))

→ Flag2(n�t) = 1 (22)

Flag2 is the output of second detector, which is 1 whenthe direction of impedance trajectory is toward the insideof Zone 3. These two detectors are used for power swingdetection.

Based on the index proposed in this paper, power swingand fault can be detected by monitoring the variation of thephase angle of the relay current. This parameter is obtained bysubtracting the angles of current (θ1) for the current sampletime (n�t) and the previous sample time ((n − 1)�t), anduse of cumulative summation (C S) over one fundamentalcycle. Cumulative summation is used to increase the method’sreliability by removing instantaneous changes.

S(n�t) = θ1(n�t) − θ1((n − 1)�t)

C S(n�t) =n=r∑

n=r−N

S(n�t) (23)

where, N is sample number in one cycle and r is the lastsample inserted in the data window. Therefore, if C S > h p ,it means that the variation trend of the phase angle of thecurrent is positive, and it is negative if C S < h p (h p is thethreshold value, which is ideally zero; however, it is considereda small positive value to improve the reliability of the proposedmethod. The exact value of this parameter is mentioned in theSimulation Results section and it is defined based on differentsimulation cases during power swing). Therefore, power swingis detected according to (24).

Flag1 = 1 & Flag2 = 1 & C S > h p (24)

If the impedance trajectory is inside Zone 3 (Flag1 = 1),the direction of the impedance movement is towards the insideZone 3 (Flag2 = 1), and if the trend of the phase angleof current is positive (C S > h p), then a power swing ispresent and the proposed method issues blocking command(Block = 1). Moreover, the blocking command is kept at 1until the impedance trajectory is inside Zone 3 (Flag1 = 1).

The proposed method should also detect fault and unblockdistance relay. Based on the explanation given in the previoussection, when a fault happens the phase angle of currentdecreases. This decrease causes a negative trend of the phaseangle of the current at the fault inception. Therefore, any faultcan be detected when:

Flag1 = 1 & C S < hn (25)


Fig. 2. Flowchart of the proposed method.

It means than if the impedance trajectory is inside Zone 3(Flag1 = 1) and the trend of the phase angle of the currentis negative (C S < hn), a fault has occurred and the proposedmethod removes the blocking command of the distance relay.hn is the threshold value, which is ideally zero , but a smallnegative value is considered, in order to improve the reliabilityof the proposed method. The flowchart of this method is shownin Fig. 2.

III. SIMULATION RESULTS

Two simulations test scenarios are prepared: the first usesa single machine to infinite bus (SMIB), whereas the secondis based on the IEEE 39-Bus power system. Both detailedtime model and simple phasor model of UPFC are used,with the latter being used to illustrate the theoretical conceptmore clearly and the former in the validation of the results.Generally, six different modes of UPFC-compensated systemare considered in the simulation sections:

• Uncom: UPFC is out of service.• Mode 1: UPFC is in service, and its reference values

(active and reactive powers) are equal to the uncompen-sated case.

• Mode 2: Reference active power is increased, but not thereactive one.

• Mode 3: Reference active power is decreased, but not thereactive one.

• Mode 4: Reference reactive power is increased, but notthe active one.

• Mode 5: Reference reactive power is decreased, but notthe active one.

A. Simulation Results in SMIB (Phasor Model of UPFC)

The compensated two-machine equivalent system (Fig. 1)is considered in this section. The phasor models of the powersystem and UPFC (programmed sing MFILE-MATLAB) areused, in order to reduce the time required for simulation.The phasor model of UPFC is programmed by a time-varyingcontrolled series voltage source, which provides desired activeand reactive powers at Bus 2. Therefore, the amplitude and thephase of series voltage are (26) and (27), respectively.

|Vse(t)| = [(|V2|cosδ2 − |V1(t)|cosδ1(t))2

+ (|V2|sinδ2 − |V1(t)|sinδ1(t))2](1/2) (26)

ρ′(t) = tan−1(|V2|sinδ2 − |V1(t)|sinδ1(t)

|V2|cosδ2 − |V1(t)|cosδ1(t)) (27)

The shunt branch of UPFC is programmed by time-varyingcontrolled current source. Shunt current consists of two parts:reactive power compensation part (I 2

sh) to keep the amplitudeof bus voltage (|V1|) at its reference and series active powercompensation part (I 1

sh), which provides the active powerdemanded by series branch. The phasor data of the powersystem are V3 = 1 � 0, V1 = 1 � δ1, Z A = 0.1 � 900, Z B =0.1 � 900, and Z Line = 0.1 + j0.9. Amplitude values are inper unit, with Vbase = 500 kV and Sbase = 100 MV A. Thefundamental frequency is 60 H z and UPFC is located at theleft end of the transmission line. In order to produce the powerswing, the displacement angle of V1 is considered as:

δ1(t) = δ01 + k sin(2 π fslip t) (28)

where, k is the scaling factor, δ01 is the initial value of δ1,

and fslip is swing frequency. Performances of three mentionedmethods are analyzed next.

1) Performance of Method 1: This section analyzes theimpact of UPFC on the performance of Method 1, whichis based on monitoring the center of circular impedancetrajectory. Here, the center of impedance trajectory during thepower swing is always outside Zone 1. In order to examine thisindex, a compensated two-machine equivalent system (Fig. 1)is simulated. The UPFC is set to maintain active and reactivepowers at Bus2 equal to their reference values (Pre f =0.8118 p.u and Qre f = 0.2352 p.u, respectively). Powerswing occurs on the power system at t = 1s, which issimulated based on (28) where k = 1, fslip = 0.5 H z andδ0

1 = 450. First, two cases are simulated to show that theproposed index in [13] is not always true if an UPFC isinstalled in the transmission line. The first case simulates thepower system with k1 = |V1|/|V3| = 0.8, and the second casesimulates the power system with k1 = 0.5. The simulationresults of these cases are shown in Fig. 3.

Fig. 3(a) shows the impedance trajectories of the firstcase (k1 = 0.8) for both the uncompensated and compen-sated conditions and the respective centers. According tothis figure, the centers of both compensated and uncompen-sated conditions are outside Zone 1. Therefore, the proposedindex operates properly in this case. However, the secondcase (k1 = 0.5) indicates that the proposed index is notvalid for UPFC-compensated lines (Fig. 3(b)). Although thecenter of impedance trajectory in uncompensated condition is


Fig. 3. Impedance trajectory and its center during power swing foruncompensated and compensated conditions; (a) the conditions in which theindex is not contradicted, (b) the conditions in which the index is contradicted.

TABLE I

SIMULATION RESULT OF METHOD 1

outside Zone 1, the center of compensated trajectory entersZone 1. Eight others cases (different K1 and Pre f ) are alsoexamined and their results are tabulated in Table I. The centerof each case is tabulated, and

√means that the center is

inside Zone 1, whereas × means that the center is outsideZone 1 (radius of Zone 1 is 0.8 × |Z Line|). According to thistable, the centers of impedance trajectories for all 8 casesin uncompensated condition are outsides Zone 1. However,in compensated condition, the centers of trajectories come toZone 1 in 4 of the 8 cases. Therefore, it can be concluded thatpresence of an UPFC invalidates the use of this method fordetecting power swing.

2) Performance of Method 2: This section, analyzes theimpact of UPFC on the performance of Method 2, whichis based on the maximum rate of change of active andreactive powers. In uncompensated condition, the index valueduring power swing is larger than 0.7, but it decreases to

Fig. 4. Index of Method 2 for uncompensated and compensated conditions.

TABLE II

SIMULATION RESULT OF METHOD 2

zero during a fault. Therefore, by considering a thresholdvalue between 0.7 and zero, power swing is discriminatedfrom fault. However, based on the explanations previouslypresented, UPFC operation decreases the value of the proposedindex to lower than 0.7 making this index impracticablefor UPFC-compensated lines. The compensated two-machineequivalent system (Fig. 1) is simulated in different conditionsto examine the index values in different conditions (theresistance part of line impedance is ignored to make theanalysis easier).

The analyses is divided into two cases that consist in havingthe UPFC out of service, uncompensated, (Case1) and inservice (Case2). Power swing occurs on the power systemat t = 1s. The simulations is based on (28) where k = 1,fslip = 0.5 H z and δ0

1 = 450. The index values obtained inthese cases are shown in Fig. 4. The figure shows that the indexvalue during the steady state (0 < t < 1s) is zero for bothuncompensated and compensated conditions. However, whenthe power swing starts at t = 1s, the index value increasesand oscillates (1 < t < 2s). According to Fig. 4, the lowestvalue of the index in the uncompensated case is 0.7 p.u;however, it decreases to 0.27 p.u in compensated condition.This decrease is a result of compensating the variations ofactive power by UPFC. Moreover, in order to widen thesimulation cases, different swing frequencies, scale factorsand initial values of δ1 in (28) are investigated. Minimumvalues of the index in different cases during power swingare tabulated in Table II. As shown, the minimum value inuncompensated condition is always equal to 0.7 p.u. However,the UPFC reduces this minimum value to lower than 0.7.Since the threshold value of this method should be consideredlower than the minimum index during power swing, it can be


Fig. 5. Impedance trajectories during power swing.

Fig. 6. Performance of the proposed PSB during power swing.

concluded that the presence of UPFC invalidates this methodin compensated line.

3) Performance of the Proposed Method: To investigatethe merit of the proposed algorithm the compensated powersystem shown in Fig. 1 is used.

a) Power swing condition: A power swing occurs in thetwo-machine equivalent system (Fig. 1) at t = 1s basedon (28); where, k = 1, fslip = 0.5 H z and δ0

1 = 450. UPFCis set to keep active and reactive powers at Bus2 equal totheir reference values. Fig. 5 shows the impedance trajectoriesfor compensated and uncompensated conditions during powerswing, whereas Fig. 6 shows the performance of the proposedmethod.

According to Fig. 5, the impedance movements of bothconditions start at the same complex value Zrelay = 1.14 +j0.33 p.u. in plane R,X at t = 1s. The reason behind thisis that the reference values of active and reactive powersof UPFC are set initially equal to uncompensated condition.As a result of the UPFC operation, the impedance trajec-tory in compensated conditions is different from uncompen-sated conditions. This leads to different entrance times ofthe impedance trajectories into Zone 3 for uncompensatedand compensated conditions. Uncompensated trajectory entersZone 3 at t = 1.21s and compensated trajectory entersZone 3 at t = 1.29s. When the impedance trajectories comeinside Zone 3, the first detector sets Flag1 to 1 and triggersthe second detector to detect if the impedance trajectory isentering or leaving Zone 3. The output of the second detector

TABLE III

CS(p.u) VALUES DURING POWER SWING IN DIFFERENT CASES

Fig. 7. Impedance trajectories for fault during power swing.

is set to 1 (Flag2 = 1) if the impedance trajectory isentering Zone 3. According to Fig. 6, C S = +0.018 att = 1.21s for uncompensated condition and C S = +0.0165at t = 1.29s for compensated condition. It is worthy tonote that, according to Fig. 6, C S becomes negative at somepoints; however, the negative part does not correspond to theimpedance trajectory entering Zone 3, so these points arenot considered in the detection process. Therefore, since bothvalues of C S for compensated and uncompensated conditionsare positive (bigger than h p = 0.005) when the impedancetrajectories enter Zone 3, the proposed method detects thisevent as power swing and sets Block = 1. Moreover, it keepsBlock = 1 until the impedance trajectories are inside Zone 3.

Additionally, in order to widen the simulation, powerswings with different swing frequencies and scale factorsin (28) are examined. C S values of all the examined casesare tabulated in Table III. According to this table, in allconditions, the C S values are positive at the point where theimpedance trajectory enters Zone 3, showing that the proposedmethod can detect power swing in all conditions accurately.Additionally, this table shows that the proposed method doesnot lose efficiency in compensated condition.

b) Fault during power swing condition: PSB shoulddetect every fault and unblock distance relay (Block = 0)during power swing. This condition is examined in this section.Power swing condition and UPFC setting are the same asthe previous case, but with a symmetrical fault occurringduring power swing. A power swing occurs on the powersystem (Fig. 1) at t = 1s, which forces the impedance trajec-tories to move slowly toward the distance relay zones (Fig. 7).As explained earlier, power swing is detected by the proposedmethod based on the positive trend of currents angle whenthe impedance trajectory’s movement during power swing istoward Zone 3.


Fig. 8. Performance of the proposed PSB for fault during power swing.

TABLE IV

CS(p.u) VALUES AT FAULT INITIATION DURINGPOWER SWING IN DIFFERENT CASES

The block command is set to 1 at t = 1.21s for uncompen-sated and t = 1.29s for compensated condition during powerswing. However, a fault is initiated at t = 1.5s at 75% oftransmission line forcing the impedance trajectories to moverapidly toward the line impedance characteristic (Fig. 7). It isworthy to mention that, as a result of UPFC operation duringa fault, distance relay experiences over-reaching (Fig. 7),so that the fault, which is at 75% of transmission line, is seencloser by the distance relay. Fig. 8 shows the performanceof the proposed method: the fault causes a severe drop ofthe angle of current for both uncompensated and compensatedconditions (C S = −0.83 for uncompensated and C S = −0.77for compensated). Since the values of C S for compensatedand uncompensated conditions become negative, the proposedmethod detects this event as fault and removes block com-mand (Block = 0) at t = 1.53s for both conditions. Moreover,36 different cases (different modes of UPFC, different faultdistances (m%), and different δ1 in which fault occurs) aresimulated and the results are tabulated in Table IV. Accordingto the table, C S values are negative (smaller than hn = −0.05),showing that the proposed method can detect fault duringpower swing for both compensated and uncompensated lines.

B. Simulation Results of IEEE 39-Bus PowerSystem (Detailed Model of UPFC)

This section provides simulation results related to IEEE39-Bus (Fig. 9) simulated in PSCAD. The power system issimulated by using a sampling frequency 30.72 KHz andthen measured voltage and current are pre-filtered by low-passfilter for preventing aliasing phenomenon. In order to modeldistance relay and PSB, sampling rate of measured voltage

Fig. 9. IEEE 39-Bus power system.

and current are decreased by integer factor 16 and are sent toDFT (Discrete Fourier Transform) to estimate their phasors.According to the output of DFT, angle of current is obtained.The detailed model of UPFC consists of two major parts,the electrical part and the control part. In the electricalpart, there are four multi-level converters, four phase shiftingtransformers and a dc capacitor; whereas in the control part,there are measurements, PI controllers, calculation blocksand limiters. UPFC is located at the right end of protectedline (between Bus14 and Bus4 at Bus14 side). The naturalpower flow (without UPFC) at Bus14 is P2 = 9.5 p.u andQ2 = 2.55 p.u. (Vbase = 345 K V and Sbase = 100 MV A).Two different conditions are analysed: the first one deals withpower swing condition and the second one is related to faultcondition.

1) Power Swing Condition: In order to trigger this con-dition, a three-phase fault is simulated at t = 1s in theline between Bus19 (B19) and Bus33 (B33), and is clearedby the circuit breakers at both ends of the line. This eventputs the power system in power swing condition. Fig. 10 showsthe impedance trajectory, C S parameter and block commandduring this condition. The impedance trajectory enters Zone 3at t = 1.94s and the proposed method detects this event as apower swing based on the positive trend of the current angle,issuing a blocking command (Block = 1).

2) Fault Condition: This section analyzes the performanceof the proposed method for detecting a fault condition.Fig. 11 shows the simulation results for this case. Accordingto Fig. 11, a three-phase fault is detected by the negativevalue of C S at fault initiation setting the block commandto zero. Hence, the proposed method can detect fault during


Fig. 10. Simulation results of IEEE 39-Bus for pure power swing condition.

Fig. 11. Simulation results of IEEE 39-Bus for fault condition.

power swing accurately. In order to examine the efficiencyof the proposed method in different conditions (different faultresistances (R) and fault locations(m%)), different cases are

TABLE V

CS(KA) VALUES AT FAULT INITIATION IN IEEE 39-BUS POWER SYSTEM

examined, with the respective results tabulated in Table V.C S values are negative in both the compensated and uncom-pensated conditions, showing that the proposed method candetect fault during power swing for different fault locationsand fault resistances.

IV. CONCLUSION

The first section of this paper demonstrated that the powerswing blocker methods based on center of impedance tra-jectory and maximum rate of change of active and reactivepowers are not applicable in UPFC-compensated transmissionlines. Thus, it can also be concluded that the negative impactof UPFC on PSB so every power swing blocking methodshould be examined in compensated condition in order to bequalified. The second section of this paper proposes a newmethod for PSB, which is based on the phase angle of currentand it is applicable in both uncompensated and compensatedconditions. The trend of the angle of current to time refer-ence at relay point (local measurement) is considered as anevent detection index. According to the mathematical analysis,the variation of angle of current during power swing relatesdirectly to the variation of δ1. If �δ1 is positive, the variationof angle of current is also positive and vice versa. On theother hand, it is shown that when a fault occurs, a downfallin the angle of current is present demonstrating the negativetrend of this parameter. According to these results, a newmethod is proposed based on three conditional statements:if the impedance trajectory is inside Zone 3, the direction ofthe impedance movement is towards the inside Zone 3 andif the trend of the phase angle of current is positive, thenpower swing is present and the proposed PSB issues blockingcommand. It is worthy to note that the blocking command iskept until the impedance trajectory is inside Zone 3. Moreover,the proposed PSB can detect fault and unblock distance relay.If the impedance trajectory is inside Zone 3 and the trendof phase angle of current is negative, the proposed methoddetects that a fault has occurred and removes the blockingcommand of the distance relay. The proposed PSB is simulatedfor both single machine to infinite bus and IEEE 39-Buspower system, using both phasor and detailed model of UPFCduring power swing and fault. According to the simulationresults, the proposed PSB show acceptable performance inboth compensated and uncompensated conditions.

REFERENCES

[1] M. Alizadeh, N. Khodabakhshi-Javinani, G. B. Gharehpetian, andH. Askarian-Abyaneh, “Performance analysis of distance relay inpresence of unified interphase power controller and voltage-sourceconverters-based interphase power controller,” IET Generat., Transmiss.Distrib., vol. 9, no. 13, pp. 1642–1651, Oct. 2015.

[2] R. Dubey, S. R. Samantaray, and B. K. Panigrahi, “Adaptive distanceprotection scheme for shunt-facts compensated line connecting windfarm,” IET Generat., Transmiss. Distrib., vol. 10, no. 1, pp. 247–256,Jan. 2016.


[3] P. K. Nayak, A. K. Pradhan, and P. Bajpai, “Wide-area measurement-based backup protection for power network with series compensation,”IEEE Trans. Power Del., vol. 29, no. 4, pp. 1970–1977, Aug. 2014.

[4] N. G. Hingorani and L. Gyugyi, Understanding FACTS: Concepts andTechnology of Exible AC Transmission Systems. Hoboken, NJ, USA:Wiley, 2000.

[5] S. R. Samantaray, “A data-mining model for protection of FACTS-basedtransmission line,” IEEE Trans. Power Del., vol. 28, no. 2, pp. 612–618,Apr. 2013.

[6] P. K. Dash, A. K. Pradhan, G. Panda, and A. C. Liew, “Adaptive relaysetting for flexible AC transmission systems (FACTS),” IEEE Trans.Power Del., vol. 15, no. 1, pp. 38–43, Jan. 2000.

[7] T. S. Sidhu, R. K. Varma, P. K. Gangadharan, F. A. Albasri, andG. R. Ortiz, “Performance of distance relays on shunt-FACTS com-pensated transmission lines,” IEEE Trans. Power Del., vol. 20, no. 3,pp. 1837–1845, Jul. 2005.

[8] M. Khederzadeh and T. S. Sidhu, “Impact of TCSC on the protection oftransmission lines,” IEEE Trans. Power Del., vol. 21, no. 1, pp. 80–87,Jan. 2006.

[9] A. Kazemi, S. Jamali, and H. Shateri, “Effects of SSSC on distancerelay tripping characteristic,” in Proc. IEEE Int. Power Energy Conf.,Nov. 2006, pp. 624–629.

[10] A. Esmaeilian, A. Ghaderi, M. Tasdighi, and A. Rouhani, “Evaluationand performance comparison of power swing detection algorithms inpresence of series compensation on transmission lines,” in Proc. 10thInt. Conf. Environ. Elect. Eng. (EEEIC), May 2011, pp. 1–4.

[11] P. K. Nayak, A. K. Pradhan, and P. Bajpai, “A fault detection techniquefor the series-compensated line during power swing,” IEEE Trans. PowerDel., vol. 28, no. 2, pp. 714–722, Apr. 2013.

[12] Z. Moravej, M. Pazoki, and M. Khederzadeh, “Impact of UPFC onpower swing characteristic and distance relay behavior,” IEEE Trans.Power Del., vol. 29, no. 1, pp. 261–268, Feb. 2014.

[13] R. Jafari, N. Moaddabi, M. Eskandari-Nasab, G. B. Gharehpetian, andM. S. Naderi, “A novel power swing detection scheme independent ofthe rate of change of power system parameters,” IEEE Trans. PowerDel., vol. 29, no. 3, pp. 1192–1202, Jun. 2014.

[14] X. Lin, Y. Gao, and P. Liu, “A novel scheme to identify symmetricalfaults occurring during power swings,” IEEE Trans. Power Del., vol. 23,no. 1, pp. 73–78, Jan. 2008.

Jalal Khodaparast was born in Iran in 1987.He received the B.Sc. and M.Sc. degrees in electricalengineering from the Shahrood University of Tech-nology, Shahrood, Iran, in 2008 and 2010, respec-tively. He is currently pursuing the Ph.D. degreewith the Electrical Engineering Department, ShahidBeheshti University, Tehran, Iran. He was a GuestPh.D. Student (visiting period) with the AalborgUniversity, Denmark, from 2015 to 2016. Hisresearch interests include wide area protection, flex-ible ac transmission systems, power system dynam-

ics, digital signal processing, and phasor estimation.

Mojtaba Khederzadeh (SM’06) received theB.Sc. degree in electrical engineering from the SharifUniversity of Technology, Tehran, Iran, in 1980, theM.Sc. degree in electrical engineering from TehranUniversity, Tehran, in 1990, and the Ph.D. degree inelectrical engineering from the Sharif University ofTechnology in 1996. He was a Post-Doctoral Fellowwith the University of Western Ontario, London,ON, Canada, from 2004 to 2005. He is currently anAssociate Professor and the Director of the PowerSystem Protection and Control Research Center,

Electrical Engineering Department, Shahid Beheshti University, Tehran. Hisresearch interests include power system protection, control and monitoring,and power system dynamics.

Filipe Faria da Silva (M’09) was born in Portugalin 1985. He received the M.Sc. degree in electricaland computers engineer from the Instituto SuperiorTécnico, Portugal, in 2008, and the Ph.D. degreein electric power systems from Aalborg University,Denmark, in 2011. He was with EDP-Labelec in2008 and with Energinet.dk from 2008 to 2011.He is currently an Associate Professor with theDepartment of Energy Technology, Aalborg Uni-versity, where he is also a Semester Coordinatorof the Electrical Power System and High Voltage

Engineering master program and the Vice-Leader of the Modern PowerTransmission Systems research program. His research focuses on powercables, electromagnetic transients, system modeling, network stability, HVDCtransmission, and HV phenomena. He has authored over 70 articles and a bookin these areas. He serves as the Active Member of CIGRE, being currentlythe Head of Denmark’s IEEE–PES and a regular member of the Cigré SCC4.

Claus Leth Bak (M’99–SM’07) was born in Århus,Denmark, in 1965. He received the B.Sc. degree(Hons.) in electrical power engineering in 1992,the M.Sc. degree in electrical power engineeringwith the Department of Energy Technology, AalborgUniversity (AAU), in 1994, and Ph.D. degree in2015. His Ph.D. thesis was EHV/HV undergroundcables in the transmission system. He was a Profes-sional Engineer with Electric Power Transmissionand Substations, NV Net Transmission Company,with specializations within the area of Power System

Protection. In 1999, he was an Assistant Professor with ET-AAU. He currentlyholds a full professor position with ET-AAU. He serves as the Head of theEnergy Technology Ph.D. Program (over 100 Ph.D.’s) and the Head of theSection of Electric Power Systems and High Voltage with AAU and is amember of the Ph.D. Board with the Faculty of Engineering and Science.He has supervised/co-supervised over 35 Ph.D.’s and over 50 M.Sc. theses.He has authored or co-authored app. 240 publications. His main researchareas include corona phenomena on overhead lines, power system modelingand transient simulations, underground cable transmission, power systemharmonics, power system protection, and HVDC-VSC offshore transmissionnetworks. He is a member of Cigré JWG C4-B4.38, Cigré SC C4, and SCB5 study committees and Danish Cigré National Committee. He received theDPSP 2014 best paper award and the PEDG 2016 best paper award.

ieee journal on emerging and selected topics in …static var compensator (svc) and static...

Documents