Paper accepted for presentation at 2003 IEEE Bologna Power Tech Conference, June 23th-26th, Bologna, Italy

Power system stability enhancement using PSS and UPFC Lyapunov-based controllers: A comparative study
S. Robak, M. Januszewski, D.D. Rasolomampionona
research centers paid a particular attention on magnitude and phase regulated device Unified Power Flow Controller (UPFC). A typical application of UPFC consists on switching to one or two parallel power system transmission lines in order to get a higher loading. Additionally the use of an appropriate control algorithm in fast dynamics allows an efficient damping of power system oscillations [1], [3]. This paper analyses and compares the control features of a PSS and UPFC from the point of view of damping of power swings. At first a theoretical analysis which shows the designing method is performed. Then the optimal control strategies resulting from Lyapunov theory for non-linear power system model are presented. Theoretical analysis is supported by the results of simulations tests and eigenvalues analysis which are obtained for the case of multi-machine system model. II. PSS AND UPFC INFLUENCE ON POWER SYSTEM STABILITY A. Effect of PSS The generator automatic voltage regulator which reacts only to the voltage error, weakens the damping introduced by damper windings. This detrimental effect of the AVR can be compensated using supplementary control loop which is the power system stabiliser [1]. Generally the main task of PSS is to add an additional signal into the excitation control loop in order to improve the damping of power swings. From Fig. 1 results that PSS provides weak damping action and helps to return the rotor quickly to the equilibrium point [1].
PE ( )
2

Abstract--Recently it has been shown that it is possible to design a Power System Stabilizer (PSS) as well as an Unified Power Flow Controller (UPFC) controller using a non-linear multi-machine system model in conjunction with Lyapunovs direct method. In this paper the features of two controllers implementing the proposed control laws have been compared: the PSS constitutes a supplementary loop to the main Automatic Voltage Regulator (AVR) and UPFC controller. Theoretical analysis is supported by the results of eigenvalues analysis and simulations tests, which highlighted the differences between the action of compared controllers. The tests performed on a multimachine system model have shown a good performance of both proposed controllers. Index Terms-- flexible AC transmission systems, power system stability, synchronous generator excitation

NHANCEMENT of electromechanical power swing damping is one of the main tasks which the power system control system should fulfill. During power swing two oscillations types appear simultaneously: local and inter-area, making the control target difficult to reach [1]. Because of the low cost of a single device, it is easy to damp oscillations using Power System Stabilizers (PSS), which is an additional loop in the automatic voltage regulation system of the generator. The attractiveness of the PSS has resulted an increasing in PSS research interest and in a wider use of this device by industry. Simultaneously the problem of PSS appropriate tuning have appeared, especially in conditions of power system deregulation. It is because there is a limited access to the indispensable data for optimization. This problem has been solved for a non-linear system using the Lyapunov method [2]. Another efficient method of enhancement of electromechanical power swing damping is the FACTS application. Among different types of FACTS devices
The research reported in this paper has been supported in part by State Committee for Scientific Research under Grant 4 T10B 027 24. S. Robak is with Institute of Power Engineering, Warsaw University of Technology, (e-mail: sylwester.robak@ien.pw.edu.pl). M. Januszewski is with Institute of Power Engineering, Warsaw University of Technology, (e-mail: Marcin.Januszewski@ien.pw.edu.pl). D.D Rasolomampionona is with Institute of Power Engineering, Warsaw University of Technology, (e-mail: desire.rasolomampionona@ien.pw.edu.pl).

I. INTRODUCTION

1 8 57 6 3 4

Pm

Fig. 1. Power oscillations with PSS damping included.

B. Effect of UPFC The main task of UPFC device is the control of active and reactive power during steady state. High speed and flexibility

0-7803-7967-5/03/$17.00 2003 IEEE

of switching processes allow the use of UPFC during postdisturbance fast dynamics [1], [3], [4]. The different schemes of generator-infinite busbar system with UPFC is shown on Fig 2.
a) g a b s

III. THEORETICAL BACKGROUND Any disturbance in the system involves a power imbalance that moves the system trajectory from the pre-fault stable equilibrium point to a transient point that has a higher energy & x level than the post-fault equilibrium point . If V is negative then Lyapunov function V(x) decreases with time and tends x towards its minimum value, the system equilibrium point . & is, the faster the system The more negative the value of V returns to the equilibrium point , i.e. the better the damping x is. Consequently any given control is optimal (in the & Lyapunov sense) if it maximises the negative value of V at each instant of the transient state. IV. THE PSS CONTROL

b) Ig g Eg Xa a Ua

Ub

b Ib Br

Xb

s Is Us

c)

Ua

UP

Ub UQ
Fig. 2. Generator-infinite busbar system with UPFC: a) schematic diagram; b) single-phase diagram; c) phasor diagram.

From the point of view of influencing the steady-state and the transient state, the thyristor controlled UPFC can be treated as an element of a very small time constant, or approximately as a proportional element [5]. Hence the series part of the UPFC (booster) can be modelled (Fig. 2b) using the ideal complex transformation ratio , while the shunt part of the UPFC can be modelled (Fig. 2b) as a controlled shunt susceptance Br . The ratio resolved into two orthogonal components: direct - corresponding to the voltage increment U Q and quadrature - corresponding to voltage increment U P . Using the previously defined components the ratio can be defined as follows: = (1 + ) + j (1) Using the model presented on Fig.2 the influence of different UPFC voltage components on the power characteristic is shown on Fig. 3.
a) P Pm

A. Control law The approach outlined above has been applied to design a PSS using initially a fourth-order non-linear single-machineinfinite-busbar system model and then third-order non-linear multi-machine system [2]. In the latter case the Lyapunov function was expressed as the sum of the system kinetic energy, potential energy, and a term proportional to the sum of squared deviation of the transient emf for all the machines. It was further proved in [2] that the derivative of the Lyapunov function can be expressed as
n n 1 1 & V = Di i2 ' E qi E qi i =1 i =1 Td 0i X di X di

& + VEf

(2)

where
n 1 1 & VEf = ' E fi E fi E qi E qi i =1 Td 0i X di X di

)(

(3)

In these equations the subscript i relates to the generator number, i is the speed deviation, Xdi and X di are the synchronous and transient d-axis generator reactance, respectively, Eqi is the q-axis synchronous emf, E qi is the q axis transient emf, Td' 0i the open-circuit transient time constant, Efi the excitation voltage. A hat on the top of a symbol corresponds to the post-fault equilibrium point. The first and the second components in (2) are negative semi-definite and always contribute to the overall system damping. The third component is given by (3) and it is this component that is influenced by the excitation control. Thus the optimal control law must be such that at any instant in the transient state each component of (3) must be negative and maximal. This condition is satisfied by the control law E E =K E E (4)

>0 =0 <0

b) P Pm

fi

fi

qi

qi

+ Fig. 3. Influence of UPFC control on power characteristics: a) direct control influence, b) quadrature control influence.

The regulation of quadrature component depending on the UPFC voltage sign makes the power characteristic move right or left. The direct component changing results in magnitude change of active power angle characteristic in a small range of load angle change.

where Ki is the controller gain. Obviously this law is optimal only for the Lyapunov function used. The control law defined by equation (4) can be re-written as (5) E fi (t ) = E q ref i Ki X adi i fi (t )

where E q ref i = X adi i f

ref i

is synchronous emf (Xad is the dref i

axis armature reaction reactance), i fi (t ) = i fi (t ) i f

is

the increment in the field current obtained using a wash-out element which eliminates the DC component from the field current. B. Important features The first important feature of the control law (4) is that the & function V , equation (2), is independent of the network & parameters. Recall that V effectively determines the system damping. This would suggest that the contribution of each generator, controlled using (4), to the overall system damping is insensitive to changes in the network topology/parameters or the pattern of network flows. This is very important as it suggests that the proposed stabilizer is robust in that it does not need re-tuning following network changes. The second important feature of control law (4) is that & each generator contributes an independent component into V given by (2), with no cross-coupling terms between generators. In other words each individual controller contributes a damping term independent of all the other controllers. This very important feature of the proposed controller is referred to as the additivity of damping. C. Block diagram of excitation controller The control strategy (5) can be executed using two different structures hierarchical and with supplementary loop, respectively [2]. Detailed discussion above mentioned structures including choosing of the parameters of the controllers can be find [6]. This paper presents the structure wiht supplementary loop in which Transient Stability Excitation Control (TSEC) block is not necessary. Fig. 4 shows the functional block diagram whit a more traditional AVR+PSS structure of a controller executing strategy (5). Symbol ST denotes a static exciter. The lower part of this diagram, denoted as PSS.
V
+

dE & dE V = P + K = D 2 dt dt (1 X SHC Br ) b sin + (1 X SHC Br )b cos + Br b X SHC sin

a 2

(6)

where X SHC = X a X b

(X

+ Xb

) and b

is the amplitude

of the power-angle characteristic without the shunt branch. In order for each of the inputs , , Br to contribute to the fast returning of the system to the equilibrium point, the control strategy should ensure that all the components in (6) are negative independently of the sign of , . This can be achieved if:

(t ) = + K [b sin ]

Br (t ) = K B [b sin ]

(t ) = K [b cos ]

(7) (8) (9)

where K , K , K B are positive coefficients. The control strategy given by (7), (8) and (9) using local measurements can be replaced, with good accuracy by the following strategies [3], [4]:

(t ) +

K dQb dQb = + K ( P) K X dt dt

(10) (11) (12)

dPb dt dQb K dQb Br (t ) B = K B( P ) K X dt dt

(t ) + K

Vref

1 + sTA 1 + sTB

KA AVR

+ -

VE

ST

Ef

if

PSS E q Tr s 1 + T1s 1 + T3 s i f X ad K 1 + Tr s 1 + T2 s 1 + T4 s
Washout with correction

B. Important features These strategies are easily executed in practice as they are based on locally measurable signals of real and reactive power. The quantitative analysis studied in [4] has been shown that strategies (10), (11) and (12) emulate reasonably well the strategies (7), (8) and (9). However it should be emphasised that damping achieved using the locally available signals is weaker that the damping achieved using the state variables. This is due to the fact that the gains K , K , K B used in the local strategies have to be constrained because of the influence of inputs , , Br on the changes in Pb , Qb . For high values of the gains the control system becomes oscillatory unstable. C. Block diagram of the UPFC controller The controller of each of the three inputs , , Br consists of two paths: (i) main path executing the required steady-state control strategy (not analysed in this paper); and (ii) supplementary control loop executing the stabilising control discussed in this paper. The main controller is usually an integrator with a negative feedback loop and a relatively large time constant. The supplementary stabilising controller (or power system stabiliser PSS) is a practical differentiator with a small time constant. Block diagram of such a controller for real power Pb and

Fig. 4. Block diagram of the AVR+PSS structure with supplementary loop.

V. THE UPFC CONTROL A. Control law Recently the Lyapunov direct method has been applied to design UPFC control. It was proved in case second-order nonlinear single-machine-infinite-busbar system model [3], [4] that the derivative of the Lyapunov function can be expressed as

input signal is shown in Fig. 5. Block diagram for the reactive power Qb and signal would be similar.
Pb K

T s 1+ T s Lim 1 TI s + +

Pref Pb _

simulation tests sixth-order generator model has been used so that sub-transient effects are considered. Two types of electromechanical oscillations, whose frequencies are respectively about 0,8 Hz and 1,2 Hz occur in the presented system. The nature of those oscillations is different. The frequency oscillation 1,2 Hz results from local mode between generators G1 and G2, whereas 0,8 Hz are inter-area mode oscillations between System 1 and System 2. Detailed dates of the test system can be found in [4].
B1 B7 L1 B4 L2 B8 B3

Fig. 5. Block diagram of controller.

The shunt part of UPFC can operate as the reactive power compensator controlling the voltage U b and signal Br . Hence the block diagram of the controller of the shunt branch would be similar as in Fig. 5 but the input signal for the main path would be voltage U b while the input signal for the PSS would be Qb . The time constant of the main control path is large enough not to create a strong signal during power swings. The integrator averages the oscillatory changes of the input signal resulting in very small changes of the output signal so that the PSS signal dominates. VI. SELECTION OF THE PARAMETERS OF THE CONTROLLERS In the case of AVR+PSS system the PSS signal is subtracted from the AVR signal to give the exciter voltage VE. Thus the gain of the PSS cannot be too high as this would reduce the effective gain of the AVR+PSS system for the voltage regulation purposes. Hence the PSS gain has been chosen equal to 6.66. More detailed information and additional references corresponding to analysis of choosing and definition of the parameters of the presented AVR+PSS system can be found in [6]. Results of researches concerning the UPFC controller parameters choosing has been shown in [4]. But it should be emphasized that the main difficulty of UPFC parameter selection is linked to the fact noted in section V that the gains K , K , K B used in the local strategies have to be constrained. For high values of the gains the control system becomes oscillatory unstable what is shown in section VIII. The detailed parameter data of both considered controllers AVR+PSS and UPFC are presented in appendix. VII. TEST SYSTEM The results of eigenvalue analysis and simulation tests for the three-machine test system are presented in Fig. 6. The analysed system comprises: System 1, which includes two generators G1 and G2, System 2 represented by means of a single generator called G3. The nominal power of generator G3 is several times higher than the nominal power of G1 and G2 generators and corresponds to infinite bus system. Additionally the excitation system and voltage regulator of G3 has been ignored. The UPFC device is switched on in the middle of the line 4. In the case of eigenvalue analysis all generators has been represented using a fourth order model, whereas in the case of

G1 L3 L4a B6 a
UPFC

G3 L5 b L4b B5 B2 G2 System 1 System 2

Fig. 6. Diagram of the three-machine test system.

VIII. EIGENVALUE ANALYSIS The influence of the shunt used for modelling one part of the assumed model of UPFC can be neglected in this type of analysis. Hence the gain of path UPFC controller responsible for susceptance Br changes has been equal to zero. The difficulties with appropriate choosing of UPFC controller parameters are shown in Table II and Table III. The presented results have been obtained in case when only the AVR and UPFC influence is considered while the PSS effect is ignored.
TABLE II EIGENVALUES FOR VARIOUS VALUES OF GAIN gain eigenvalues local -0.551j7.584 -0.568j7.517 -0.523j7.458 -0.487j7.448 inter-area -0.299j5.084 -0.305j5.044 -0.287j5.005 -0.268j4.989 TABLE III unstable 2.021j17.297 6.641j16.285

K
0.25 0.30 0.35 0.38

EIGENVALUES FOR VARIOUS VALUES OF GAIN gain eigenvalues local -0.551j7.584 -0.570j7.563 -0.572j7.501 -0.562j7.486 inter-area -0.299j5.084 -0.309j5.070 -0.316j5.035 -0.314j5.025

K
unstable 0.428j24.038 5.295j23.405

K
0.20 0.25 0.35 0.38

Both Table II and Table III present results for the changes of only one of the following parameters: K and K . In base case their respective values are: K = 0.25 and K = 0.20 . In this case no unstable eigenvalues occur. Increasing the gain

value of K or K makes a small increasing of both electromechanical eigenvalues damping, then a further increasing results in decreasing of damping electromechanical eigenvalues. At the same time eigenvalues linked to UPFC control mode pass to the right complex half-plane and the system becomes unstable. The appearing of permanent oscillations linked to instability control mode is called hunting phenomenon [7]. Above inconveniences can be overcome by using adaptive control which has been proposed in [8] for the case of SVC element. A. UPFC and PSS comparative study A comparative study of control efficiency of both PSS and UPFC controllers are shown in Table IV. Results of analysis included in the table presented below concern the following cases: i) Case 1 - only AVR influence is considered, ii) Case 2 both AVR and PSS influence are considered, UPFC action is ignored, iii) Case 3 both AVR and UPFC influence are considered, PSS influence is ignored, iv) Case 4 - all elements (AVR, PSS, UPFC) influence is considered.
TABLE IV EIGENVALUES FOR VARIOUS CONTROL SYSTEMS Eigenvalues Local Inter-area -0.399j7.588 -0.166j5.129 -0.788j7.489 -0.507j4.998 -0.551j7.584 -0.299j5.084 -0.939j7.540 -0.654j5.024

Fig.7 and Fig. 8 show the post-fault variation of the real power and generator terminal voltage of generator G1, situated relatively far from the fault node, whereas G2 is close to the fault. It results from Fig. 7 that both discussed controllers greatly reduced the real power settling time as well as terminal voltage. In case of generator G1 real power: the first overshoot was reduced and both compared controllers increase swing damping at the same degree. The best features were achieved during simultaneous action of both PSS and UPFC. From the generator terminal voltage control point of view, the UPFC controller provides better features.
Real power P1 (p.u.) 1.8 1.4 1.0 0.6 0.2 0 1
2

3 Time (s)

Terminal voltage V1 (p.u.)

1.15 1.1 1.05 1.0 0.95 0.9 0

1 2

Case 1 Case 2 Case 3 Case 4

3 Time (s)

Fig.7. Damping of local swings: generator G1.

In all considered cases test system is stable. Each proposed controller contributes to increase the damping of electromechanical eigenvalues. Clearly the PSS achieved better damping than the UPFC. The best results have been obtained for Case 4, in which both PSS and UPFC influence is considered. The above-mentioned remark leads to the conclusion that the proposed control of PSS and UPFC, resulting from the Laypunov stability theory provide additive damping. IX. SIMULATION RESULTS The simulation test has been carried out for various types of disturbances, among others non-zero initial condition, step change in the generator voltage reference value, various network short-circuits. All conducted tests lead to the similar conclusions and confirm one another. Having regard to this fact, in this paper only the results of temporary short circuit at node 5 will be shown. The fault duration was assumed to be equal to 150 ms. System responses are shown in Fig .7, Fig. 8 and Fig. 9. All control cases, which have been analysed in the previous section, have been compared. The dotted line corresponds to Case 1 when both PSS and UPFC were not active. The dashed-dotted line corresponds to Case 2. The dashed line corresponds to Case 3 while the solid line shows influence both PSS and UPFC (Case 4).

Fig. 8 shows the power and voltage swings for generator G2 that is situated closer to the fault. Again the settling time of real power was reduced and simultaneous action of PSS and UPFC brings the best result of damping effect. Also it has been shown that a better reduction in first overshoot is provided by the PSS. The UPFC control is clearly worst. It is due to the fact that control strategy (10)- (12) is optimal after the disturbance, whereas during the disturbance itself its control influence may not be efficient. In order to eliminate the mentioned disadvantage a certain modification should be introduced in its control strategy. The controller with supplementary control loop shown in Fig. 4, which executes the strategy (5) does not have the above-mentioned inconvenience. Fig. 9 shows additional results corresponding to real power variation at line L4, at which a UPFC element is installed. The second curve represents the bus 5 voltage variation. Both curves confirm the previous results obtained for generators G1 and G2. Both controllers contribute to the enhancement of power swing damping, however UPFC control effect during disturbance was not appropriate. From the voltage control point of view, it has been observed that the UPFC device has better control features than PSS. The general conclusion drawn out from the above discussed results is that UPFC and PSS controllers provide post-fault additive damping while during the disturbance itself the PSS positive action is reduced by inappropriate UPFC

control. This inconvenience is compensated by significant reduction of settling time.

4.0

3.5 3.0
2.5 2.0 1.5

1.0

AVR and UPFC controller. The robustness of both proposed controllers has been proved. It was also shown that the proposed controllers used local available measurements to execute the derived control strategy and could be easily applied to power system. Both PSS and UPFC controllers provide post-fault additive damping while PSS action is better then UPFC when performing control during disturbance. The above-presented theoretical analysis has been supported by test results, which have shown a very good performance of both discussed controllers. XII. REFERENCES
[1] [2] [3] J. Machowski, J.W. Bialek, J.R. Bumby, Power System Dynamics and Stability, John Wiley & Sons, 1997. J. Machowski, S. Robak, J. W. Bialek, J.R. Bumby, N. Abi-Samra, N, "Decentralized stabilizing control of synchronous generators," IEEE Trans. on Power Systems, Vol. 15, No.4, pp.1336-1344, Nov. 2000. M. Januszewski, J. Machowski, J.W. Biaek, "Application of direct Lyapunov method to improve damping of power swings by control of UPFC, " in Proc. International Symposium Modern Electric Power Systems, Wrocaw, September 11-13, 2002, pp. 93-98. M. Januszewski, "FACTS devices as means of transient stability enhancement of power systems," - in Polish. PhD dissertation, Warsaw University of Technology, Warsaw, 2002. Modeling of Power Electronics Equipment (FACTS) in Load Flow and Satbility Programs, Task Force 38.01.08 Cigre 1999. S. Robak, J. W. Bialek, J. Machowski, "Comparison of different control structures for Lyapunov-based power system stabilizer," in Proc. The 22nd International Conference on Power Industry Computer Applications, PICA 2001, Sydney, Australia. E. Agneholm: The Restoration Process following a major Breakdown in a Power System Report No. 230L, May 1996 (Chalmers UT) J. Machowski, D. Nelles, "Simple robust adaptive control of static VAR compensator," European Transaction on Electric Power Engineering, No 6, Vol.3, 1993.

Real power P2 (p.u.)

3 Time (s)

Terminal voltage V2 (p.u.)

1.15 1.1
1.05

1.0 0.95 0.9 0

1 2 3 Time (s) 4 5 6

[4] [5] [6]

Fig.8. Damping of local swings: generator G2.

Real power PL4 (p.u.)

2.5 2.0 1.5 1.0 0.5 0.0 0

1 2

[7] [8]

3 Time (s)

XIII. BIOGRAPHIES
S Sylwester Robak was born in 1971 in Poland. He received his M.Sc. (1996) and Ph.D. (1999) degrees in Electrical Engineering from Warsaw University of Technology (Poland). Since 1999 he has been with Institute of Power Engineering, Warsaw University of Technology. His research interests are in the power system stability and in the automatic control..

Voltage V5 (p.u.)

1.05 1.0 0.95 0.9 0.85 0

1 2

4 3 5 6 Time (s) Fig.9. Other simulation results: (a) real power at line L4, voltage at B5.

X. APPENDIX The controllers were characterized by the following parameters: AVR: K A = 100 , TA = 1 s, TB = 10 s; PSS:
Tr = 10 s, K = 6.66 ; T1 = 0.05 s, T2 = 0.02 s, Tr = 3 s, T4 = 5.4 s, T = 0.2 s,

Marcin Januszewski was born in 1972 in Poland. He received his M.Sc. (1996) and Ph.D. (2002) degrees in Electrical Engineering from Warsaw University of Technology (Poland). Since 1995 he has been with Warsaw University of Technology faculty at the Power System Protection Division, Institute of Power Engineering. His research interests is in FACTS elements.

UPFC:

T = 0.4 s,

K = 0.25 ,

K = 0.2 , TB = 0.2 s, K B = 2 , TI = 10 s.

XI. CONCLUSIONS In this paper the features of two controllers implementing the resulting from Lyapunovs method control laws have been compared: the PSS constitutes a supplementary loop to the

Desire Dauphin Rasolomampionona was born in 1963 in Madagascar. He received his MSc (1988) and PhD (1994) in Electrical Engineering from Warsaw University of Technology. He joined the Warsaw University of Technology faculty in 1994 at the Power System Protection Division, Institute of Power Engineering.