Sunteți pe pagina 1din 8

Electric Power Systems Research 78 (2008) 18061813

Contents lists available at ScienceDirect

Electric Power Systems Research


journal homepage: www.elsevier.com/locate/epsr

A current-based model of the static synchronous series compensator (SSSC) for NewtonRaphson power ow
Anton Vinkovic, Rafael Mihalic
Faculty of Electrical Engineering, University of Ljubljana, Trzaska 25, 1000 Ljubljana, Slovenia

a r t i c l e

i n f o

a b s t r a c t
In this paper, a new approach to modeling a static synchronous series compensator (SSSC) for power-ow calculations by applying the NewtonRaphson method is presented. This new approach differs from known methods in terms of the interpretation of the devices branch. It is considered on the basis of its current and is therefore denoted as a current-based model of an SSSC. This approach might in principle be applicable also for other FACTS devices (i.e., UPFC, IPFC, GUPFC). In the paper, the current-based model of an SSSC is presented as the models of this device have difculties with convergence in power-ow calculations and there are very few references covering these topics. First, the basic features of an SSSC are presented, as it is the basis for the current-based model that is incorporated into the NewtonRaphson load-ow model. The results of the tests at the IEEE 57-bus system are discussed in detail and compared with the existing injection SSSC load-ow model [X.P. Zhang, Advanced modeling of the multicontrol functional static synchronous series compensator (SSSC) in Newton power ow, IEEE Trans. Power Syst. 18 (November (4)) 2003]. 2008 Elsevier B.V. All rights reserved.

Article history: Received 21 September 2007 Accepted 9 March 2008 Available online 9 May 2008 Keywords: Flexible ac transmission systems Power system analysis Power ow SSSC

1. Introduction In the past few years, the literature has given considerable attention to FACTS devices because in the competitive environment of power generation and transmission power-ow control has become an attractive alternative as a result of the developments in power electronics devices. However, the rst step when studying new possibilities in the system should be the power-ow calculation, and in order to be able to carry out this type of analysis power-ow models for FACTS devices have to be available. For typical FACTS devices, i.e., SSSC [6], STATCOM [5,7], IPFC [5,12,13], UPFC [5,8,9,10,11,17], and GUPFC [5,12], such models have already been constructed for applications in Newton power ow. The existing power-ow models for FACTS devices are mostly incorporated into node equations describing the power system via the injected voltage and the devices impedance. Such models are subsequently referred to in the text as injection models. The problem of SSSC modeling in Newton power ow is described in ref. [6]. SSSC is, rst of all, meant for dynamic, active power-ow control in power networks. Its impact can be assumed to be the consequence of the control of the voltageangle difference between the devices terminals. The angle in question is controlled by varying

Corresponding author. Tel.: +386 1 4768 438. E-mail address: rafael.mihalic@fe.uni-lj.si (R. Mihalic). 0378-7796/$ see front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2008.03.006

the SSSCs series voltage injection. Therefore, our early attempts at SSSC modeling for power-ow purposes (some 11 years ago) were based on the principle of the inclusion of a controllable series voltage source in the form of a network element (some programs offer such a possibility). Its voltage magnitude was a controllable parameter, while its angle was set so as to be perpendicular to the sources throughput current. The angle correction was performed with one iteration delay. It has been proven that such approaches are inappropriate as convergence effectiveness was more or less coincidental; moreover, in some SSSC operating regions convergence in general could not be achieved (e.g., overcompensated region). As such the approach has only limited applicability, and investigations in this direction have been abandoned. For the model in ref. [6], our calculations have shown that it may have difculties with convergence in some regions. Therefore, a new approach to SSSC power-ow modeling was implemented. The so-called current-based model was developed. However, this approach should not be mistaken for the one described in refs. [15,16] in which other current-based power-ow formulations, also suitable for FACTS controllers, have been proposed. The basic idea of our approach is to represent the presence of the device by the current in the series branch and the terminal voltages instead of the injected voltage. The presented current-based SSSC power-ow model seems to be effective and might also be used for the models of some other FACTS devices.

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

1807

2. Operating characteristics of the SSSC In order to be able to understand SSSC modeling and its impact on power ow, its basic characteristics should be described. An SSSC is a FACTS device that impacts the power systems operating point by injecting a voltage U T into the line [1]. Basically, it consists of a series transformer, an ac/dc converter and a dc circuit (capacitor). The device is schematically presented in Fig. 1, U S and U R being the devices terminal voltage phasors (S, sending; R, receiving). Let us, for readers convenience, at this point neglect the devices losses (active and reactive). The ac/dc converter generates the voltage U T [2,3], which is (neglecting device losses) perpendicular to the series branch current, -. The relations between the voltage and I current phasors are presented in Fig. 2. According to the phasor diagram, the SSSCs impact on the power ow can be considered the same as the impact of classical series compensation. However, if the devices limitations are taken into consideration, its impact differs considerably from classical series compensation [2,4].

Fig. 3. Basic SSSC model.

In order to imagine the SSSCs impact on power ow, let us calculate the devices equivalent reactance (the ratio of UT and I) for a simple SSSC lossless model, included into a longitudinal system, which is presented in Fig. 3. Let the systems equivalent series reactance be XL = 1 P.U., and the voltage magnitudes be US = 1.0 P.U. and UL = 0.8 P.U. The equivalent SSSC reactance XSSSC as a function of the angle difference = S L (the angle between U S and U L ) for two injected-voltage magnitudes UT (0.4 P.U., 0.4 P.U.) is presented in Fig. 4. It is evident that for UT = 0.4 P.U. for all transmission angles , the SSSC exhibits a capacitive character. For UT = 0.4 P.U., in the range between LC and LC it exhibits a capacitive character of high capacitance (the overcompensation area), and outside this transmission angle area it has an inductive character [3]. LC is some kind of singular point at which XSSSC increases toward an innite value and changes its character between inductive and capacitive. The impact of the SSSC on the active power, P, transmission characteristics is presented in Fig. 5 for various magnitudes of UT . The power ow changes depending on the SSSCs operating point. The SSSCs operating range can be divided into three areas (P denotes active power ow with the SSSC, P0 denotes active power ow without the SSSC): P > P0 ; 0 < < 180 and P < P0 ; 180 < < 0 capacitive character; P < P0 ; LC < < 180 and P > P0 ; 180 < < LC inductive character; P < 0; 0 < < LC and P > 0; LC < < 0capacitive character (overcompensated area). As already mentioned, an SSSC is an effective tool for active power-ow control. However, this does not mean that it cannot

Fig. 1. General scheme of an SSSC.

Fig. 2. An example of the phasor diagram of a lossless SSSC.

Fig. 4. Equivalent SSSC reactance.

1808

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

QST = Im[U S (I ) ] = US (Re[I ] sin(S ) Im[I ] cos(S )) S RT = PRT + jQRT = U R (I )

(4) (5) (6) (7) (8)

S RT = UR (cos(R ) + jsin(R )) ((Re[I ] jIm[I ])) PRT = Re[U S - ] = UR (Re[I ] cos(R ) + Im[I ] sin(R )) - I QRT = Im[U S - ] = UR (Re[I ] sin(R ) Im[I ] cos(R )) - I 4. Current-based model of an SSSC in NewtonRaphson power ow

The incorporation of the current-based SSSC model into the power system is presented schematically in Fig. 7. The SSSC is connected between nodes D and L (D as device, L as line). The SSSCs impedance, representing the SSSCs active and reactive losses, is represented by ZT between nodes D and S. The system of equations describing the line model is supplemented by the current-based SSSC model. This system is represented by Eqs. (9)(16).
2 PD = UD GT UD US (GT cos(D S ) + BT sin(D S )) 2 QD = UD BT UD US (GT sin(D S ) BT cos(D S )) 2 PS = US GT US UD (GT cos(S D ) + BT sin(S D )) PST 2 QS = US BT US UD (GT sin(S D ) BT cos(S D )) QST

(9) (10) (11) (12)

2 PR = UR

GL + GLP 2

UR UL (GL cos(R L ) (13)

Fig. 5. Transmission characteristics for various SSSC injected voltages UT .

+ BV sin(R L )) PRT BL + BLP 2

enable the control of other quantities, e.g., reactive power-ow control or voltage control. However, in this situation it is not as effective as a shunt device (e.g., SVC, STATCOM). 3. Current-based model of an SSSC The system of equations for load ow that models a power system consists of equations that describe the equilibrium of active and reactive power in the power systems nodes. In accordance with this, the current-based SSSC load-ow model is modeled using the products of the terminal voltages (U S , U R ) and the throughput cur- rent, -, which represent the SSSCs active and reactive powers in the I SSSCs terminal nodes (S, R). The SSSCs scheme and its powers are presented in Fig. 6. The SSSC current, -, is presented in the form of the real and imagiI nary components, i.e., Re[I ] and Im[I ], respectively. The system from Fig. 6 (the SSSC between the nodes S and R) can be described using the following equations: S ST = PST + jQST = U S (I ) S ST = US (cos(S ) + jsin(S )) ((Re[I ] jIm[I ])) PST = Re[U S (I ) ] = US (Re[I ] cos(S ) + Im[I ] sin(S )) (1) (2) (3)

2 QS = UR

UR UL (GL sin(R L ) (14)

BL cos(R L )) QRT GL + GLP 2 BL + BLP 2

2 PL = UL 2 QL = UL

UL UR (GL cos(L R ) + BL sin(L R )) (15) UL UR (GL sin(L R ) BL cos(L R )) (16)

By incorporating the SSSC into the power line, the following equations are added to the above system of equations: The equation for the reference quantity REF, which is inuenced by the SSSC and is expressed using other quantities. The equation describing a lossless SSSC, which describes the situation when there is no active power generation (nor absorption), i.e., PT = 0, which is expressed as PST + PRT = 0. It should be noted that the SSSCs losses are described by ZT , see Fig. 7. Using the SSSC only one quantity can be controlled independently, as it has only one controllable parameter. The current model was tested with the control variables that were already discussed in the references (e.g., [5,6,7,13]): line active power ows (PST , PRT ): REF = PST = US (Re[I ] cos(S ) + Im[I ] sin(S )) REF = PRT = UR (Re[I ] cos(R ) + Im[I ] sin(R )) reactive power ows (QST , QRT ): (17) (18)

Fig. 6. Current-based SSSC model.

REF = QST = US (Re[I ] sin(S ) Im[I ] cos(S )) -

(19)

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

1809

Fig. 7. Current-based SSSC model incorporated into a power system.

REF = QRT = UR (Re[I ] sin(R ) Im[I ] cos(R )) injected-voltage magnitude (UT ): REF = UT =
2 2 US + UR 2US UR cos(S R )

(20)

[A] = [B] =

(21)

SSSC equivalent reactance (XSSSC ): REF = XSSSC = Re[I ](US sin(S ) UR sin(R )) Re[I ]2 + Im[I ]2 (22)

Im[I ](US cos(S ) + UR cos(R )) Re[I ]2 + Im[I ]2 -

PD D PL D PS D PR D QD D QL D QS D QR D PD Re[I ] PL Re[I ] PS Re[I ] PR Re[I ] QD Re[I ] QL Re[I ] QS Re[I ] QR Re[I ] -

PD L PL L PS L PR L QD L QL L QS L QR L

PD S PL S PS S PR S QD S QL S QS S QR S

As a controllable variable, XSSSC has no practical value; however, it can be used for test purposes the terminal voltage magnitudes (UD , UR ): REF = UD UD REF = 0 REF = UR UR REF = 0 (23) (24)

The presented equations are non-linear and are, in accordance with the NewtonRaphson method, developed into a Taylor series. Due to the inclusion of the SSSC, additional derivatives for buses S and R, as well as: derivatives of the real component of the SSSC current Re[I ], derivatives of the imaginary component of the SSSC current Im[I ] appear. Consequently, the dimension of the Jacobian matrix increases by 6, and it is referred to as an extended Jacobian matrix. It is represented by Eqs. (25)(29), where Pi and Qi represent the sums of the active and reactive powers in the power-system nodes, respectively. In the right-hand mismatch vector, REF denotes the referenceparameter mismatch.

PD Im[I ] PL Im[I ] PS Im[I ] PR Im[I ] QD Im[I ] QL Im[I ] QS Im[I ] QR Im[I ] -

PD R PL R PS R PR R QD R QL R QS R QR R

PD UD PL UD PS UD PR UD QD UD QL UD QS UD QR UD

PD UL PL UL PS UL PR UL QD UL QL UL QS UL QR UL

PD US PL US PS US PR US QD US QL US QS US QR US

PD UR PL UR PS UR PR UR QD UR QL UR QS UR QR UR

(26)

(27)

REF
[C] = D PT D

REF REF REF REF REF REF REF L S R UD UL US UR PT PT PT PT PT PT PT L S R UD UL US UR


(28)

[A] [B] [C] [D]

D L S R UD = UL US UR Re[I ] Im[I ] -

PD PL PS PR QD QL QS QR REF PT

REF [D] = Re[I ]


PT Re[I ] -

REF Im[I ] PT Im[I ] -


(29)

(25)

A successful convergence of the NewtonRaphson method depends on the initial conditions. For power systems without FACTS devices, it is normally possible to use the bus-voltage magnitudes 1 P.U. and the bus-voltage angles 0 as the starting values (at start). If the SSSC is included the initial values for the real and imaginary components of the SSSCs current (Re[I ] and Im[I ], respectively) have to be set. This was done in the following way.

1810

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

In a power system the bus-voltage magnitudes are about 1 P.U. The angle between the SSSC terminal voltages (U S in U R ) in most cases should not exceed 15 (an SSSCs injected-voltage magnitude is about 0.3 P.U.). The SSSCs current, -, is perpendicular to the I injected voltage, U T . This situation is represented by the phasor diagram in Fig. 8. As can be seen in Fig. 8, the real component of the SSSCs current, Re[I ], is considerably larger than the imaginary component, Im[I ]. Therefore, the proposed initial values for the SSSCs current real and imaginary components are Re[I ] = 0.1 P.U. and Im[I ] = 0, respectively. These values were conrmed in the appropriate numerical tests. In the tests, the initial value was varied toward 0 and the changes in the iteration success were marginal. However, the initial point for Re[I ] must not be 0, as the Jacobian becomes singular. For handling constraints, the same procedure may be taken as proposed in refs. [6] and [14], i.e., in case one of the quantities (e.g., the magnitude of the injected voltage UT or current I) violates its limit the original control objective is released according to a different control mode, while the quantity whose limit is violated is kept at its limit. The tests have shown that this procedure has a minor impact on the number of iterations needed to achieve a satisfactory accuracy. However, one has to take into consideration that the variable is changed and the appliance of a new control variable might change the speed of the iteration. 5. Numerical examples The SSSC current-based model was tested on the small ve-bus (same as in ref. [5]) system, IEEE 57-, 118- and 300-bus system. The testing procedure consisted of inserting the SSSC into various locations and changing the required line active power ow in all the three SSSC operating regions (capacitive, inductive and overcompensated). In all tests in all three regions, convergence was achieved and no problems with the convergence occurred. Although, the power ows were changed considerably with the SSSC devices whose ratings were in some cases unrealistically high (UT > 0.5 P.U., sometimes even UT > 0.9 P.U.), normally only ve to six iterations were enough to achieve mismatch levels below 1012 . In the paper for presentation and demonstration purposes, the IEEE 57 test system, the scheme of which is presented in Fig. 9, was considered. The lines and nodes at which, in the following considerations, the SSSC is assumed to be included are marked. In the following text, the results are presented for an SSSC included in power lines between nodes 15 and 14 at the location of node 15. The SSSCs impedance is represented by ZT = 0.1 P.U. The calculations were carried out by applying the classical NewtonRaphson method. The quantities are in per unit, the base power being SB = 100 MVA. The model was tested for all six control variables described in Section 4 (i.e., the SSSCs receiving end active power, PRT ; SSSCs receiving end reactive power, QRT ; SSSCs injected voltage, UT ; the SSSCs equivalent reactance, XSSSC ; the SSSCs input terminal voltage, UD ; and the receiving end voltage, UR ). In all cases, the model exhibited very good convergence. The number of iterations needed to achieve the prescribed accuracy varied between 5 and 6. The iteration process was somewhat worse only when control variable

Fig. 9. IEEE 57 system.

were node-voltage magnitudes UD and UR , respectively. The reason for this might be the bad coupling between the series compensation and the bus-voltage magnitude (normally, for voltage control, shunt devices like SVC or STATCOM are used). Let us analyze the two typical examples. 5.1. The control variable is the active power PRT In the IEEE 57 system, not considering an SSSC, the power ow between the nodes 15 and 16 amounts to P0 = 69.72 MW. For the control variable in the three examples, the following values were used: PRT1 = 70.00 MW; PRT2 = 35.00 MW; PRT3 = 140.00 MW. These values for PRT have been chosen in order to cover all the three SSSC operating areas, i.e., the overcompensated area, the inductive area and the capacitive area (see Section 2). The rst value (PRT1 ) is negative (100% P0 in the reverse direction at a positive transmission angle) and is therefore situated in the overcompensated area. The second value (PRT2 ) represents the inductive area, as the SSSC extends the electrical length of the line. The power ow is reduced to 50% P0 . The third value (PRT3 ) is in the capacitive area, the electrical length of the line is reduced and the power ow is increased to 200% P0 . The mismatch levels of the NewtonRaphson power ow for the IEEE 57 system without the SSSC (P0 ) as well as when considering all three values in question for the control variable (PRT1 , PRT2 , PRT3 ) are presented in Fig. 10. In the gure, max| P, Q| [P.U.] represents a mismatch as a function of the number of iterations, N. The iteration procedure is stopped when the mismatch drops below 1012 . 5.2. The control variable is the SSSCs injected-voltage magnitude UT In the next step, the magnitude of the SSSCs injected voltage was chosen as a control variable. It is the logical choice, as it repre-

Fig. 8. SSSC phasor diagram.

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

1811

Fig. 10. Mismatch levels for various reference powers PRT .

Fig. 11. Mismatch levels for various reference-voltage magnitudes UT .

sents one of the devices limitations. When the limitations need to be considered in the power-ow calculations then the control variable has to be changed as the limit is exceeded. If, e.g., the SSSCs maximum injected voltage amounts to UTmax , but a higher UT is required, the control variable is no longer PRT , but UT (like, e.g., the generators representation changes from the PV node to the PQ node, if the generators reactive power limitation is exceeded). Let us have a look at the convergence course for the following values of the control variable UT : UT1 = 0.10 P.U. UT2 = 0.20 P.U. UT3 = 0.30 P.U. The mismatch levels for the system without an SSSC as well as for the three reference values for UT is presented in Fig. 11. From Figs. 10 and 11, it can be concluded that the convergences of the system with the SSSC included are fast and do not differ signicantly from the case without an SSSC. However, this fact does not tell us anything about how effective the proposed model is compared to other existing models. Therefore, in the subsequent text a comparison between the proposed and the injection models is carried out. 6. Comparison between the current-based and injection SSSC models Zhang in ref. [6] presents an SSSC model that may, according to the SSSCs interpretation, be classied as an injection model. For the readers convenience, let us briey summarize the idea. The SSSC from Fig. 12 is characterized by its impedance, ZT , and the injected voltage, UT , and is described by the following equa-

tions:
2 PD = UD GT UD UR (GT cos(D R ) + BT sin(D R ))

UD UT (GT cos(D T ) + BT sin(D T ))

(30)

2 QD = UD BT UD UR (GT sin(D R ) BT cos(D R ))

UD UT (GT sin(D T ) BT cos(D T ))

(31)

2 PRT = UR GT UR UD (GT cos(R D ) + BT sin(R D ))

+ UR UT (GT cos(R T ) + BT sin(R T ))

(32)

2 QRT = UR BT UR UD (GT sin(R D ) BT cos(R D ))

+ UR UT (GT sin(R T ) BT cos(R T ))

(33)

These equations, together with the other system equations and the equations for the control variable X and the generated

Fig. 12. Injection SSSC model.

1812

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

Fig. 13. Number of iterations, N, needed for calculation at various values of PRT SSSC between nodes 1 and 16.

Fig. 14. Number of iterations, N, needed for the calculation for various values of PRT SSSC between nodes 2 and 3.

6.2. An SSSC between two generator PV nodes SSSCs active power (i.e., PT = 0) are developed into a Taylor series. The two variables are included in an extended Jacobian matrix, which in the proposed current-based model are not, i.e., UT (SSSC injected-voltage magnitude) and T (SSSC injected-voltage angle). The equations developed into a Taylor series build a matrix form, with the extended Jacobian matrix as presented in Eq. (34). In the second case, an SSSC is inserted into the line between the PV nodes 2 and 3 at node 2. The value of the control variable PRT is varied between 400 and +120 MW, with a step of 20 MW. The results presents Fig. 14. In this case, in the majority of the areas examined, the necessary number of iterations, N, is higher for the injection model than for the current-based model, but not by very much. However, the area of negative powers at which the model converges is considerably narrower for the injection model (300 MW compared to 400 MW) than for the current-based model. 6.3. An SSSC between PQ nodes far from generators If an SSSC is connected between PQ nodes 24 and 25 at node 24, an extreme situation is achieved. The control variable is varied between 4 and 28 MW with a step of 1 MW. The results are presented in Fig. 15. In this case, the injection model almost fails. The number of iterations needed is considerably higher than if the current-based model is used. But the most problematic factor of all is that the area of the control variables for which the method converges successfully is more than two times narrower than for the current-based model, i.e., 717 MW for the injection model compared to 428 MW for the current-based model. 6.4. Conclusion regarding the comparison between the current-based model and the injection SSSC model 6.1. An SSSC between a slack bus and a load PQ bus First, the SSSC is inserted between the slack node 1 and the load PQ bus 16 (see the system in Fig. 9). The value of the control variable PRT is varied between 100 and +160 MW, with steps of 10 MW. The results presents Fig. 13. For this case, it is typical that the number of iterations the injection model needs is very variable and considerably higher than the current-based model needs. The range of control-variable values at which the model still converges is, for the current-based model, 10 MW broader on both sides. The calculations showed that the proposed current-based model represents a good alternative to the existing SSSC models. In all our tests on the IEEE 57-, 118- and 300-bus system (always using the same initial conditions) it exhibits fast convergence. Comparable tests on the IEEE 57 system showed that the convergence of the current-based model was considerably faster (on average, twice as fast) than the existing injection model. This by itself does not prove that in all possible conditions the proposed model exhibits better convergence, but in all extensive tests the convergence of the proposed model was very fast.

Pi i Qi i REF i PT i

Pi Ui Qi Ui REF Ui PT Ui

Pi T Qi T REF T PT T

Pi UT Qi UT REF UT PT UT

i Pi U i Qi = REF T PT UT

(34)

In order to evaluate our model, we compared it with the existing one [6]. Therefore, calculations were performed in various locations and with various control variables in the IEEE 57 model. In the following text, some results for controlled power, PRT , are presented. The results are given in the form of diagrams that present the number of iterations needed to achieve a power mismatch max| P, Q| [P.U.] of less than 1012 . The number of iterations, N, for the injection model is marked with , and for the current-based model with . From the numerous calculations, the three typical cases were selected.

A. Vinkovic, R. Mihalic / Electric Power Systems Research 78 (2008) 18061813

1813

the derivatives by the real Re[I ] and imaginary Im[I ] SSSC current components. As a controlled variable, we can take the SSSCs active or reactive power, the injected-voltage magnitude, the bus voltages or the SSSCs equivalent reactance. The model tests were carried out on the IEEE 57 model for various locations, control variables and their values. The current-based model was compared to the injection model presented in ref. [6]. The numerous calculations showed that the current-based SSSC model represents a step forward compared to the injection SSSC model. In our tests, it exhibits faster convergence, the number of iterations required is almost constant over a wide area of reference variables and the operating areas in which convergence is achieved is wider than when the injection model is used. References
[1] L. Gyugyi, C.D. Schauder, K.K. Sen, Static synchronous series compensator: a solid-state approach to the series compensation of transmission lines, IEEE Trans. Power Deliv. 12 (January (1)) (1997). [2] R. Mihalic, I. PapiC, Mathematical models and simulation of a static synchronous series compensator, in: International Conference on Electric Power Engineering [also] PowerTech 99, Budapest, August 29September 2, 1999. [3] R. Mihalic, I. Papic, Static synchronous series compensatora mean for dynamic power ow control in electric power systems, Elect. Power Syst. Res. 45 (1998) 6572. [4] R. Mihalic, Power ow control with controllable reactive series elements, IEE Proc. Gen. Transm. Distribut. 145 (1998) 493498. [5] E. Acha, C.R. Fuerte-Esquivel, H. Amirez-Perez, C. Angeles-Camacho, FACTSModelling and Simulation in Power Networks, John Wiley & Sons, Chichester, 2004. [6] X.P. Zhang, Advanced modeling of the multicontrol functional static synchronous series compensator (SSSC) in Newton power ow, IEEE Trans. Power Syst. 18 (November (4)) (2003). [7] X.P. Zhang, E. Handschin, M. Yao, Multi-control functional static synchronous compensator (STATCOM) in power system steady-state operations, Elect. Power Syst. Res. 72 (July) (2004) 269278. [8] S.H. Lee, C.C. Chu, D.H. Chang, Comprehensive UPFC models for power ow calculations in practical power systems, in: Power Engineering Society Summer Meeting 2001 IEEE, vol. 1, 1519 July, 2001, pp. 2132. [9] S.H. Lee, C.C. Chu, Power ow models of unied power ow controllers in various operation modes, in: Transmission and Distribution Conference and Exposition 2003 IEEE PES, vol. 1, 712 September, 2003, pp. 157162. [10] X.P. Zhang, K.R. Godfrey, Advanced unied power ow controller model for power system steady state control, in: Electric Utility Deregulation, Restructuring and Power Technologies 2004 IEEE, vol. 1, 58 April, 2004, pp. 228 233. [11] X. Wei, J.H. Chow, B. Ferdanesh, A.A. Edris, A common framework of voltagesourced converters for load ow, sensitivity and dispatch analysis, Sensitivity Dispatch Anal., IEEE Trans. Power Syst. 19 (2004) 934941. [12] X.P. Zhang, Modelling of the interline power ow controller and the generalised unied power ow controller in Newton power ow, IEE Proc. Generat. Transm. Distribut. 150 (May (3)) (2003) 268274. [13] X. Wei, J.H. Chow, B. Ferdanesh, A.A. Edris, A dispatch strategy for an interline power ow controller operating at rated capacity, in: Power Systems Conference and Exposition, 2004 IEEE PES, vol. 3, 1013 October, 2004, pp. 14591466. [14] Y. Zhang, Y. Zhang, A novel power injection model of embedded SSSC with multi-control modes for power ow analysis inclusive of practical constraints, Elect. Power Syst. Res. 76 (March (5)) (2006) 374381. [15] V.M. Da Costa, J.L.R. Pereira, N. Martins, An augmented NewtonRaphson power ow formulation based on current injections, Int. J. Elect. Power Energy Syst. 23 (May (4)) (2001) 305312. [16] A.M. Variz, V.M. da Costa, J.L.R. Pereira, N. Martins, Improved representation of control adjustments into the NewtonRaphson power ow, Int. J. Elect. Power Energy Syst. 25 (September (7)) (2003) 501513. [17] C.R. Fuerte-Esquivel, E. Acha, The unied power ow controller: a critical comparison of NewtonRaphson UPFC algorithms in power ow studies, IEE Proc. Generat. Transm. Distribut. 144 (September (5)) (1997) 457464.

Fig. 15. Number of iterations, N, needed for the calculation for various values of PRT SSSC between nodes 24 and 25.

The advantage of the current-based model in this sense depends on the SSSCs location and the control variable value. However, during numerous tests on the IEEE 57 system, there was not even a single case when the current-based model converged more slowly than the injection model. The next interesting feature of the current-based model is that in the majority of the areas where convergence can be achieved the number of iterations needed is almost constant (this is valid also for tests on the IEEE 118 and 300 system), while for the injection model the number changes considerably (see Fig. 13). Considering the area of the control variable values where convergence can be achieved, comparable tests on the IEEE 57 system showed that the current-based model allows a wider set of values than the injection model. How much wider the area in question is depends on the location, but in our tests there was not even a single case when the current-based models area was smaller than that for the injection model. 7. Conclusions In the paper, a new approach to the modeling of an SSSC for NewtonRaphson power ow is presented. This approach might also be applicable for some other FACTS devices. The motivation for developing such a model is the fact that there have only been a few solutions proposed so far. The current-based SSSC model uses the devices current (the real component. Re[I ], and the imaginary component, Im[I ]) for its representation instead of the injected-voltage magnitude and angle, as used in the injection model. The system of equations describing a power system is supplemented by equations of nodes S and R, between which the SSSC is connected, by equations for the controlled value REF, which should be inuenced by the SSSC and the equation that describes the SSSC feature, i.e., PT = 0. The Jacobian matrix is extended with

S-ar putea să vă placă și