Limit Induced Bifurcation Caused by SVC Capacity Limit
Lei Chen, Yong Min
State Key Lab of Power Systems Dept. of Electrical Engineering Tsinghua University Beijing 100084, China Email: chenlei99@[Link], minyong@[Link]
AbstractThe paper demonstrates and studies limit induced bifurcation (LIB) caused by SVC capacity limit. If the capacity limit is greater than a critical value, LIB occurs when an SVC reaches limit. In static analysis, the operating point is on the lower half of the PV curve after limit. Dynamic analysis reveals the immediate change of stability when LIB occurs. A stable equilibrium point and an unstable equilibrium point meet and vanish at the bifurcation. The consistency of the results of static and dynamic approaches veries the validity of static voltage stability analysis. LIB caused by SVC capacity limit in the IEEE 30-bus system is also presented.
In this paper we show that SVC reaching capacity limit may also cause LIB and immediate voltage collapse in the system. Reactive power limit induced bifurcation is a common phenomenon of dynamic reactive sources. When reaching limit, a dynamic reactive source loses the capability of regulating voltage and turns to a static reactive source, changing the structure of the system. The structural stability of the system is destroyed when LIB occurs. II. M ATHEMATICAL PRELIMINARY There are two types of limit in power systems, windup limit and non-windup limit [11]. Systems with these two types of limit behave quite differently. The case of windup limit is studied in [12] in detail. The limit of SVC used in this study is non-windup limit, so the case of non-windup limit is studied in this section. In the case of a non-windup limit, the output of the limiter is constrained if the measured output exceeds the specied limit and the derivative of the output is greater than zero. With nonwindup limits, the output comes off the limit as the sign of the derivative changes [11]. A non-windup limit directly affects the status of a state variable. When a state with a non-windup limit reaches its limit, the state variable gets stuck at this limit value until the dynamics drives the state to move away from the limit value. In other words, when a non-windup limit is encountered, and is active (the free dynamics is trying to push the state past the limit), then the state variable becomes a constant in the model, and the system loses a degree of freedom (system dimension in the sense of dynamics drops by one) [12]. The dynamic system with non-windup limits can be represented as: x = f (x, y, ) 0 = g (x, y, ) (1)
I. I NTRODUCTION In recent years, voltage stability has become a more important subject in electric power systems [1], [2]. Small disturbance voltage instability caused by slow change of system parameters, e.g. continuous increase of load, is one of the main research topics. Bifurcation theory gets a lot of applications in the investigation of voltage stability. Small disturbance voltage instability is commonly related to bifurcations in the system [3]. Many types of bifurcations may occur in nonlinear dynamic systems. Saddle Node Bifurcation (SNB), Hopf Bifurcation (HB) and Singularity Induced Bifurcation (SIB) are studied a lot in power system analysis. Besides them, another type of bifurcation, Limit Induced Bifurcation (LIB) [4][9] is also considered one of the important causes of small disturbance voltage instability. Many researchers have studied generator reactive power limit induced bifurcation, whose common behavior is described as follows. As the load increases, the system voltage drops and the generator reactive power outputs increase. When the reactive power of a certain generator reaches the limit, voltage collapse immediately occurs in the system. The explanation in static analysis is that the operation point is on the lower half of the PV curve after reactive limit [4]. Static VAr Compensator (SVC) is another common type of dynamic reactive source in real power systems. As the most widely used FACTS device, the SVC can rapidly regulate its reactive power output to control voltage. However, any actual SVC device has its capacity limit. It can only regulate the reactive power output in a limited regulating range. When reaching limit, the SVC becomes a xed capacitor [10], which is a static reactive source.
(1) is called the before-limit system. When the constrained state variable xk reaches its limit, the vector of state variables becomes x = (x1 , x2 , , xk1 , xk+1 , , xn )T and xk is replaced by its limit xlim k . The after-limit system is x = f (x , xlim k , y, ) 0 = g (x , xlim k , y, ) (2)
Because of the changes of the structure of system equations, the same equilibrium points will have different Jacobian matrices and different eigenvalues in the two systems before and after limit. The equilibrium point which is stable in (1) may be unstable in (2). This type of immediate change of stability is LIB. The bifurcation is the negation of system structural stability. The essence of LIB is that when variables reach their limits, the stability of equilibrium immediately changes because of the change of system structure. III. LIB
CAUSED BY
SVC CAPACITY LIMIT
Fig. 2. V-I characteristic of the SVC
In this section, we will show that SVC reaching capacity limit can also cause LIB. Both static analysis and dynamic analysis are investigated. A. Static analysis The 2-bus system in Fig. 1 is studied. The load bus is connected to the innite bus with a lossless line. A TCR-FC (Thyristor Controlled Reactor and Fixed Capacitor) type SVC is installed at the load bus.
Fig. 3. SVC model with slope representation using conventional power ow PV bus
Fig. 1.
A 2-bus system
In the regulating range, an SVC varies its reactive power output so as to maintain the bus voltage. However, the terminal voltage is commonly not maintained constant, but allowed to vary with the compensating current with a specic regulation slope, as shown in Fig. 2. In static power ow analysis, an SVC in the regulating range can be represented as an auxiliary PV bus connected to the terminal bus with a small reactance equal to the regulation slope in p.u., as shown in Fig. 3 [13]. When reaching capacitive capacity limit, an SVC becomes a xed capacitor, whose capacity is equal to the capacitive limit Bmax . The reactive power output decreases with the square of voltage decrease and the capability of regulating voltage is lost. Immediate voltage collapse may be caused by this structural change. In the 2-bus system in Fig. 1, the voltage of the innite bus is 1.05. The reactance of line 1-2 is 0.5. The reference voltage Vref = 1.00. The reactance in Fig. 3 XSL = 0.05. That is, the regulation slope is also 0.05. The load of bus 2 increases slowly and is represented by . The active load P = . The reactive load Q = 0.5. In the static analysis of generator reactive power limit induced bifurcation, the operating point is on the lower half of the PV curve after generator reactive limit. We will show that this can also happen when the SVC reaches capacity limit. Figure 4 shows the PV curves of bus 2 when increases. Curve 1 is the PV curve when the SVC is in the regulating range. As the load increases, the voltage decreases slightly
because of the regulation slope. When the capacitive limit Bmax = 0.4, the SVC reaches its limit at point A and becomes a xed capacitor. The PV curve turns to curve 2. Point A is on the upper half of curve 2, so the system is still stable. However, when Bmax = 1.0, the SVC reaches its limit at point B, which is on the lower half of the PV curve after limit and unstable. The system immediately loses stability. Similar to the case of the generator, LIB occurs at point B and causes voltage instability. The value of at point B is the load margin.
2.5
2 Bmax=1.0 3 1.5 V (p.u.) Bmax=0.4 2 1 A SVC Regulating 1 B
0.5
0 0
0.2
0.4
0.6
0.8 (p.u.)
1.2
1.4
1.6
Fig. 4.
PV curves of the 2-bus system
B. Dynamic analysis Bifurcation is essentially a dynamic problem. In this section LIB caused by SVC capacity limit is studied with dynamic
approaches to understand the mechanism. The dynamic analysis utilizes the simplied SVC model described in [13], as shown in Fig. 5. The voltage measurement module and the thyristor susceptance control module are ignored. The dynamic equation of the SVC is: = 1 [KR (Vref Vt ) B ] B TR (3)
In (3), KR = 1/XSL , reecting the regulation slope of the SVC. The state variable B is restricted by non-windup limit. As described in Section II, B is removed from the state variables and considered a constant equal to Bmax when the upper limit is reached. This causes the change of system dynamic equations. The before-limit system is different from the after-limit system. The same equilibrium point may have different stabilities in the two systems.
Increase the capacity limit to Bmax = 1.0. The SVC reaches its limit at = 1.2708. The four eigenvalues of the before-limit system are -55.9195, -13.0710, -2.3203+j7.2558, -2.3203-j7.2558. The operating point is stable. The three eigenvalues of the after-limit system are -3.5406+j8.2558, 3.5406-j8.2558, 1.0081. There is a positive eigenvalue and the operating point is unstable. Because of the change of the system structure, the stability of the operating point changes. When the limit is reached, the system undergoes instability because of LIB. The locus of bus 2 voltage at the equilibrium point as the load increases is plotted in Fig. 6, which is also called PV curve in this paper. Before the bifurcation, there are two equilibrium points. One is stable and the other is unstable. As the load increases, the variable at the stable equilibrium point reaches limit. LIB occurs and the two equilibrium points meet and vanish. There is no equilibrium point after the bifurcation. This is the same as that of generator reactive power limit induced bifurcation. From the analysis above, it is found that LIB occurs only when the limit is greater than a critical value. LIB does not occur when the limit is small. We can also observe this phenomenon in the simulations of large power systems.
Fig. 5.
Block diagram of SVC
The load is represented as an induction motor using 3-order model described below. TJ s = Tm Te 1 Ex = 2f0 sEy [Ex + (X X )Iy ] (4) T E y = 2f0 sEx 1 [Ey (X X )Ix ] T where Tm = k [ + (1 )(1 s) ], Te = Ex Ix + Ey Iy , and k can be determined if the initial slip s0 is known. The parameters, mainly the impedances, should be converted into those in system MVA base when the model is used in the system computation. The load increase is not represented by the increase of Tm , but by the decrease of the impedances of the model. This is more reasonable because the load increase is often the result of the start-up of more motors, not the load increase of the running motors. The parameters of the SVC model are KR = 1/XSL = 20.0, TR = 0.05s. The parameters of the induction motor model are TJ = 2.0s, T = 0.576s, f0 = 50Hz, X = 3.7941, X = 0.4110, = 0.15, = 2.0, s0 = 0.0116. When Bmax = 0.4, the SVC reaches its limit at = 0.7554. Linearize the before-limit and the after-limit system at the operating point respectively and compute the eigenvalues of the Jacobian matrices. To the before-limit system, the four eigenvalues are -81.4900, -13.6969, -3.6043+j9.9555, -3.6043j9.9555. The operating point is stable. To the after-limit system, the three eigenvalues are -4.3318+j10.3179, -4.3318j10.3179, -2.4026. Although the system structure changes, the operating point is still stable.
0.8 V2(p.u.)
0.6
0.4
0.2
0 0
0.2
0.4
0.6
0.8 (p.u.)
1.2
1.4
Fig. 6.
Dynamic PV curve when LIB occurs
C. Comparison between static and dynamic results The results of the static analysis using power ow equations and the dynamic analysis using dynamic equations are compared in this section. When Bmax = 0.4, the SVC reaches its limit at = 0.7550 in static analysis and = 0.7554 in dynamic analysis. No LIB occurs in the two cases. When Bmax = 1.0, the SVC reaches its limit at = 1.2723 in static analysis and = 1.2708 in dynamic analysis. LIB occurs and the system becomes unstable when the limit is encountered in the two cases. Figure 7 shows the PV curves of static approach and dynamic approach. The result of dynamic approach is marked with +. The remarkable consistency of the results veries the validity of static approaches in voltage stability analysis.
2.5
TABLE I SVC Bus 19 21 30 Vref 1.03 1.04 1.00
PARAMETERS
Regulation Slope 0.05 0.02 0.02
Capacitive Limit (MVAr) 40.0 40.0 30.0
1.5 V (p.u.)
0.5
0 0
0.2
0.4
0.6
0.8 (p.u.)
1.2
1.4
1.6
Fig. 7.
Comparison between static and dynamic PV curves
Voltage of Bus 24(p.u.)
IV. E XAMPLE In this section we use the IEEE 30-bus test system to show LIB caused by SVC capacity limit in large power systems. The test system is shown in Fig. 8. Based on the analysis of Section III-C, the result of static analysis is consonant with that of dynamic analysis. Moreover, static analysis is still widely used in voltage stability analysis because of its simpleness and speed. In this section LIB caused by SVC capacity limit is studied with the static approach.
LIB occurs and the system undergoes voltage collapse when the SVC reaches capacity limit. The PV curve of bus 30 is also shown in Fig. 10. The result veries that in large power systems SVC reaching capacity limit may cause LIB and voltage instability. Furthermore, LIB only occurs when the SVC capacity limit is great. There is a critical limit, which is mentioned in Section III-B, playing an important role in LIB. In addition, it is found in the simulation that if the generators have no reactive power limits, SVC reaching limit does not cause LIB. The probability of LIB in a system with enough dynamic reactive sources is very little.
1.3 1.2 1.1 8 1 0.9 0.8 0.7 0.6 0.5 2.5 5 2 SVC 21 13 11 SVC 19 SVC 30
3.5
4.5 5 (p.u.)
5.5
6.5
Fig. 9.
PV curve of bus 24 when no LIB occurs
1.3 Bus 24 Bus 30 1.2
Fig. 8.
The IEEE 30-bus test system
1.1 8 Voltage(p.u.) 1 0.9 0.8 0.7 0.6 0.5 2.5 SVC 30 5 2 SVC 21 13 11 SVC 19
SVCs are installed at bus 19, 21 and 30. The reference voltages, the regulation slopes and the capacitive limits are shown in Table I. is the total active load. All the loads and the active power outputs of the generators are assumed to change uniformly in proportion to the value of . The reactive power limits of the generators and the SVCs are considered in the computation. Figure 9 is the PV curve. The points where the generators or the SVCs reach reactive power limit are marked with +. The SVC at bus 30 is the last one reaching capacity limit, but it does not cause LIB. Increase the capacitive limit of the SVC at bus 30 to 50.0 MVAr. The PV curves are shown in Fig. 10. In this case,
3.5
4.5 5 (p.u.)
5.5
6.5
Fig. 10.
PV curves when LIB occurs
V. C ONCLUSION A new type of LIB caused by SVC capacity limit is investigated in the paper. The SVC is another type of dynamic reactive source in real power systems besides the generator. In the regulating range, an SVC can rapidly varies the reactive output to control voltage. However, an SVC becomes a xed capacitor and loses the capability of regulating voltage when reaching limit. It turns to a static reactive source. The structural change may cause LIB and voltage instability in the system. In static analysis, the operating point where the limit is reached is on the lower half of the after-limit PV curve when LIB occurs. It is similar to the case of generator reactive power limit induced bifurcation. The dynamic analysis shows that when variables reach limits, the system structure changes and the same equilibrium point will have different Jacobian matrices and different eigenvalues in the two systems before and after limit. The essence of LIB is that the stable equilibrium point before limit immediately becomes unstable after limit. When LIB occurs, a stable equilibrium point and an unstable equilibrium point meet and vanish at the bifurcation. The comparison between static and dynamic results veries the validity of static analysis. The simulation in IEEE 30-bus test system also reveals LIB caused by SVC capacity limit. It can be concluded that LIB is a common phenomenon of dynamic reactive sources such as the generator, the SVC and the STATCOM. A dynamic reactive source loses the capability of regulating voltage and turns to a static reactive source when reaching reactive power limit. The Generator, the SVC and the STATCOM turn to static reactive sources of constant power, constant impedance and constant current, respectively. The role of the dynamic reactive source in maintaining system voltage stability is sometimes very important. Then the sudden loss of the dynamic reactive source may cause voltage instability, which is LIB in the analysis of the system. Furthermore, the preventive control measures of LIB will be the work in the future.
ACKNOWLEDGMENT The authors would like to acknowledge the support of Special Fund of the National Priority Basic Research of China (2004CB217904) and the National Science Foundation of China (50323002). R EFERENCES
[1] C. Taylor, Power System Voltage Stability. New York: McGraw-Hill, 1994. [2] T. Van Cutsem and C. Vournas, Voltage Stability of Electric Power Systems. Norwell, MA: Kluwer, 1998. [3] C. Canizares, I. Dobson, N. Miller et al. Voltage stability assessment: concepts, practices and tools, IEEE/PES Power System Stability Subcommittee Special Publication, 2002. [4] I. Dobson, L. Lu, Voltage collapse precipitated by the immediate change in stability when generator reactive power limits are encountered, IEEE Trans. Circuits Syst. I, vol. 39, no. 9, pp. 762-766, Sep. 1992. [5] M. L. Crow, J. Ayyagari, The effect of excitation limits on voltage stability, IEEE Trans. Circuits Syst. I, vol. 42, no. 12, pp. 1022-1026, Dec. 1995. [6] I. A. Hiskens, B. B. Chakrabarti, Direct calculation of reactive power limit points, Electr. Power Energy Syst., vol. 18, no. 2, pp. 121-129, 1996. [7] A. A. P. Lerm, C. A. Canizares, N. Mithulananthan, Effects of limits in small signal stability analysis of power systems, IEEE PES Summer Meeting 2001, vol. 3, pp. 1805-1810, Jul. 2001. [8] N. Yorino, H. Q. Li, H. Sasaki, A predictor/corrector scheme for obtaining q-limit points for power ow studies, IEEE Trans. Power Syst., vol. 20, no. 1, pp. 130-137, Feb. 2005. [9] Y. Kataoka, Y. Shinoda, Voltage stability limit of electric power systems with generator reactive power constraints considered, IEEE Trans. Power Syst., vol. 20, no. 2, pp. 951-962, May 2005. [10] N. G. Hingorani and L. Gyugyi, Understanding FACTS: Concepts and Technology of Flexible AC Transimmison Systems. New York: IEEE Press, 1999. [11] P. Kundur, Power System Stability and Control. New York: McGrawHill, 1994. [12] V. Venkatasubramanian, H. Schattler, J. Zaborszky, Dynamics of large constrained nonlinear systems - a taxonomy theory, Proc. IEEE, vol. 83, no. 11, pp. 1530-1561, Nov. 1995. [13] IEEE Special Stability Controls Working Group, Static var compensator models for power ow and dynamic performance simulation, IEEE Trans. Power Syst., vol. 9, no. 1, pp. 229-240, Feb. 1994.