Influence of Governors On Power System Transient Stability
Influence of Governors On Power System Transient Stability
Influence of Governors On Power System Transient Stability
J. L. Dineley, M.Sc.Tech., Associate Member, and M. W. Kennedy, B.Sc, Graduate Synopsis The paper describes an investigation into the contribution that a conventional velocity governor controlling the input power to a synchronous generator makes to the transient stability of a typical power-system configuration. It seeks to ascertain the corresponding transient-stability effect using an alternative operating signal, derived from rotor acceleration, for the control of input power. A governor actuated from a compound of velocity and acceleration signals is described, and its effect on transient stability is compared. The effects on this stability of varying some of the parameters of the system, the machine and the governor are described. The paper concludes with a brief study of the effect of various governors on the transient stability of a synchronous generator connected to a large system by a single faulted transmission line that is fitted with autoreclosing circuit breakers. List of symbols 8 = rotor angle, electrical deg / = system frequency, c/s H = inertia constant, kWs/kVA M = H/l80f Pe0 = initial electrical output, p.u. Pe = electrical output, p.u. Pj = mechanical power input, p.u. Kd = damping torque coefficient Gg = velocity governor gain Gag = acceleration governor gain T| = servomechanism time constant, sec T 2 governor-valve time constant, sec T3 = turbine time constant, sec t = time, sec tc = clearing time, sec Tc = critical clearing time, sec Xd machine direct-axis synchronous reactance, p.u. X'd = machine direct-axis transient reactance, p.u. Xq = machine quadrature-axis synchronous reactance, p.u. rd0 = field time constant, sec X, = transmission-line reactance, p.u. Rt transmission-line resistance, p.u. Xc transmission-line shunt reactance, p.u. Xti, Xt2 = transformer reactances, p.u. oi = angular velocity, p.u. p d/dt Ku K2 voltage gains associated with constituent parts of automatic-voltage-regulator main loop AT3, K4, K5 = voltage gains associated with automaticvoltage-regulator stabilising loops T4, T5, T6, T 7 = time constants associated with magnetic amplifier and control exciter T 8 = main-exciter time constant T T 9> io> T n = iimQ constants associated with stabilising loops V = comparator error voltage Vc = control-exciter open-circuit voltage Vf = main-exciter output voltage $ = angular separation between machine terminal voltage and current Paper 4346 P, first received 21st March and in revised form 6th August 1963 Mr. Dineley and Mr. Kennedy are with the Department of Electrical Engineering, University of Newcastle upon Tyne 98 Introduction For many years it has been realised that more sophisticated control of the input power to synchronous generators would materially contribute to the stability of large electric power systems.1 In more recent years the introduction of fast-acting automatic voltage regulators has increased the transient-stability capabilities of synchronous generators, and has thereby made a longer time available after a disturbance, during which the mechanical power regulator or governor may adjust the prime-mover input power. This has given the governor a more significant role in synchronous-power-system transient stability. Since the stability of power systems ultimately depends on the rapid and accurate matching of the input and output powers of each individual machine in all conditions of operation, it is important that the input power to a machine should be made to respond to the difference between input and output powers to reduce the difference to zero as soon as possible. The contribution that is made to transient stability by the control of input power has not been widely recognised, and forms the subject matter of the paper. Initially the study was designed to investigate the contribution that conventional velocity governors make to synchronous-machine stability, but on the results, the work was extended to the exploration of those aspects of governor design that can be modified to improve performance. The paper therefore seeks to indicate likely trends in power-system performance when various feedback signals are used in input-power control to improve transient stability. Attention was directed in the first instance to the governor rather than the automatic voltage regulator, since the effects of the governor on transient stability appear to be less well known. The studies were carried out on a combined analogue computer and network analyser that has been described elsewhere.2 2
2.1
Velocity governor
General
The governor that responds to changes in velocity is believed to have been invented by James Watt for use on early reciprocating steam engines; a schematic of a modern velocity-sensitive governor for use on steam turboalternators is shown in Fig. 1. Governors fitted to synchronous generators must usually meet an empirical requirement. The appropriate British PROC. IEE, Vol. Ill, No. 1, JANUARY 1964
Standard3 defines a governor gain of 25, since a speed change of 004p.u. must produce an input-power change of 1 Op.u. at full load. A general definition of velocity governor gain is thus
This specified sensitivity has been found necessary in order to protect turboalternators from dangerous overspeed when all the electrical load is suddenly removed. As shown in Fig. 1, the actuating device usually operates an oil servo-
double-circuit transmission line, as shown in Fig. 2. A salient-pole generator is represented in view of the general nature of the paper. This Figure shows the position at which the faults were applied whilst studying their effects on the transient stability of the controlled generator. Fault clearance is effected by the simultaneous 3-phase tripping of the circuit breakers at each end of the faulted line, without their subsequent reclosure, except as described in Section 7. In order to present in a concise form the large number of swing curves that were computed to establish stability trends, a stability boundary is defined9 such that, when a fault occurs
mechanism, which in turn operates the steam valve or valves that actually control the input to the turbine. In some published works dealing with the transient stability of synchronous generators^5 it has been assumed, in solving the equation of disturbed motion of the generator's rotating system, that the mechanical power input to the generator remains unchanged during the transient disturbance being investigated, i.e. in the equation
p. - pe = Mp2h (2)
and is disconnected by circuit-breaker operation, a longer clearing time or a higher initial or prefault load on the generator would cause instability and loss of synchronism, whilst shorter clearance time or lower initial load would
transformer synchronous machine
X i X. R. t "t KSWUv/WV
transformer To
i X
Pi remains constant. Other authors,6"8 however, have assumed a variable input power, and described a transfer function to represent the turbine and the governor. The same method has been adopted here, since a separate investigation has verified it and led to the conclusions that the transfer function used (see Appendix 11.2) is valid for power-system transient-stability studies. The point is referred to again in Section 6. Many large steam turboalternators now include reheating of steam during the turbine cycle. If this reheated steam is not governed, a very long time delay is introduced into the control of power to the generator. Therefore, for this study it has been necessary to assume that there is no reheat cycle or that the reheat cycle is governed exactly as the input power. The time delay in the input-power response to governor movement introduced by an ungoverned reheat cycle must render the governor contribution to transient stability almost negligible. 2.2 Study details The system chosen for study is that of a synchronous generator connected to an infinite power system through a
infinite system
i
Fig. 2 System studied
-X
| /vflflft;
y/WV
X-
enable the generator to remain in ultimate synchronism with the system. Fig. 3 shows such stability boundaries for the generator and system tested, without automatic voltage regulator or governor, with automatic voltage regulator and governor separately, and with both together, on the occurrence of a 3-phase-to-earth fault. For these curves the rotating system has been assumed to have an inertia constant of 5kWs/kVA rating, and the governor has been taken as velocity-actuated with a gain of 20. . These curves demonstrate that the automatic voltage regulator and governor both together and separately increase the generator stability above that for the ungoverned, unregulated machine. The automatic voltage regulator and excitation system considered is described in Appendix 11.3, 99
and, in order to compare different governor characteristics, this same representation of voltage regulator has been employed in all the studies described in the paper. Figs. 4 and 5 have been prepared to show the effect on stability of different machine inertia constants and different
1-5-
velocity governor has a significant effect up to a value of about 30, higher gain producing little improvement. These curves suggest that, when inertia constants are below about 4kWs/kVA, the velocity governor becomes less satisfactory, so far as its contribution to first-swing stability is concerned.
O-3r
10
'0-2
10-1
025 05 0-75 critical clearing time, sec Fig. 3
10
24
28
Effect of velocity governor gain on critical clearing time for two values of inertia
3-phase-to-earth fault. Peo = 1 - 2 3 p.u. a H = 5-0kWs/kVA b H = 3-0kWs/kVA
0-5
At fast clearing times of less than about 200ms, only a limited contribution to first-swing stability can be expected, even over a wide range of inertia constants and governor gains. The influence of inertia and governor gain can also be demonstrated by their effects on rotor transients. Fig. 6 relates the acceleration of the rotor from an initial load angle, the prime-mover input power, and the rotor angle when 2-phase-to-earth faults are applied to machines of differing inertias, differing governor gains and differing fault clearance times being used. 10 4 6 8 The reduced governor gain permits larger rotor-angle inertia constant, kWs/kVA excursions and longer periods of oscillation, whilst lower Fig. 4 inertia gives greater initial rotor acceleration. The initial Effect of inertia on critical clearing time, for various velocity governor rotor acceleration, of course, corresponds to the loss of load gains caused by fault application, and the point at which the fault 3-phase-to-earth fault. Pe0 = 1 -09 p.u. is cleared gives a similarly clearly defined rotor deceleration. a Gg = 40 b Gg - 20 In Fig. 6, curve d has been prepared for comparison with c Gg - 10 d Gg-A curves a, b and c to show the effect of a less severe 2-phase-toearth disturbance occurring on a machine with an initial load velocity governor gains. Fig. 4 relates the governor gain, of 0-73p.u. The longer permissible clearance time of 0-6sec critical clearing time and inertia constant for a prefault enables the governor to make a more significant contribution power level of 1-09 p.u. (see Appendix 11.4 for details of tofirst-swingstability. initial conditions), whilst Fig. 5 relates these quantities in a Fig. 6 also indicates that the velocity governor makes a different presentation for a prefault power level of 1 -23 p.u. contribution to the positive damping of the rotor oscillations, It will be seen from these figures that, when the machine and from these curves and Fig. 3 it is evident that, at lower inertia is low, the governor gain tends to be less significant initial power levels and in less severe disturbance conditions, than at higher values. With an inertia constant of between 3 the extended clearing times then possible would permit the and 4kWs/kVA, increase of governor gain makes only a velocity governor to make a more significant contribution to marginal contribution to stability as long as the gain exceeds transient stability. about 6, although Fig. 3 shows that the improvement over The most important contribution of a velocity governor an ungoverned machine is much more apparent. If the inertia to stability seems to lie in its ability to add to the positive constant is higher than about 4kWs/kVA, the gain of a damping of the rotor oscillations, rather than in the reduction 100 PROC. 1EE, Vol. Ill, No. 1, JANUARY 1964
of energy to the rotating system during the first quarter of a second or so of a disturbance. Stability after heavy faults seems to be decided in this first short period, and a sufficiently sensitive velocity governor gives a significant contribution to
disappears, so that a gain of 50 at usual values of inertia constant will probably render a governor susceptible to self induced oscillation unless special preventive measures are taken. Such special measures are suggested later in the paper.
20
30
Fig. 6
Rotor swing, input power and rotor acceleration with velocity governor
2-phase-to-earth fault Initial load p.u. 1-23 1-23 1-23 0-73 Clearance time sec 0-2 015 015 0-6 Governor gain, Gg 20 6 20 20 Inertia constant H 5 3-5 3-5 3-5
overall transient stability, as may be seen from Fig. 3. The conjoint effect of governor and regulator action is here seen to greatly exceed their separate effects. It was found that too high a gain of the velocity-governor control loop could lead to self induced oscillations, especially with low inertia constants and small viscous damping. Except at very low inertia constants, these oscillations never occurred with gains of 30 or less, and increasing the governor gain led to reduction of the positive damping contribution. At higher gains the positive damping inherent in the governor action
PROC. IEE, Vol. Ill, No. 1, JANUARY 1964
It was essential to keep the study free from the extraneous effects of self oscillation, and accordingly attention was mainly confined to the lower velocity-governor gains shown in the Figures.
3
3.1
Acceleration governor
General
On competition of the study of the widely used velocity governor, it was concluded that a more substantial 101
improvement in the transient performance of the synchronous generator might be produced by replacing the velocityactuating signal by one proportional to the rotor acceleration, since acceleration is directly related to the power difference Pl Pe on which stability ultimately depends. The effect of 3-phase-to-earth, 2-phase-to-earth, phase-to-phase and single-phase-to-earth faults on machines fitted with velocity and acceleration governors have been studied, but only the least severe and most severe have been reproduced in curves b and c in Figs. 7 and 8, which show, respectively,
the acceleration governor seems to be more effective in the maintenance of stability than the velocity governor, because of the greater initial acceleration and the more limited time during which governor action may be effective. It is evident that velocity-governor action is superior when long clearing times are possible, although these clearing times may not be permissible in practice. 3.2 Study details Fig. 9 relates the critical clearing time to the acceleration governor gain for a generator with an initial load of
0-5
1-5
10
0-25
30
l-5p.u., subjected to a 2-phase-to-earth fault, alternative inertia constants not being considered. The rate at which improvement in system transient performance increases with acceleration governor gain does not justify gains in excess of about 25. In conditions of high initial loads or severe faults, the acceleration governor gives appreciably improved performance as compared with the velocity governor, for it is immediately sensitive to the acceleration produced by the loss of load. In the limit, increasing gain produces little improvement, chiefly because of the time delays inherent in the governor prime-mover control loop. Fig. 10 relates the critical clearing time and the inertia
0 Fig. 8
10 . 0-5
the stability trends for velocity and acceleration governor restraint. Curves a each represent the effect of a combined governor response as described in Section 4, whilst curve d in Fig. 7 indicates the stability trend for the position governor described in Section 5. An inertia constant of 5kWs/kVA has been chosen for all these curves. A gain of 20 for a velocity governor at this value of inertia gives a measurable improvement in stability without any likelihood of self induced oscillations. Acceleration governor gain is defined in a similar way to velocity governor gain. An acceleration of 0-04p.u. (persisting for the appropriate period) will completely close the input, valve with an acceleration governor gain of 25. 1 p.u. acceleration is defined as unit change of speed in unit time. At high prefault power levels and severe fault conditions,
102
Fig. 10
10
constant for an initial power level of l-36p.u. for different types of governor characteristic; curves b and c describe the effects of acceleration and velocity governors, respectively. Curve a is for the combined governor described in Section 4. The acceleration governor shows less improvement in critical
PROC. IEE, Vol. Ill, No. 1, JANUARY 1964
clearing time than does the velocity governor at inertia constants greater than about 5kWs/kVA. For any given disturbance, the initial rotor acceleration will be inversely proportional to the inertia constant, so that for high inertias the acceleration governor will be less effective whilst the machine is inherently slower in response to disturbance, so that these effects tend to cancel. Fig. 11 relates rotor acceleration, prime-mover input and
100
significant contribution to transient stability than does the conventional velocity governor. 4
4.1
A combined governor
General
Curves b and c in Figs. 7 and 8 demonstrate that neither a velocity governor nor an acceleration governor as
time, sec
. 5 = 0-5-
time, sec
0-1 ^
d.
-0-05 o
"-0-05
-0-1
1-0
time, sec
20
30
Fig. 11
Rotor swing, input power and rotor acceleration with acceleration governor
2-phase-to-earth fault Clearance time, sec 0-35 0-25 0-35 Governor gain, Gag 10 3 10 Inertia constant H 5 5 3-5
rotor angle for a machine fitted with an acceleration governor, subjected to a 2-phase-to-earth fault, with an initial load of I 23 p.u. These are the same load iand fault conditions as for the velocity-governed machine whose swing curves are shown in Fig. 6, but the faults can be allowed to persist for longer periods, with less disturbance to the rotor. The acceleration governor is seen to give more heavily damped rotor oscillations and much reduced frequency of oscillation compared with the velocity governor, i.e. the acceleration governor has the effect of greatly increasing the apparent machine inertia. It is concluded that, except at long clearance times, the acceleration governor makes a more
PROC. IEE, Vol. Ill, No, 1, JANUARY 1964
described above has the best characteristics for all fault types, clearing times and initial loads, but suggest that a combination of their characteristics might be beneficial. There are many conceivable ways of effecting a combination, e.g. by making the greater signal of the two the operating signal or by arranging that initial load or duration of fault decides the signal selected, but in the event it was decided simply to add the two signals and investigate the effects of varying the gain of each. It was found that a combined governor offers the possibility of greater assistance, not only to transient stability but also to the elimination of self induced oscillations arising from
103
. 1-5
large gains in velocity governors, as discussed in Section 2.2, so that an acceleration signal has a substantial stabilising effect on a velocity governor that tends to self oscillation. 4.2 Study details Curves a in Figs. 7 and 8 show the improvement in generator transient performance produced by compound governing over that obtained by either a velocity or an acceleration governor. Fig. 12 shows the stability boundaries for a generator with an inertia constant of 5kWs/kVA when subjected to a 3-phase-to-earth fault. These curves show the stability effects of different relative amounts of the two control signals produced by varying their gains. Fig. 10 curve a shows the effects of inertia on stability for a machine with a compound governor of acceleration gain 10 and velocity gain 20, and permits comparison with the performance of purely velocity and purely acceleration governors. Fig. 13 relates rotor acceleration, prime-mover input and
0-5.
10
time, sec
2-0
30
time, sec Fig. 13 Rotor swing, input power and rotor acceleration with compound governor
2-phase-to-earth fault. PeQ = l-23p.u. Clearance time, sec 0-35 0-35 Governor gain
Gg Gag
20 60
Inertia constant H 5 5
104
rotor angle for a machine with the same initial load and subjected to the same fault conditions as Figs. 6 and 11. Damping is increased still further with the compound governor than with the simple acceleration governor. Fig 136 is especially interesting since, with this velocity-governor gain, self induced oscillations were apparent in the absence of acceleration feedback. Rotor-angle governor The rotor angle has been proposed as a possible inputpower controlling signal10 and has been investigated as part of this study. It was found to be much less satisfactory than velocity, as might be expected, since rotor displacement, the integral of velocity, introduces a further phase shift into the input-power control loop, giving instability problems even at lower levels of gain. Fig. Id gives some indication of the stability limit with a rotor-angle governor, compared with velocity and acceleration governors. This curve explains why the rotor-angle governor was not regarded as making any practicable contribution to transient stability. 6 Effects of alternative transfer function for governor loop Appendix 11.2 describes the transfer function for the governor used in the main study, and also postulates an alternative transfer function for representing the governor loop that takes account of the time delay inherent in the movement of the input valve and a longer delay in the oil servomechanism. These longer time constants seemed likely to reduce the governor contribution to transient stability, and their effect was therefore investigated. Fig. 14 includes a 5
transmission line. Fast automatic reclosure after fault tripping can often restore the synchronous tie and maintain generator stability. The stability boundary of a single-line system can be defined in the same way as in Section 2, if 'reclose' is substituted for 'clearing'. A typical single-line system was assumed, whose transmission-line length was half that shown in Fig. 2, so that the initial conditions defined in Appendix 11.4 can be retained. A 3-phase-fault clearing time of 0 1 sec and successful 3-phase reclosure is assumed to occur. Fig. 15 shows the
1-5
:10
05
1-0
Fig. 15
Effect of various compound governors on stability boundary for single-line system
3-phase-to-earth fault. H = 3-5kWs/kVA, tc = 0-1 sec a (Jff = 60, Gag = 5 6 C , = 40, Ga<l = 10
C Gg = 20, Gag = 20
d Velocity governor, Gg = 10
.1-5
stability boundaries for a machine subjected to a 3-phaseto-earth fault at the local end of a single tie line, with three alternative compound governors and, for comparison, a velocity-actuated governor of gain 10. The inertia constant is 3 5kWs/kVA for all the curves.
Conclusions
0-5
1-0
Fig. 14
Effect of alternative transfer-function representation of turbine and acceleration governor on stability boundary
H 5kWs/kVA a Acceleration governor, Gag 10, first transfer function b Acceleration governor, Gag 10, second transfer function
stability boundary for a 2-phase-to-earth fault occurring when the generator prime mover is controlled by this slower response governor and compares it with the governor transfer function assumed earlier in the paper. The governor is acuated by an acceleration signal, and it is seen that the trend is not altered by increased governor time constants, but the stability boundary is reduced. It should be noted that a pessimistic time constant has been used for the oil-servomechanism representation. 7 Single-line faults An acute transient-stability problem can arise when a generator is connected to a large power system by a single PROC. 1EE, Vol. Ill, No. 1, JANUARY 1964 7 P4
With the aid of an appropriate combined computer, the effect of conventional velocity governors on synchronousgenerator transient stability shows that the potential improvement of power-system stability by these governors is limited, but that substantial improvements will be effected by the inclusion of acceleration signals in governor-actuating mechanisms. These, signals may be used alone or combined with velocity signals, but the manifest advantages of the compound governor have been demonstrated. At practicable fault-clearing times, system stability will be improved and system design problems correspondingly relieved by employing the appropriate forms of compound governing outlined in the paper, with optimisation of the proportion of the component signals. In addition, improvement of the steady-state stability of high-gain velocity governors seems to be possible by the use of acceleration feedback techniques. It is believed that a purely acceleration-operated governor will prove less useful than a compound governor as some velocity signal must be used to control power input in conditions of slowly changing load. Load-rejection problems cannot be solved on the computer, which employs a reference frame of synchronous speed, but the improved performance obtained with acceleration and compound governors on severe faults suggests that loadrejection performance will be similarly improved. 105
Acknowledgments This work is part of a comprehensive programme of studies on power-system transient stability at the University of Newcastle upon Tyne. The authors would like to acknowledge the interest and encouragement of Prof. R. L. Russell and Mr. W. D. Horsley, and the very important work of Dr. W. D. Humpage on the combined computer. The Department of Scientific and Industrial Research, C. A. Parsons and Co. Ltd., and the International Research and Development Co. Ltd. are thanked for their financial support and advice. 10
1
2
Eqn. 3 is applicable for the study of Section 2, eqn. 4 for Section 3 and eqn. 5 for Section 4. (6) Eqn. 6 describes the alternative transfer function discussed in Section 6. Table 1 gives the details of the various governor time constants used throughout the study.
Table 1
GOVERNOR TIME CONSTANTS Equation
T2
References
HORSLEY, w. D.: 'The stability characteristics of alternators and of large interconnected systems', / . IEE, 1935, 77, p. 577
DINELEY, j . L., and HUMPAGE, w. D. : 'Aids in the calculation of
3
4 5 6
3 4 5 6 11.3
01 01 01 0-5
01
lator on the study-state and transient stability of a synchronous generator', ibid., 1958, 105 A, p. 420
CONCORDIA, c , CRARY, s. B., and PARKER, E. E. : 'Effect of prime
mover speed control characteristics on power-system frequency variations and time-line power swings', Trans Amer. Inst. Elect. Engrs, 1941,60, p. 559 7 ALDRED, A. s.: 'Electronic analogue computer simulation of multi-machine power-system networks', Proc. IEE, 1962, 109 A, p. 195 8 MILES, J. G.: 'Analysis of overall stability of multi-machine power systems', ibid., 1962, 109 A, p. 203
9 10 DINELEY, j . L., and KENNEDY, M. w.: 'Concept of synchronous
The automatic voltage regulator, though not forming the subject of the paper, was present during the study, unless otherwise stated. The transfer function is summarised below in terms of the gain and time constants of the component part: pr6)(l +pr7). K,V
cmi
of displacement governing in synchronous power systems', Proc. I.E.E. 1961, 108 C, p. 471
"
. pK5 . .
max
(')
11
11.1
Appendixes
Test generator and system principal details
8
Vfmax
.( 8 )
The principal details of the system studied, which remained unchanged throughout the study, are: Generator: Xd = 08p.u. X'd = 0-3p.u. Xq = 0-5p.u. rd0 = 5 sec
Transformers: Xtl = Xl2 = O l p . u . Double-circuit transmission lines: X, = 0-4p.u. R, = O l p . u . Xc = lOOp.u.
pK5
(1 +pr9)
vd
+ (1
+/>T |0 )
(1
(9)
y=
V V Y y m g
(10)
where Vd, Vm and Vg are the demanded terminal voltage, actual terminal voltage and stabilising regulator feedback voltage, respectively.
11.4 Initial conditions
All steady-state initial system variables were computed digitally. The rotor angle was chosen as the independent variable, and the corresponding prefault powers are shown in Table 2. Table 2
INITIAL CONDITIONS
'
P,o
P 8Gag
10 20 30 40 50 60
0-73
106