IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 21, NO. 2, MAY 2006
791
Emmanuel G. Potamianakis and Costas D. Vournas, Fellow, IEEE
Abstract—This paper discusses a few aspects of shortterm voltage instability: modeling of transient overexcitation limiters of synchronous generators; possible outcomes of instability, such as induction motor stalling, or generator loss of synchronism leading to a local blackout; eigenvalue tracking and use of eigenvector to identify and characterize the instability mode; and ﬁnally, induction motor disconnection to counteract shortterm voltage instability. Results on a system with large penetration of induction wind generators, as well as on a small test system, commonly used in voltage stability studies, are included.
Index Terms—Eigenvalue tracking, eigenvector analysis, induc tion motor stalling, loss of synchronism, overexcitation limiter, voltage stability.
I. INTRODUCTION
S HORTTERM voltage stability as deﬁned in [1]–[3] and [12] is relatively less well known than its more traditional
longterm counterpart, mainly because there are not as many welldocumented incidents of shortterm voltage instability as longterm ones. However, several factors, such as increasing proportion of air conditioning and in general induction motor load, growing penetration of dispersed generation consuming reactive power without voltage control, increased use of various types of elec tronically controlled loads, and ﬁnally, introduction of HVDC ties linking weak areas, point out at an increasing risk of voltage instability in the short term. It was even suggested that part of the problems that led to the recent North American blackout of August 2003 might be linked to shortterm voltage instability. Voltage instability on the short term is driven by fast recov ering load components that tend to restore power consumption in the time scale of a second after a voltage drop caused by a con tingency. A typical such component is the induction motor. The active power consumption of an induction motor restores almost to constant power after a voltage drop [3]. Moreover, if the dis turbance is such that the electrical torque cannot overcome the mechanical load, the motor stalls, i.e., it decelerates, absorbing a high reactive current, thus precipitating a further voltage drop and occasionally a voltage collapse. Induction generators produce active power, but similarly to motors, they consume reactive power. In new wind farm
Manuscript received September 6, 2005; revised January 5, 2006. This work was supported in part by the EU Programs Dispower and MicroGrids. Paper no.
TPWRS005642005.
The authors are with the School of Electrical and Computer Engineering, Na tional Technical University of Athens (NTUA), Athens 100 22, Greece (email:
manpot@power.ece.ntua.gr; vournas@power.ece.ntua.gr). Digital Object Identiﬁer 10.1109/TPWRS.2006.873022
installations, the trend is to use variable speed wind generators, which are controlled through power electronics converters and are thus able to regulate their own voltage by producing reactive power. However, many existing installations of wind farms in Greece and several other countries continue to be based on simple, uncontrolled (constant speed) induction generators. These machines behave in terms of voltage stability similarly to induction motors, with the difference being that when the torque equilibrium is lost, due for instance to a sustained voltage drop, the generator tends to overaccelerate instead of stalling. Thus, even if not literally a load, the induction generator is a factor contributing to shortterm voltage instability. Induction machines are usually shuntcapacitor compensated to improve their power factor. The reactive support provided by shunt capacitors varies with the square of the voltage, and consequently, during a voltage drop, reactive support is greatly reduced. Thus, in order to avoid induction machine instability, dynamic and fast reactive compensation, such as provided by an SVC or a STATCOM, may become necessary [13]. Although voltage instability in the cases discussed in this paper is driven by induction machine loss of equilibrium, the latter is not likely to happen unless the voltage support pro vided by local synchronous generators is lost or reduced. This is usually the result of rotor current limitation brought about by the overexcitation limiter (OEL) of the synchronous machine. Usually the OEL acts as a slow device allowing the transient overexcitation of the generator for about 20 s before actually en forcing the rotor current limit. In this case, other control mech anisms, such as load tap changers (LTCs) have time to act, and the voltage stability problem becomes a longterm one. How ever, even transient overexcitation is not allowed above an in stantaneous limit that must be enforced in the short term, i.e., in a fraction of a second. When this limit is overstepped, for instance, after a severe contingency, or if it is inappropriately set to a relatively low value, a shortterm voltage stability problem can arise. Thus, in Section II, we review OEL models and in particular the mechanism of applying the fast overexcitation limit. In case of induction motor stalling or generator overspeed, protection equipment will normally disconnect the affected machine, thus relieving the system from the instability. It is noted, however, that small motors, in particular, are protected by thermal elements and may thus remain on stalled con ditions for several seconds. In such and other similar cases, there is a danger that during the shortterm instability phase and before protection equipment has the time to act, nearby synchronous machines may lose synchronism, thus leading to a local blackout.
08858950/$20.00 © 2006 IEEE
792
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 21, NO. 2, MAY 2006
In the ﬁrst system examined in [13] and presented here in Sec tion III, shortterm instability is contained to induction machines only, while the synchronous generators remain in synchronism. In Section IV, we will show a shortterm voltage instability case followed by loss of synchronism. In this latter case, timedo main simulations cannot provide a clear distinction between in duction motor and synchronous generator instability. It is thus necessary to develop special tools and in particular an eigen value tracking routine, which together with eigenvector and par ticipation factor analysis is used to identify the components and systems variables involved in the instability. Section V provides the modal analysis of the system in case of local generator transient ﬁeld limitation. Section VI presents the linearization of the system during simulation (online lin earization routine), as well as the results of eigenvalue tracking in the examined instability case. An undervoltage motor shed ding preventing the abovementioned instability is analyzed in Section VII. Finally, Section VIII presents the conclusions of this paper.
II. OEL MODELING _{A}_{S}_{P}_{E}_{C}_{T}_{S}
The simulation tool used in this paper for the analysis of shortterm voltage stability stems from an educational purpose simulation software developed in Matlab/Simulink in conjunc tion with the University of Liege [11] and was adapted for re search purposes in a package called WHSSP suitable for wind and hybrid power systems [10]. Of particular interest to short term voltage stability is the modeling of synchronous generator OELs shortly presented below.
A. Transient OEL Modeling
The modeling of OEL is detailed in [5] and has a signiﬁ cant effect on voltage stability analysis [6]. A certain level of overexcitation (i.e., ﬁeld current exceeding the longterm limit ) can be tolerated for several seconds. There is, how ever, a level of overexcitation that is not allowed to remain even for a short time and must be limited with very small delay. We note the corresponding ﬁeld current as . In the sequel, we concentrate on this transient OEL activation, assuming a pro portional, summedtype OEL [5], [6]. The ﬁeld current fed back into the OEL is given in the per unit system of the synchronous machine [3] by the following equation:
(1)
Fig. 1.
OEL transient block diagram.
repeated transient activations and deactivations, in case the ﬁeld current is close to the value . The limiter steadystate block is similar to the transient block, that is, when the ﬁeld current exceeds the respective value for a longer period (ranging from several seconds to one minute), the discrete output signal changes from zero to unity. As shown in Fig. 2, the output of the transient block is added with that of the steadystate limiter block through a summing junction. Consequently, if either block has a positive discrete output signal or , the OEL output signal will be
subtracted from the reference voltage input of the AVR, thus forcing the ﬁeld current within the corresponding limit. Note that the two discrete variables or can both be zero but cannot normally take the value of one at the same time, because when the steady block is activated, the ﬁeld current re
duces below
. Therefore, the OEL output signal is
given by the following function:
under no limitation
under transient limitation
under steadystate limitation.
(2)
The OEL model proposed in [3] introduces a block limiting
the OEL output to positive values only. In [13], this block is proven to have a detrimental effect on voltage stability if ap plied during oscillatory transients and is thus removed in the
model. This means that the OEL is allowed to provide tran
siently negative values to the AVR corresponding to ex
where is the transient EMF proportional to ﬁeld ﬂux linkage, and , , and are, respectively, the direct axis stator cur rent component, synchronous, and transient reactance. Fig. 1 shows in block diagram form the transient OEL com ponent that limits the ﬁeld current to the maximum transient permissible value. When the ﬁeld current exceeds , a time counter is activated, and after a short time delay of (tenths of a second), the discrete output signal changes from zero 
citation boosting. The typical ﬁeld current response of a synchronous generator equipped with the described proportional, summedtype OEL model is shown in Fig. 3. Note that a slight offset in the ﬁeld current limitation remains for both transient and steadystate OEL cases. These voltage errors are typical of proportional limiters [6]. 

to one. When the ﬁeld current returns below the transient limit for some time, the value of returns to zero. The small positive constant subtracted from in the input of the block diagram shown in Fig. 1 is used to avoid 
In the following case studies, note that no synchronous ma chine ﬁeld current reaches its steadystate limit value , be cause both examined networks collapse a short time after the respective transient limitation. Consequently, the steadystate block of the respective OEL remains inactive during simulation. 
POTAMIANAKIS AND VOURNAS: SHORTTERM VOLTAGE INSTABILITY
793
Fig. 3.
Evolution of synchronous generator rotor current under limitation.
Fig. 4.
Simulation results for South Evia network.
III. CASE STUDY I: INSTABILITY OF _{I}_{N}_{D}_{U}_{C}_{T}_{I}_{O}_{N} _{W}_{I}_{N}_{D} GENERATORS IN A REALISTIC SYSTEM
The system studied here corresponds roughly to the South Evia region of the Hellenic Interconnected System [8]. The examined network consists of a local conventional steam power plant with two synchronous generators of 176.5 MVA each and 19 wind farms of total nominal capacity of 200 MW connected to the distribution network. Each wind farm is modeled by an equivalent induction generator. The system interconnection to the main Hellenic Interconnected transmission system is simulated via two equivalent lines, with total impedance corre sponding to the shortcircuit level at the interconnection point. It was seen in [8] that with proper capacitor compensation, the system is voltage stable in the long term, even for very se vere contingencies. It is important to note that longterm voltage stability is maintained mainly because the LTCs at the HV/MV substations where wind generators are connected are able to regulate MV voltage, thus decreasing reactive power consump tion by the induction generators, before the activation of the re maining synchronous generator OEL. These results led to the conclusion that normal shunt capacitor banks were sufﬁcient to maintain stability by allowing the synchronous machines a wider margin of reactive support, and no dynamic reactive com pensation was deemed necessary. In the present paper, we consider a slight variant of the above scenario considering the effect of a transient limitation of the generator excitation ﬁeld, such as that performed by the OEL shown in Fig. 1. In order to investigate the risk of shortterm voltage instability, we assume a rather low value of . It should be noted that this is done as an academic exercise and does not reﬂect the degree of security of the actual power system. With this setting, the system is unstable after the following double contingency.
clear from the terminal voltage response of a selected wind farm in Livadi [8], as shown in Fig. 4(c). Even though the driving force of the shortterm instability is provided by the induction generators, this instability could not occur without the transient overexcitation limitation of the synchronous machine presented in Fig. 4(a). The shortterm voltage instability in this case involves only the induction generators as shown in Fig. 4(b) and (d), where the synchronous generator rotor angle and the speed of the se lected equivalent induction generator are plotted. Clearly, the synchronous machine remains in synchronism while the induc tion generators lose their equilibrium and overaccelerate. Dis connection of the wind farms will be the only harm done in this unstable case. Thus, the cause of the instability in this case is the transient excitation limitation of the synchronous machine, the instability driving force is the induction generator dynamics, and the ﬁnal outcome is the overacceleration and disconnection of induc tion generators, without any further effect on the interconnected system.
IV. CASE STUDY II: VOLTAGE INSTABILITY _{A}_{F}_{F}_{E}_{C}_{T}_{I}_{N}_{G} SYNCHRONOUS MACHINE
Not all shortterm instability cases are as simple to charac terize as the one examined in the previous section. In this sec tion, we demonstrate that even in a simple system, it may be difﬁcult to distinguish the cause, the driving force, and the ﬁnal outcome of shortterm instability.
A. Test System Description
The test system analyzed is the 11bus network presented in [1] and commonly used in voltage stability studies. The oneline equivalent diagram is shown in Fig. 5.
• 
At 

, there is an outage of one local generator. 
The system consists of an area fed through ﬁve 500kV par 
• 
At 
, there is loss of one interconnection line. 
allel lines by a remote generator SG2 of nominal capacity 2200 

The voltage instability resulting from the transient overex citation limitation of the synchronous machine ﬁeld winding is 
MVA and an inﬁnite bus (SG1) representing a large intercon nection. The presence of the inﬁnite bus ensures that the system 
794
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 21, NO. 2, MAY 2006
Fig. 5.
Oneline equivalent diagram of the examined test system.
frequency will remain at its nominal value during the simula tion. The abovementioned area is heavily shunt compensated and contains a 1600MVA local synchronous generator SG3 and two aggregate loads, one industrial directly served via the off nom inal constant ratio transformer T4, and one commercialresiden tial on bus 6. The terminal voltage of the local generator is controlled by an AVR of type AC4 according to IEEE deﬁnitions [2] (the tran sient gain reduction block is neglected), equipped with propor tional, summedtype OEL with a transient ﬁeld current limita tion block. The industrial load is represented by two equivalent industrial induction motors with different parameters: a large one (IM1) of 3375 MVA and a small one (IM2) of 500 MVA. The 3000MW commercialresidential load is half resistive of constant admit tance and half motor. The 2440MVA single motor equivalent (IM3) is an aggregate of motors heavily dominated by aircondi tioning load. This load is connected to the transmission network through two transformers (T5 and T6) and a 115kV transmis sion line between buses 10 and 11. The latter transformer T6 is equipped with LTC mechanism, in order to keep the bus voltage of the commercialresidential load in a desired range of values. In the simulation, the synchronous generators are modeled through a fourthorder model, representing the ﬁeld winding and one damper winding in the rotor quadrature axis. The induction motors are described through their thirdorder model, including rotor mechanical and electrical transients. The motor mechan ical load includes both constant and quadratic torque compo nents. Since the time frame of interest extends to a few seconds, the response of the synchronous generator prime movers can be neglected. As a result, the mechanical input torque of generators is considered constant during the simulation. In the initial operation point, the remote generator and the inﬁnite bus deliver about 5000 MW to the load area.
B. Simulation Results
Typically, transient limitation of ﬁeld current is applied at 140% of rated with a time delay of about 1 s [7]. However, as in the previous case study, we assume a much tighter tran sient ﬁeld current limitation on the local synchronous generator
p.u. for 100 ms). The disturbance simulated is the tripping of one line circuit
between buses 8 and 9 at
. This leads to shortterm
Fig. 6.
Simulation results (a), (b) Induction motors. (c) ,(d) Static loads.
voltage instability as seen in Fig. 6(a), while the simulation is terminated a short time after , due to the unacceptably low value of the local generator terminal voltage. To demonstrate the prominent role of the induction motors in the instability, the system is simulated again with the mo tors modeled as static loads (constant active power, exponential function of voltage with exponent of 1.5 for ). The result is shown in Fig. 6(c). Even though this is a stiff load model, the responses are stable, and voltage is maintained. The cause of the instability is again the transient overexcita tion limitation of the local synchronous generator. This is shown in Fig. 6(b), where again, the effect of induction machine insta bility in overexciting the generator is clear. However, for the static load representation [see Fig. 6(d)], the ﬁeld current is limited to its transient value without any problem. Thus, it can be deduced that the driving force of the instability is induction motor dynamics. Of particular interest is that the ﬁnal outcome of instability in the above case is the local generator (SG3) loss of synchronism at about the same time that the induction motors are stalling. This is shown in Fig. 7(a) and (b). From this ﬁgure, it could be argued that we have a transient angle instability case as well as motor stalling. The generator loss of synchronism makes this case more se vere than the previous one examined in Section III, as the loss of synchronism, if not contained, will lead to a local blackout. At the same time, this demonstrates that clear distinction between induction motor and synchronous generator stability is not al ways possible at ﬁrst glance. A deeper examination of this case using modal analysis and an online linearization tool will follow.
V. MODAL ANALYSIS
A. Saddle Node Bifurcation
As discussed in the previous section, the cause of induction motor instability is the overexcitation limitation of the local gen
erator. In this section, we investigate further the driving force of
POTAMIANAKIS AND VOURNAS: SHORTTERM VOLTAGE INSTABILITY
795
Fig. 7.
(a) Synchronous generator rotor angle. (b) Induction motor speed.
the instability mechanism. For this reason, the transient OEL limit value is increased so that the system is able to reach postdisturbance equilibrium, and for equilibrium points corre sponding to different values of , the system is linearized, and the respective eigenvalues are computed.
In Fig. 8, we present the evolution of the real eigenvalue closest to the origin (critical eigenvalue) as a function of the local generator transient limit value ( ). Note in Fig. 8 that as the limited value of ﬁeld current is reduced, the critical eigenvalue is approaching zero; thus, the system tends to a saddle node bifurcation (SNB) marked by the loss of postdisturbance equilibrium. The last eigenvalue before the SNB is equal to , cor
p.u. or equivalently to
is only slightly reduced (i.e.,
p.u.), the system will become unstable, and the
equilibrium will disappear.
Clearly, the critical eigenvalue corresponds to a monotonic
mode. In order to characterize this mode, an eigenvector anal ysis is necessary.
B. Eigenanalysis
The eigenanalysis of the postdisturbance equilibrium point for p.u. is summarized in Table I, where the eigen values, the dominant state variables, and the respective eigen vectors and participation factors are given.
Ten eigenvalues are associated with the two synchronous gen
erators SG2 and SG3, giving four oscillatory and two mono
tonic modes. Similarly, nine eigenvalues are related to induc
tion motor dynamics, giving rise to three oscillatory and three monotonic modes.
796
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 21, NO. 2, MAY 2006
The modes in Table I are characterized based upon the relative phase of the right eigenvector elements, as well as the magnitude of the participation factors. In multimachine systems, the combined modes of response for interacting synchronous or induction machines may form common modes (where the same variables of different machines respond in phase) and interaction modes (where similar vari ables of different machines respond in opposite directions, i.e.,
180 out of phase) [15].
As seen in Table I, the industrial motors IM1 and IM2 exhibit
large induction machine dynamics [14], i.e., they have an oscil latory response that involves slip and internal EMF angle and a monotonic mode involving magnetic ﬂux. This is evident in the participation factors of relevant modes where the slip ( ) and
state variables dominate, while the monotonic responses, as expressed by the real eigenvalues, are related to variables. On the other hand, the modes of the residentialcommercial motor IM3 correspond to small induction machine dynamics [14]. This means that it exhibits an oscillatory behavior of the electrical state variables representing the rotor windings (i.e., an oscillation involving and ), while the mechanical speed response is relatively slower and monotonic. It is interesting to note that the critical eigenvalue, which is prone to enter the righthand side of the complex plane, corre sponds to the industrial motors common ﬂux mode. Note ﬁnally that the frequency of the local generator excita tion oscillation mode is very high, due to the transient excita tion limitation, and that the interaction synchronous generator electromechanical oscillatory mode appears to be less damped than the respective common mode. The above analysis clearly shows that the industrial motors form the driving force of the instability. However, this analysis required the modiﬁcation of the system in order to establish an equilibrium point. The following section describes a tech nique that allows online linearization and tracking of eigen values during simulation.
VI. EIGENVALUE _{T}_{R}_{A}_{C}_{K}_{I}_{N}_{G}
A. Online Linearization
A special feature incorporated in the WHSSP simulation package is the eigenvalue tracking facility. This program lin earizes the simulated system by building the state Jacobian matrix [4] at selected time instants during the simulation and computes its eigenvalues. To be more speciﬁc, the dynamics of a power system are cap tured by the following differential equations:
(3)
where are smooth functions, is the state vector, is the vector of algebraic variables, and is the input vector. Because the system is not necessarily at equilibrium during the simulation, the state Jacobian computed during simulation can be thought to belong to an implicit equilibrium point [3], for which the inputs have been changed to obtain an equilibrium point for the current values of the state variables. This means that at a speciﬁc time instant , the input vector is considered
Fig. 9.
Evolution of critical eigenvalue in the instability case.
to have changed from its actual value order to satisfy the following equation: 

to a value 

in 

(4) 
In case of synchronous generators, the input vector includes the reference voltage of AVR, the limiting values of OEL, as well as the prime mover reference. For the induction machines, the input vector contains the mechanical torque input. The information provided by the online linearization eigen values is not a strictly rigorous indication of stability, but it is quite useful in practice in order to characterize the nature of a possible instability, as will be done in the sequel for the case of Section IV.
B. Induction Motor Stalling
Fig. 9 shows the evolution of the dominant critical real eigen value in the instability case of Fig. 6(a). The examined eigenvalue corresponds to industrial motor IM1+IM2 common ﬂux mode and becomes positive at
. After this point, the motor demagnetization and stalling process has started, and even reduction of torque to achieve an equilibrium will not sufﬁce to contain the instability, as the achieved equilibrium would be unstable. Thus, it is clear from the online linearization that the induction motors drive the shortterm instability. The initial oscillations of the critical eigenvalue are due to the oscillations of the limited ﬁeld current of the local generator
[see Fig. 6(b)].
C. Synchronous Generator Loss of Synchronism
As seen in Section IV, the outcome of the examined instability scenario was the local generator loss of synchronism. Eigen value tracking is used again to explain the mechanism of angle instability in this case. Fig. 10 presents an enlargement of the root locus diagram during simulation for the time interval be tween (marked with circle) and (marked
with square).
POTAMIANAKIS AND VOURNAS: SHORTTERM VOLTAGE INSTABILITY
797
Fig. 10.
Root locus from to in the instability case.
The root locus contains three eigenvalues, which appear ei ther as a complex pair and a real one or as three real ones. At the start of the root locus ( ), the complex pair of the eigenvalues corresponds to the common electromechan ical oscillation mode, while the real one is the monotonic mode of the commercialresidential induction motor IM3. While the simulation is running, the complex pair is moving to the real axis with almost constant real part, while the real eigenvalue is moving to the right. At , the complex pair of eigenvalues collapses on the real axis, and the resulting two real eigenvalues are moving to opposite directions. At this point, the local generator is losing synchronism, as the equivalent synchronizing coefﬁ cient is rapidly approaching zero. Following the loss of synchronism, one real eigenvalue enters the righthand side of the complex plane, describing a mono tonic angle instability. Note that this is the second real positive eigenvalue, as the one shown in Fig. 9 already has a value close to 4 . Thus, the induction motors stall ﬁrst, and the generator loss of synchronism follows after roughly 3–4 s.
Fig. 11.
Machines terminal voltage (IM2 undervoltage shedding).
Fig. 12.
Local generator rotor current (IM2 undervoltage shedding).
VII. UNDERVOLTAGE MOTOR SHEDDING
The analysis of instability driving force and mechanism of the previous sections is not a mere academic exercise, as it may lead to the correct assessment of possible countermeasures to contain the instability. For instance, if the above case was diag nosed as angle instability, based solely on Fig. 7(a), the means of control would be sought in generation rejection. Since, how ever, the case has been identiﬁed as induction motor instability, the measures to be taken are related to motor disconnection. Thus, the instability can be corrected by undervoltage shed ding of the small industrial motor IM2. It is assumed that IM2 is disconnected from the network, if its terminal voltage drops below 0.9 p.u. for a time interval of few seconds. However, this delay should not exceed a critical value, because after this time, the remaining motors will be no longer attracted to their postdisturbance equilibrium point. This critical time is calcu lated using repeated simulations, as shown in Fig. 11, where the
terminal voltages of the machines are shown, as well as the un dervoltage threshold and the critical time delay with dashed line in Fig. 11(c). The solidline curves marked with 1 in Fig. 11 present the simulation results of a stable scenario, considering that the time delay of IM2 disconnection is equal to the critical interval of 3.54 s. On the other hand, assuming that the examined time delay is equal to 3.55 s, the posttripping response becomes un stable, as shown in dottedline curves marked with 2 in Fig. 11. As seen in Fig. 11, the small industrial motor IM2 is discon nected from the network a little before , leading to the reduction of system total load. As a result, the bus voltages are restored, and the network reaches a stable postdisturbance equi librium point. The rotor current of the local generator SG3 is shown in Fig. 12. For the stable scenario 1, note that following the IM2 discon nection, the local generator ﬁeld current is reduced below its
798
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 21, NO. 2, MAY 2006
Fig. 13.
(a), (b) Synchronous generator rotor angle. (c), (d) Induction motor
speed (IM2 undervoltage shedding).
transient limit. As a result, the OEL transient block is deacti vated after 600 ms (at about ), leading to further reduc tion of the respective rotor current. Synchronous generator rotor angles, as well as the rotor speed of the two remaining motors, are shown in Fig. 13. Note that the unsuccessful delayed IM2 undervoltage tripping results in shortterm instability of the remaining motors, leading again to local generator loss of synchronism.
VIII. CONCLUSION
In this paper, we presented shortterm voltage stability results on two test systems. In both cases, the instability was initiated after a contingency that forced a local synchronous generator to its transient overexcitation limit, which was taken to be lower than is usual in practice. The driving force of the instability was identiﬁed in both cases to be the induction machines, either wind generators or equivalent motors representing industrial and res idential components of load. Of particular interest in the second case was that the instability of induction motors was also affecting the local syn chronous generator that was losing synchronism, thus leading to a local blackout. Eigenvalue tracking and eigenvector and participation factor analysis was used to help identify the components and system variables involved in the instability. Finally, an induction undervoltage motor shedding was pro posed, in order to prevent the detected instability from leading to a local blackout.
REFERENCES
[1] C. W. Taylor, Power System Voltage Stability. GrawHill, 1994.
New York: EPRI/Mc
[2] P. Kundur, Power System Stability. New York: McGrawHill, 1994,
[3] 
EPRI Power System Engineering Series. T. Van Cutsem and C. Vournas, Voltage Stability of Electric Power Sys tems. Norwell, MA: Kluwer, 1998. 
[4] 
P. W. Sauer and M. A. Pai, Power System Dynamics and Stability. En glewood Cliffs, NJ: PrenticeHall, 1998. 
[5] IEEE Task Force on Excitation Limiters, “Recommended models for overexcitation limiting devices,” IEEE Trans. Energy Convers., vol. 10, no. 4, pp. 706–713, Dec. 1995.
[6]
C. D. Vournas, G. A. Manos, P. W. Sauer, and M. A. Pai, “Effect of overexcitation limiters on power system longterm modeling,” IEEE Trans. Energy Convers., vol. 14, no. 4, pp. 1529–1536, Dec. 1999.
[7] A. Murdoch, R. W. Delmerico, S. Venkataraman, R. A. Lawson, J. E. Curran, and W. R. Pearson, “Excitation system protective limiters and their effect on volt/var control—Design, computer modeling, and ﬁeld testing,” IEEE Trans. Energy Convers., vol. 15, no. 4, pp. 440–450, Dec.
2000. 

[8] 
C. D. Vournas, G. A. Manos, E. G. Potamianakis, and J. Kabouris, 
“Voltage security assessment of Greek interconnected power system with large wind penetration,” in Proc. Eur. Wind Energy Conf. Exhib., Copenhagen, Denmark, Jul. 2001. [9] J. A. Diaz de Leon II and C. W. Taylor, “Understanding and solving shortterm voltage stability problems,” in Proc. IEEE PES Summer 

Meeting, Chicago, IL, Jul. 21–25, 2002. [10] E. G. Potamianakis and C. D. Vournas, “Modeling and simulation of 

[11] 
small hybrid power systems,” in Proc. IEEE Bologna Power Tech Conf., Bologna, Italy, Jun. 23–26, 2003. C. D. Vournas, E. G. Potamianakis, C. Moors, and T. Van Cutsem, “An educational simulation tool for power system control and stability,” 
IEEE Trans. Power Syst., Special Section on Power Engineering Edu cation, vol. 19, no. 1, pp. 48–55, Feb. 2004. [12] P. Kundur et al., “Deﬁnitions and classiﬁcation of power system sta 

bility: IEEE/CIGRE Joint Task Force on Stability Terms and Deﬁni tions,” IEEE Trans. Power Syst., vol. 19, no. 2, pp. 1387–1401, May 

2004. 
[13] E. G. Potamianakis and C. D. Vournas, “Investigation of shortterm
[14]
voltage stability problems and countermeasures,” in Proc. IEEE PSCE Conf., New York, Oct. 2004. B. M. Nomikos and C. D. Vournas, “Investigation of induction machine
contribution to power system oscillations,” IEEE Trans. Power Syst., vol. 20, no. 2, pp. 916–925, May 2005. [15] B. M. Nomikos, E. G. Potamianakis, and C. D. Vournas, “Oscillatory stability and limit cycle in an autonomous system with wind generation,” in Proc. IEEE St. Petersburg Power Tech Conf., St. Petersburg, Russia, Jun. 27–30, 2005.
Emmanuel G. Potamianakis was born in 1978. He received the Diploma of Electrical Engineering from the National Technical University of Athens (NTUA), Athens, Greece, in 2000. He is currently working toward the Ph.D. degree at NTUA. His research mainly deals with the integration of distributed generation to the electric grids from the viewpoint of stability.
Costas D. Vournas (S’77–M’87–SM’95–F’05) received the Diploma of Elec trical and Mechanical Engineering from the National Technical University of Athens (NTUA), Athens, Greece. in 1975, the M.Sc. degree in electrical engi neering from the University of Saskatchewan, Saskatoon, SK, Canada, in 1976, and the D.Eng. from NTUA in 1986. He is currently a Professor in the Electrical Energy Systems Laboratory, School of Electrical and Computer Engineering, NTUA. His research interests are in the area of power system dynamics, stability, and control and include voltage stability analysis, wind generator integration in power systems, and the effect of deregulation on power system operation and control.
Viel mehr als nur Dokumente.
Entdecken, was Scribd alles zu bieten hat, inklusive Bücher und Hörbücher von großen Verlagen.
Jederzeit kündbar.