Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
AbstractThe classical theory of representation of power In steady state, the current flowing in the rotor winding is
swings in the impedance plane is based on the representation of equal to the exciter voltage divided by the winding resistance:
synchronous generators as constant voltage sources. The classical
model of a synchronous generator represents the machine as a Vf
If = (2)
constant voltage source behind the transient reactance in the rf
direct axis. The classical model of synchronous generators is
based on the assumption that the rotor flux linkage will not The field winding has a self-inductance Lff. A fundamental
change during a short period of time following a major distur- characteristic of a synchronous generator is the direct-axis
bance. In reality, with constant excitation voltage, the rotor flux open-circuit transient time constant T'd0, the ratio of the field
linkage will decrease and the internal generator voltage will de-
crease accordingly. The addition of an Automatic Voltage Regu-
self inductance over its dc resistance:
lator (AVR) boosts the excitation voltage following a disturbance L ff
so that the rotor flux linkage will be sustained and the generator Td' 0 = (3)
internal voltage will be prevented from collapsing. The purpose
rf
of this paper is to show how power swing representation in the The order of magnitude of this time constant, which is typ-
impedance plane will depart from the classical theory when com-
ically a few seconds, indicates that the voltage at the syn-
plex AVRs are used on modern generators.
chronous generator terminals cannot be changed instanta-
I. BASIC SYNCHRONOUS GENERATOR PHYSICAL PRINCIPLES neously; in other words, the current in the field winding varies
according to the field open-circuit time constant.
This paper is not intended to be an exhaustive review of When the generator is loaded, the three-phase currents will
synchronous generator physical and engineering principles, create a flux represented in Fig. 2 as ia+ib+ic. This flux is
but rather an overview of fundamental and essential facts. It is known as the armature reaction. The vectorial addition of the
assumed here that the reader is familiar with the two-axis field flux and the armature reaction is the air-gap flux
model representation of synchronous machines [1].
represented as ag.
Direct Quadrature Direct Quadrature
Axis Axis
ia Axis Axis
ia
f f
ag
ic Eq ib ic ib
If Stator If
Phase
vf ia + ib + ic
Vf Windings
ib ic ib ic
ia ia
Fig. 1 represents a two-pole synchronous generator. The The projection of the air-gap flux on the direct and quadra-
rotor winding is supplied with a dc exciter voltage Vf. A dc ture axis, as in Fig. 3, defines the fluxes in the direct and qua-
current If flows in the rotor winding to create a flux f in the drature axes d and q.
generator direct axis. When the primary mover drives the ge- The fluxes in the direct and quadrature axes create the cor-
nerator is and the generator rotates at synchronous speed, this responding voltages vd and va along the same axes. We can
flux will induce a voltage in the three-phase stator windings. demonstrate [1] that these voltages:
When the generator is unloaded, we can measure an excitation d d
or generator internal voltage at the terminals. This internal vd = ra id + q (4)
dt
voltage, Eq in Fig. 1, is proportional to the current If and lies in
and
the quadrature axis 90 f. Eq is given by (1).
d q
Eq = M f I f (1) vq = ra iq + + d (5)
dt
2
We define the following: tage. The dc output is connected to the machine field winding
Tm: variation of the mechanical power input to the by brushes and collector rings.
generator in pu In manual mode, either the level of the generator output
voltage, or the field current level (as in Fig. 11), is under the
H: inertia constant, seconds manual control of the operator. Although manual control of
M: inertia coefficient = 2H, seconds the excitation system still occurs on some old machines, or-
ganizations such as NERC in North America recommend
base rotor electrical speed in radians per
0: against this practice today because of the drawbacks and
second (377 rad/s)
shortcomings that this mode of operation entails.
The control block diagram that Fig. 10 represents allows us In automatic mode, a voltage set point is introduced in the
to use the technique of small-signal analysis [2], [5] to study summing point of the AVR. The excitation system compares
the dynamics of the elementary network in Fig.. this voltage set point to the generator output voltage mea-
The total electrical torque that opposes the mechanical surement, and the comparison produces an error signal that
power input the synchronous machine produces is the sum of adjusts the timing of the firing of the silicon-controlled rectifi-
two synchronous and damping torques and is equal to (18): ers until the output voltage Vt equals the voltage set point. In
Te = Tsync + Tdamp = K1 + K D (18) steady state, the generator output voltage is therefore equal to
the voltage set point.
The synchronous torque is proportional to the machine in-
ternal angle variation in (19):
SCR Bridge
Tsync = K 1 (19) Field PPT
Circuit
The damping torque is proportional to the machine speed Terminal
Voltage Vt
variation in (20):
Tdamp = K D (20) PT
the generator internal voltage during a short circuit at the ge- C. The Underexcitation Limiter
nerator terminals. The purpose of an underexcitation limiter (UEL) is to pre-
Fig. 13 represents a simplified model of an AVR by a gain vent a generator from being operated in steady state in a de-
KA, again with a time constant. termined underexcited region.
VUEL V*UEL V*UEL VOEL Consider the UEL model (type UEL2) as in Fig. 15. This
VS* model comes from the recommended models in [11]. We can
VIMAX VAMAX (Et VRMAX KC IFD) determine the UEL static or steady-state characteristic by set-
VS
ting the Laplacian operator s to zero and by looking at the
VI HV (1 + sTC)(1 + sTC1) KA VA HV LV
Vref
+
Gate (1 + sTB)(1 + sTB1) 1 + sTA
+
Gate Gate
EFD condition when the error signal from the UEL circuit will be
VC
VIMIN VAMIN Et VRMIN
zero [1]. Equation (22) provides this condition:
VF KF P Et k1 KUP Etk 2 KUV Q Et k1 KUQ = 0 (22)
1 + sTF
V*UEL, V*OEL: alternate inputs KLR + IFD Expressing Q as a function of P, we obtain (23):
0 KUP KUV
ILR
Q=P Et( k1+ k 2) (23)
KUQ KUQ
Fig. 12. IEEE type ST1A excitation system Equation (23) describes a straight line as in Fig. 16 and
represents the UEL characteristic in the P-Q plane. When the
Vref KA
+ EFD generator operating point falls below the line segment, the
1 + sTA UEL produces a positive error that the system supplies to the
AVR summing point. This positive error, in turn, has the ef-
Terminal Voltage, Vc
fect of increasing the voltage setting or AVR voltage reference
Fig. 13. Simplified representation of a static excitation system
so that the generator terminal voltage increases until the gen-
erator operating point goes above the UEL limit straight-line
B. The Power System Stabilizer characteristic [3], [5].
One negative effect of an AVR installed on a generator is
QxF1 KUQ
that it decreases the internal damping torque when its gain KA Q
1 + sTUQ
is increased (the synchronizing torque is, however, increased).
Because of this, the regulator gain must be limited to some
value between 15 and 25 in most situations [4], and this limi-
F2 KUV VUEL to AVR
tation then reduces the dynamic stability of the generator. Two F2 = (Et)k2 Summing
1 + sTUV
solutions exist for this AVR gain limitation: 1) limit the AVR et F2 = (Et)k2 Point
gain or 2) supplement the AVR with a power system stabilizer
or PSS.
The PSS basic principle, as Fig. 15 shows, consists of mea- P
PxF1 KUV
suring the change with respect to synchronous speed and send- 1 + sTUV
ing a signal derived from this speed variation to the summing Fig. 15. Example of a UEL2 type straight-line underexcitation limiter
point of the AVR [2]. The net effect of the PSS action is to
increase the generator damping torque in both steady and tran- Q
sient state. Another result of using the PSS is being able to
increase the AVR gain KA without affecting the overall gene- KUV
KUP
rator stability. P
Vref
+
KUV (k1+k2)
Vt
Vfmax KUQ KUP
slope =
KUQ
Terminal
KA
Vf
Voltage Vc 1 + sTA
+
Vfmin
Fig. 16. UEL2 type straight-line characteristic
Phase
Gain Washout Compensation Reference [11] describes two additional UEL characteris-
sTw 1 + sT1
tics, one circular (type UEL1) and one multisegment straight
R KPSS
1 + sTw 1 + sT2
line (type UEL3), that work on the same principles as type
UEL2.
Fig. 14. Principle of a power system stabilizer
6
Keep in mind that, while the AVR and the PSS will im-
prove the dynamic stability of a generator, the purpose of a j (XdXd) Id
UEL is to help prevent the generator from undergoing steady- j xd It Eq=E Eq
Iq
state instability resulting from its operation in the underex-
cited region. Ei
D. High-Speed, High-Ceiling Voltage Excitation Systems Vt j xq Iq=j xq Iq
Id
A network disturbance resulting from a fault close to a ge- It
nerator reduces the generators terminal voltage. If the genera- Es j xd Id
tor excitation system is operated in manual mode, its internal j Xe It
voltage will decrease according to (15).
If the generator excitation system is under the control of an Fig. 18. Vector diagram of network in Fig. 17
AVR, the voltage the AVR imposes on the field winding dur-
Equation (24) allows us to determine the maximum power
ing the fault will depend upon the AVR speed, gain, and ceil-
that we can transfer from the generator before reaching
ing voltage. Consequently, the amount of boost that the gene-
steady-state instability. We can also plot the stability limit in
rator internal voltage will receive depends upon these three
the P-Q plane. With manual operation, and if we assume that
factors.
saliency has been removed, the classical steady-state stability
It is generally recognized that a high-speed, high-gain, and
limit (SSSL) is a circle with center and radius as in Fig. 19 [3],
high-ceiling AVR supplemented with a PSS is presently one
[5].
of the best means to improve generator transient stability [2].
Q (pu)
III. REVIEW OF CLASSICAL STEADY STATE
AND TRANSIENT STABILITY METHODS Vt2 1 1
Center = j
A. Steady-State Stability 2 Xe Xd
3
For this condition of operation, we can compute the gene-
Vt=1.0
Xe=0.4
Xd=1.6 Xq=1.55 Xd'=0.32
Te=0.05 H=3 T'do=6
rator internal and infinite bus voltages in Fig. 18.
Manual SSSL
sient state corresponding to (25) and (26). In transient state,
we can apply the equal-area criterion on the two transient state
0
curves. One can see that the two curves in transient state exhi-
bit a much higher peak value, so that we can expect a better
transient stability if we were to use the curve in steady state.
-1 2.5
Fig. 20. Various steady-state stability limits depending upon the nature of 1.5
the exciter Power P (pu)
B. Transient Stability
Transient stability is the ability of the power system to
1
maintain synchronism when subjected to a severe transient
disturbance such as a fault on a transmission line, loss of gen- Steady-State
eration, or loss of a large load [2].
Modern techniques allow us to use such tools as Transient 0.5
Stability Programs or EMTP to study power network transient
stability. These two program types introduce extensive models
of generators. Classical methods to study transient stability
have used the equal-area criterion and have necessitated a 0
We can obtain a formula for the power transfer equation in ANGLE (Degrees)
Xd Vt Xtr Ze X
Electrical
ZT Ze Center
Ei Es0 d = 180
ZT
Ej >
Xtr Es
Fig. 22. Generator classical model Ei >
P Es
B. Basic Theory of Swing Impedance
R
For the network model corresponding to Fig. 22, the clas-
Xd Ei >
sical theory of swing impedance allows us to determine the Es
swing impedance characteristics in Fig. 23. The swing imped-
ance characteristic depends primarily on the ratio n of the two A
source magnitudes:
Ei Fig. 23. Loss-of-synchronism characteristic in the impedance plane
n= (30)
Es
2
When the two source magnitudes are equal (n = 1), the
swing trajectory is a straight line perpendicular to the total n=1.25
impedance ZT segment. The trajectory crosses the ZT segment 1.5
n=1.5
at its middle point, when the phase angle between the two
sources is 180. This point is called the swing center. 1
n=2
When ratio n is greater than one, the swing impedance is a
circle in the upper part of the impedance plane. When n is 0.5
B
smaller than one, the swing impedance trajectory will also be
X (pu)
Man
Gen VGT
TM
Man Ef
VGT
AVR
If
AVR = Auto
1
PSS
2
0
PSS = Y
Fig. 25. A high-level representation of the generator AVR and PSS control-
lers used in the case studies Fig. 27. Voltage and current profiles for a three-phase fault with autoreclos-
ing after 500 ms
We use a static excitation model as the AVR because of the
rapid response of this type of excitation system. The model The voltages and currents shown in Fig. 27 are the second-
does not include a governor model because of the slow re- ary voltages and currents; the CT and PT ratios are 400 and
sponse of the governor compared to the exciter (AVR). The 3000, respectively. By using the voltages and currents from
mechanical torque applied to the generator is kept constant Fig. 27, we can obtain the positive-sequence impedance and
(torque prefault = torque postfault). plot this in the impedance plane. Fig. 28 is a plot of the posi-
We obtained the power system used in this test from [2], tive-sequence impedance after the breaker is closed (faulted
Chapter 12, and show this system in Fig. 26. line switched back into service).
Line 1
Fig. 26. Representation of the model power system used for these case stu-
dies
Fig. 30. Response of the generator speed, active power, and reactive power
outputs before, during, and after the applied fault
Fig. 32. The voltage and current profile for the three-phase fault whereby
the faulted line is reclosed after 1 second (unstable swing case)
Fig. 35. A plot of the positive-sequence impedance magnitude for the case
when the generator becomes unstable
Fig. 40 is the response of the generator for the applied con- Fig. 41 shows the response of the AVR. The three-phase
ditions. Initially, the speed, active power, and reactive power fault on the line reduces the terminal voltage to near zero dur-
are at a constant level (prefault condition). During the fault, ing this fault, the AVR tries to compensate for the low termin-
the machine speed increases. This is to be expected because al voltage by boosting the generator internal voltage (Ei). Dur-
the amount of electrical active power the generator exports is ing the fault, the AVR reaches the voltage maximum limit (see
not the same as the amount of mechanical power it imports. Fig. 12).
Also, during the fault the active power export decreases and
the reactive power export increases, which agrees with the
theory. Once the fault clears, the machine speed begins to sta-
bilize and the machine begins to export more active power
(MWs) than it did during the prefault condition. This is be-
cause the machine has stored up kinetic energy in the rotor
during the fault, energy that is now also being exported to the
power system. Because the machine is exporting more elec-
trical power than it is receiving mechanical power, the rotor
speed begins to decrease as the generator uses up the stored
kinetic energy. Up to this point (27 cycles after fault clear- Fig. 41. Output of the AVR
ance (450 ms), the machine behaves almost identically for the
manually excited and the automatically excited case. Moving This boost in internal voltage helps maintain generator sta-
forward, however, differences begin to appear. In the manual- bility during the time the faulted line breaker is open, allowing
ly excited case, the generator active power export begins to the generator to export the same amount of active power as the
decrease, which results in increasing speed. This downward amount of mechanical power it imports. As mentioned before,
spiral continues until the machine slips a pole and becomes the current the generator exports for the first 27 cycles after
unstable. The instability occurs because the increased voltage the fault is cleared is almost similar (Fig. 42) for both cases,
drop inside the generator collapses the terminal voltage, pre- but after this point the manually excited generator current con-
venting the generator from exporting the electrical active tinues to increase because the machine is trying to export real
power that it is receiving in the form of mechanical energy. power. So, even though reactive power is linked to voltage,
When the AVR is enabled, it keeps the terminal voltage at a voltage is necessary to export active power from a generator.
more or less constant level after the fault is cleared. Thus, the boost in internal voltage resulting from the AVR
gives the generator the synchronizing torque to maintain syn-
chronism with the system.
Fig. 40. A plot of the generator speed, active power, and reactive power for
the case where the AVR is enabled Fig. 42. A comparison of the line current in the unfaulted lines for the case
with and without the AVR enabled
From Fig. 40, we can see that the generator does not settle
back to its prefault condition rapidly after the fault is cleared.
As mentioned before, the generator achieves its prefault state
after about 15 seconds.
14
C. Examples of Swings With AVR and Power System If we enlarge Fig. 44, we see that the impedance variation
Stabilizer (PSS) Enabled after each swing decreases more (Fig. 45) than for the case
For this third and final case, both the AVR and the power with only the AVR. The plot in Fig. 45 occurs over the same
system stabilizer (PSS) will be enabled, and we will subject time frame, 3 seconds or 180 cycles, as the one in Fig. 38.
the system to the same condition as before. Fig. 43 is a voltage From this plot, we can clearly see the effect of the PSS. The
and current plot for the applied fault and the reclose. PSS results in the impedance returning more rapidly to its
prefault state.
Fig. 43. A plot of the voltage and current with the AVR and PSS in service
The voltage and current plot shows that the generator re-
mains stable during the open interval time and when the Fig. 45. Enlarged plot of Fig. 44 to better show the trajectory of the positive-
sequence impedance after the line is reclosed
breaker recloses. If we now plot the positive-sequence imped-
ance (Fig. 43 and Fig. 44), we see that the swing is more If we examine the positive-sequence magnitude plot
damped and the impedance does not traverse a large section of (Fig. 46), we see that the rate of change of impedance |dZ|/dt,
the impedance plane, as was the case when only the AVR was is initially as large (for the first swing) as for the plot without
enabled. the PSS. However, we can see that this is no longer the case
for subsequent swings, and these oscillations are damped out
more rapidly than in the plot without the PSS. Therefore, we
can see that the PSS damps out the oscillations.
Fig. 46. A plot of the positive-sequence impedance magnitude for the case
where AVR and PSS are enabled
VI. CONCLUSIONS
1. For short-circuit studies, where only the fault current
magnitude is of concern, we can use a constant voltage
source instead of a complete generator model, reducing
the complexity of the system model.
2. Generators with manual excitation experience a decrease
of the generator internal voltage (Ei) following a system
disturbance, such as a fault, close to the generator termin-
als. This decrease in internal voltage significantly reduces
the generators synchronizing ability after the disturbance.
3. An AVR significantly improves generator steady-state
stability, provided that the gain of the AVR is limited and
that the AVR is not operating at its limits before the dis-
turbance. The AVR boosts the generator internal voltage
during a system disturbance; this boost increases the ge-
nerator synchronizing torque, allowing the generator to
Fig. 47. A plot of the machine speed, active power, and reactive power return to synchronism after the disturbance.
when the AVR and PSS are both enabled
4. An AVR will help increase the synchronizing torque of a
Fig. 48 shows the output of the PSS and the AVR for the generator but will reduce the damping torque. Because of
above condition. At fault inception, the machine initially loses this, the AVR gain will have to be limited in most cases.
speed and then gains speed. This initial drop in speed results 5. A PSS increases the damping torque of a generator and
in the PSS issuing a negative gain in the excitation. After this allows increasing the AVR gain without compromising
initial decrease in speed, the machine speed increases and the the generators dynamic stability. A PSS therefore im-
PSS and AVR both drive the AVR to its upper limit. Also proves the dynamic stability of the generator more than an
interesting to see is that the PSS causes the AVR to be driven AVR alone.
to the upper and lower limits more often than when only the 6. In transient stability studies, generators modeled with
AVR was enabled alone. This occurs because the PSS wants constant excitation will constitute the worst-case scenario
to drive the change in speed of the machine to zero. with respect to system stability following a disturbance.
AVRs and PSSs in the generators should substantially
improve system stability.
Fig. 50. Per unit conversion between excitation system and synchronous
machine
The following sketch illustrates how to move from the non- Now let us revisit some generator flux equations:
reciprocal system to the reciprocal system and vice versa. d = M f I f Ld id (38a)
(xd xd)Id [11] Standard for Requirements for Salient-Pole Synchronous Generators and
Generator/Motors for Hydraulic Turbine Applications, 1982. ANSI Std.
Ed E C50.121982.
Quadrature [12] R. Sandoval, A. Guzmn, H. J. Altuve, Dynamic simulations help
Iq Eq Eq Axis
improve generator protection, 33rd Annual Western Protective Relay
Conference, Spokane, WA, October 1719, 2006.
Ed E xqIq
xdId
xqIq Ei X. BIOGRAPHY
Vt xdI
rI xdId Normann Fischer received a Higher Diploma in Technology, with honors,
from Witwatersrand Technikon, Johannesburg in 1988, a B.S.E.E., with hon-
Id I ors, from the University of Cape Town in 1993, and an M.S.E.E. from the
University of Idaho in 2005. He joined Eskom as a protection technician in
Direct 1984 and was a senior design engineer in Eskoms Protection Design Depart-
Axis ment for three years. He then joined IST Energy as a senior design engineer in
1996. In 1999, he joined Schweitzer Engineering Laboratories, Inc. as a pow-
Fig. 52. Phasor diagram of a synchronous machine during the transient state er engineer in the Research and Development Division. He was a registered
professional engineer in South Africa and a member of the South Africa Insti-
IX. REFERENCES tute of Electrical Engineers.
[1] E. W. Kimbark, Power System Stability: Synchronous Machines, Dover
Publications, Inc., New York. Gabriel Benmouyal, P.E. received his B.A.S. in Electrical Engineering and
his M.A.S. in Control Engineering from Ecole Polytechnique, Universit de
[2] P. Kundur, Power System Stability and Control, New York: McGraw- Montral, Canada in 1968 and 1970, respectively. In 1969, he joined Hydro-
Hill, 1994. Qubec as an Instrumentation and Control Specialist. He worked on different
[3] G.Benmouyal, The Impact of Synchronous Generators Excitation projects in the field of substation control systems and dispatching centers. In
Supply on Protection and Relays, 34th Annual Western Protective Re- 1978, he joined IREQ, where his main field of activity has been the applica-
lay Conference, Spokane, WA, October 1719, 2007. tion of microprocessors and digital techniques for substation and generating-
[4] IEEE Recommended Practice for Excitation System Models for Power station control and protection systems. In 1997, he joined Schweitzer Engi-
neering Laboratories, Inc. in the position of Principal Research Engineer. He
system Stability Studies, IEEE Standard 421.51992.
is a registered professional engineer in the Province of Qubec, is an IEEE
[5] F. P. DeMello and C. Concordia, Concepts of synchronous machine Senior Member, and has served on the Power System Relaying Committee
stability as affected by excitation control, IEEE Trans. Power App. since May 1989. He holds over six patents and is the author or co-author of
Syst., vol. PAS88, No. 4, pp. 316329, April 1969. several papers in the field of signal processing and power networks protection
[6] IEEE Task Force on Excitation Limiters, Underexcitation Limiter and control.
Model for Power System Stability Studies, IEEE Trans. On Energy
Conversion, Vol. 10, No. 3, September 1995. Satish Samineni received his B.E. degree in electrical and electronics engi-
[7] D. Reimert, Protective Relaying for Power Generation Systems, Boca neering from Andhra University College of Engineering, Visakhapatnam,
Raton: CRC Press, 2006. India. He received his Masters degree in Electrical Engineering from the
University of Idaho, in 2003. Since 2003 he has been with Schweitzer Engi-
[8] IEEE Power Engineering Society, IEEE Tutorial on the Protection of neering Laboratories, Inc. in Pullman, Washington, where he presently is a
Synchronous Generators, 95 TP 102. Lead Power Engineer. His research interests include power electronics and
[9] Guide for AC Generator Protection, IEEE Standard C37.102/D7200X, drives, power system protection, synchrophasor-based control applications,
April 2006. and power system stability.
[10] Requirements for Cylindrical Rotor Synchronous Generators, 1989.
ANSI Std. C50.131989.