IEEE Transactions on Power Apparatus and Systems, Vol. PAS100, No. 2, February 1981
EXCITATION SYSTEM MODELS FOR POWER SYSTEM STABILITY STUDIES
IEEE COMMITTEE REPORT
ABSTRACT
Excitation system models suitable for use in large scale system stability studies are presented in this paper. With these models, most of the excitation systems currently in widespread use on large, system connected generators in North America can be represented.
In 1968, models for the systems in use at that time were presented by the Excitation System Subcommittee, and have since been widely used by the indusWhile such models are still adequate for many try. types of system stability studies, improved models of those systems are presented which reflect current knowledge and modelling practices. In addition, several new excitation systems are now in use which cannot be adequately represented by the older models. Models for these systems are developed in the paper, as well as models for some of the stpplementary excitation control features commonly used with them.
The models are valid for frequency deviations of 5% from rated frequency and oscillation frequencies up to about 3 Hz. However, the analysis of subsynchronous resonance and the shaft torsional spectrum is beyond the scope of these models. Care should be taken in using them outside these limits.
INTRODUCTION
When the behaviour of synchronous machines is to be accurately simulated in power system stability studies, it is essential that their excitation systems be modelled in sufficient detail(l]. The desired models must be suitable for representing the actual excitation equipment performance for large, severe disturbances as well as for small perturbations.
An earlier IEEE Committee Report on Excitation System models[2] has provided a reference for manufacturers, owners and system analysts since 1968. It established a common nomenclature, presented mathematical models for excitation systems then in common use, and defined parameters for those models. This report is an extension of that work. It provides models for new types of excitation equipment not covered previously as well as improved models for older equipment. To some extent, the model structures presented are intended to facilitate the use of field test data as a means of obtaining model parameters. Although the earlier report contained typical model parameters, this report does not. The wide variety of parameters applicable to some of the models makes the definition of such typical data difficult.
Three distinctive types of excitation systems are identified on the basis of excitation power source:
*
Type DC Excitation Systems which utilize a direct current generator with a commutator as the source of excitation system power. Type AC Excitation Systems which use an alternator and either stationary or rotating rectifiers to produce the direct current needed for the generator field.
Type
ST
Excitation
Systems
in
which
excitation
tifiers.
VREF
F 80 2584 A paper recommended and approved by the EEE Power Generation Committee of the IEEE Power ingineering Society for presentation at the IEEE PES W;inter Meeting, New York, NY, February 38, 1980. Manuscript submitted November 5, 1979; made available for printing December 28, 1979.
) 1981 IEEE
495
In addition, the following key accessory functions common to all excitation systems are identified and described:
* Voltage Sensing and Load Compensation * Power System Stabilizer
The terms "Excitation System Stabilizer" and "Transient Gain Reduction" are used to describe circuits in several of the models which affect the stability and response of those systems. The meaning of the terms and a description of some of the ways in which parameters of such circuits affect excitation system performance are discussed in Appendix D.
GENERATOR TERMINAL VOLTAGE TRANSDUCER LOAD COMPENSATOR MODELS
VR EF
Fig. 2.
ate voltage at a point part way through the stepup transformer. For these cases, Rc and Xc would take on the appropriate negative values.
A block diagram of the terminal voltage transducer and the load compensator is shown in Fig. 2. These model elements are common to all excitation system models described in this document.
(RC
Xr
cuit and comparator. The terminal voltage of the generator is sensed, and usually reduced to a dc quantity. While the filtering associated with the voltage transducer may be complex, for modelling purposes it can usually be reduced to the single time constant shown. For many systems, this time constant is very small and provision should be made to set it to zero.
Some compensator circuits act to modify terminal voltage as a function of reactive and real power, instead of reactive and real components of current. Although the model provided will be equivalent to these circuits only near rated terminal voltage, more precise representation has not been deemed worthwhile.
These and other forms of compensation are described in detail in Reference [3].
The terminal voltage transducer output is compared with a reference which represents the desired terminal voltage setting. The equivalent voltage regulator reference signal, VREF, is calculated to satisfy the initial operating conditions. It will, therefore, take on a value unique to the generator load condition being studied. The resulting error signal is amplified as described in the appropriate excitation system model to provide the field voltage and subsequent terminal voltage to satisfy the steady state loop equations. Without load compensation, the excitation system, within its regulation characteristics, attempts to maintain a terminal voltage determined by the reference signal. When compensation is desired, the appropriate values of Rc and Xc are entered. The input variables of generator voltage and current must be in phasor Care must be form for the compensator calculation. taken to ensure that a consistent per unit system is utilized for the compensator parameters and the generator current base.
This type of compensation is normally used in one of the following two ways:
1.
When units are bussed together with no impedance between them, the compensator is used to create an artificial coupling impedance so that the units will share reactive power This corresponds to the appropriately. choice of a regulating point within the generator. For this case, Rc and Xc would have positive values. (This function can also be achieved in practice with interconnected compensators as described in Reference 3).
Allis Chalmers
Westinghouse
General Electric
Regulex regulator Amplidyne regulator GDA regulator MagAStat regulator Rototrol regulator Silverstat regulator
TRA regulator
VRMAX
V,
EFD
2.
When a single unit is connected through a significant impedance to the system, or when two or more units are connected through individual transformers, it may be desirable to regulate voltage at a point beyond the machine terminals. For example, it may be desirable to compensate for a portion of the transformer impedance and effectively regul
Fig. 3.
496
The principal input to this model is the terminal voltage error signal VERR from the Generator Terminal Voltage Transducer and Load Compensator Model described above.
vs
T RMAX
The stabilizing feedback VF is subtracted and the power system stabilizing signal V5 is added to VERR. In the, steadystate these last two signals are zero, leaving only the terminal voltage error signal. The resulting signal is amplified in the regulaThe 'major time constant, TA, and gain, KA, tor. associated with the voltage regulator are shown incorporating nonwindup limits typical of saturation or amplifier power. supply limitations. A discussion of windup and nonwindup limits is provided in Appendix These voltage regulators utilize power sources F. that are essentially unaffected'by brief transients on the generator or auxil'iaries buses.' The time constants TB and T0 may be used to model equivalent time constants inherent in the voltage regulator; but ,these time constants are frequently small enough to be neglectedand provision should be made for zero input data. The voltage regulator output, VR is used to control the exciter, which may be either separately or selfexcited as dis'cussed inAppendix A. When a selfexcited shunt field is used, KE represents the setting of the shunt field rheostat.
Most of these exciters utilize selfexcited shunt fields with the voltage regulator operating in a mode commonly t'ermed "buckboost". The ma'jority of station operators manually track the voltage regulator byperiodically trimming the rheostat set point so as to zero the voltage regulator 'output. This may be simulated by selecting 'the value of KE so that, initial conditions' are satisfied with VR of zero, as described in Appendix A.The term SE represents exciter saturation as described in Appendix C and its value is a function of exciter output voltage EFD. The signal VF derived from EFD is normally used to provide excitation system stabilization asdiscussed'in Appendix' D.
Type DC2  DC Commutator Exciter regulators having supplies obvoltage ously acting tained from the generator or auxi7iaries bus voltage. It differs from Type DCl only in the voltage regulator output limits which are now proportional to VT. It is representative of solid state, controlled rectifier replacements for various forms of older mechanical and rotating amplifier equipments.
Fig. 4.
The systems discussed in the previous section are representative of the first generation of highgain, fastacting excitation sour'ces. The Type DC3 system is used to represent' older systems, in particular those DC Commutator Exciters with noncontinuously acting regulators that were commonly used before the development of the' continuously, acting varieties. Some examples'of these systems are:
General Electric Westinghouse

The per unit system and the relationship between regulator limits and field voltage limits are developed in Appendix B. Type DC2 Excitation System Model
These systems respond at basically two different rates, depending upon 'the magnitude of voltage error. For small errors, adjustment is made periodically with a signal to a motoroperated rheostat. Larger errors cause resistors to be'quickly shorted or inserte'd and a strong forcing signal appl'ied to the exciter. Continuous motion of the motoroperated rheostat occurs for these' larger error signals even' though it is bypassed by contactor action. Fig. 5 illustrates this control action.
The exciter representation is similar to that_of previously described systems. Note that no excitation system stabilizer is represented. V RMAX
AXK
The model shown in Fig. 4 is used to represent Field Controlled DC Commutator Erciters with continuK v
ER
r C
<K}
V
rFVR
VRMIIN
VRH
II
sK
Iv
VRMI N
II
IF IVERR
IF
VR= VRH
VR IVRMIN
RMAX
EFD
Fig. 5.
497
VERR
<
EFD
I
sT
+S
V VMIN
sKF i@sTF
+
Type ACl, Alternat:orRectifier Excitation System with NonCo ntrolled Rectifiers and Feedback from Exci iter Field Current acteristic in the exciter output imposes a lower limit Depending upon the magnitude of voltage error, of zero on the exciter output voltage as shown in Fig. VERR, different regulator modes come into play. If 6. This model is applicable for'simulating the perthe voltage error is larger than the fast raise/lower formance of Westinghouse brushless excitation systems. contact setting, KV, (typically five per cent), VR MAX or VR MIN is applied to the exciter, depenFor large power system stability studies, the exding upon'the sign of the voltage error. For an absolute value of voltage error less than KV, the exciciter alternator synchronous machine can be represented by the simplified model shown here. The demagnetter input equals the rheostat setting VRH. The rheostat setting is notched up or down, depending upon izing effect of load current (IFD) on the dynamics the sign of the error. The travel time representing of the exciter al'ternator output voltage (VE) is accounted for in the feedback path which includes the continuous motion of the rheostat drive motor is constant KD. This constant is a function of the exTRH. A nonwindup limit is shown around this block, to rep,resent the fact that when the rheostat reaches citer'alternator synchronous and transient reactances either limit, it is ready to come off the limit immed[4,5]. iately when the input signal reverses. Additional refinements, such as dead band for small errors, 'have Exciter output voltage drop due to rectifier regulation is simulated by inclusion of the constant been considered, but were not deemed justified for the relatively few, older machines using these voltage reKC (which is a function of commutating reactance) gulators. and the approximation to the rectifier regulation curve FEX, as described in Appendix E. The model assumes that the quick raiselower In the model, a signal VFE proportional to exlimits are the same as the rheostat limits. It does citer field current is derived from 'the summation of not'account for time constant changes in the exciter field as a result of changes in field resistance. The signals from exciter 'output voltage VE (multiplied field resistanc,e changes 'with rheostat movement and by the term KE + SE. where SE represents saturawith operation of quick action contacts. tion as described in Appendix C) andIFD (multiplied by the demagnetization term Kn). The exciter field current signal VFE is' used as the input to the exciTYPE AC  ALTERNATOR SUPPLIED RECTIFIER tation system stabilizer'in the Type ACI model. EXCITATION SYSTEMS
Fig. 6.
These excitation systems use an ac alternator and either stationary or rotating rectifiers to produce the direct current needed for the generator field. Loading effects on such exciters are significant and the use of generator field current as an input to these model allows their effects 'to be accurately represented. While few of these systems (Type AC4) allow negative field forcing, most do not supply' negaFor studies involving induced tive field current. negative field current (eg asynchronous operation) more detailed modelling is required.
Type AC1 Excitation System Model The model shown in Fig. 6 represents the Field Controlled Alternator Rectifier Excitation System designated Type AC1. This excitation system consists of an alternator ma'in exciter with noncontrolled rectiThe exciter does not employ selfexcitation fiers. and the voltage regulator power is taken from a source not affected by external transients. The diode char
AC2, represents a High Initial Response Field Controlled AlternatorRectifier Excitation System. The alternator m,ain exciter is used with noncontrolled rectifiers. The Type AC2 model is similar to that of
Type AC1 except for the inclusion of two additional exciter field current feedback loops simulating exciter time constant compensation and exciter field curThis model is rent limiting elements respectively. applicable for simulating the performance of Westinghouse High Initial Response Brushless excitation systems.
The exciter time constant compensation consists of a direct negative feedback (v.) around the exciter field time constant reducing its effective value and thereby increasing the bandwidth of the excitation system small signal response. The KB KHj time constant is reduced by the gain (1
essentially
498
Fig. 7.
Type AC2, High Initial Response AlternatorRectifier Excitation System with NonControlled Rectifiers and Feedback from Exciter Field Current
tion from regulator control to limiter control of excitation at the limit point. As explained by the description of the LV gate function in the nomenclature, excitation is controlled by the more negative of the two control signals.
of the compensation loop and is normally more than an order of magnitude lower than the time constant without compensation.
To obtain high initial response with this system very high forcing voltage (VR MAX) is applied to the exciter field. A limiter sensing exciter field current serves to allow high forcing but limits the current. By limiting the exciter field current, exciter output voltage (VE) is limited to a selected value (VLR) which is usually determined by the specified excitation system response ratio. The output signals from the voltage regulator (VA) and time constant compensation (VH) elements are compared with the output signal (VL) from the limiter in control logic circuitry which functions to provide a sharp transi
The model shown in Fig. 8, represents the Field Alternator Rectifier Excitation System designated Type AC3. This excitation system includes an alternator main exciter with noncontrolled rectiThe exciter employs selfexcitation and the fiers. voltage regulator power is derived from the exciter output voltage. Therefore, this system has an additional nonlinearity, simulated by the use of a multi
V ERR
Fig. 8.
499
plier whose inputs are the voltage regulator command signal, VA, and the exciter output voltage, EFD, times KR. This model is applicable to systems such as the General Electric ALTERREX excitation  systems employing static voltage regulators. For large power system stability studies, the exciter alternator synchronous machine model is simplified. The demagnetizing effect of load current (IFD) on the dynamics of the exciter alternator output voltThe feedback path inage (yE) is accounted for. cludes the constant KD, which is a function of the exciter alternator synchronous and transient reactances.
TYPE ST

These excitation systems utilize transformers to convert voltage (and also current in compounded systems) to an appropriate level. Rectifiers, either controlled or noncontrolled, provide the necessary direct current for the generator field.
While many of these systems allow negative field voltage forcing, most do not supply negative field
current.
Exciter output voltage drop due to rectifier regulation is simulated by inclusion of the constant KC (which is a function of commutating reactance) and the approximation to the regulation curve FEX, as described in Appendix E.
In the model, a signal proportional to exciter field current VFE is derived from the summation of signals from exciter output voltage VE (multiplied by the term KE + SE, where SE represents saturation as described in Appendix C) and IFD (multiplied by the demagnetization term KD).
For many of the static systems, exciter ceiling For such systems there are additional field current limit circuits, not shown in these models, which protect the exciter and the generator rotor.
The excitation system stabilizer also has a nonlinear characteristic. The gain is KF with exciter output voltage less than EFDN. When exciter output exceeds EFDN, the value of this gain becomes KN.
Type AC4 Excitation System Model
The Type AC4 Alternator Supplied Controlled Rectifier Excitation System illustrated in Fig. 9 is quite different from the other Type AC systems.
vs
Fig. 10.
VIMAX
(VRMAX KcIFD)
VERR
V
N
am
sTc II + #
sT.
K ]A/ _FD I
(VRM,IN  KCIIF)
Fig. 9.
This high initial response excitation system utilizes a full thyristor bridge in the exciter output circuit. The voltage regulator controls the firing of thyristor bridges. The exciter alternator uses an independent voltage regulator to control its output voltage to a constant value. These effects are not modelled, however transient loading effects on the exciter alternator are included. Exciter loading is confined to the region described as mode 1 in Appendix E and loading effects can be accounted for by using the exciter load current and commutating reactance to modify excitation limits. The excitation system stabilization is frequently accomplished in thyrstor systems by a series laglead network rather than through rate feedback (Appendix D). The time constants TB and TC would be used to simulate this control function. The overall equivalent gain and the time constant associated with the regulator and/or firing of the thyristors would be simulated by KA and TA respectively.
Systems utilizing this simulation model include the General Electric ALTHYREX and Rotating Thyristor Excitation systems.
500
vs
VRMAX
EFDMAX
IN
KC
VjI
=f (IN) FE EXX
Fig.
11.
Exciter
While for the majority of these excitation sysa fully controlled bridge is employed, the model is applicable to semicontrolled systems as well, in which case the negative ceiling is set to zero.
tems
are:
Canadian General Electric Silcomatic Exciters Solid State Canada Westinghouse Thyristor Exciter Westinghouse Type PS Static Excitation Systems with type WTA or WHS regulators
For most of these systems, the cosine characteristic relating thyristor firing angle to bridge output is compensated for by an inversecosine function between regulator output and firing angle. In such systems the gain KA is a constant and is independent of In a few systems, this inexciter supply voltage. versecosine compensation is not employed and the term KA becomes a cosine function, dependent on supply The STl model can be used to approximate voltage. these systems for some types of studies, but more accurate representation may be required for others.
+~~~~~~
ERR
XVE =KPVTj(K+
X XIT
IN
FI
cVE
EX
ON)
Source Controlled Rectifier Exciter
Fig. 12.
501
rent) to form the source of excitation power. Such Compound Source ControlledRectifier Excitation Systems employing controlled rectifiers in the exciter output circuit are designated Type ST3 and are represented by the model shown in Fig. 12.
The excitation system stabilizer for these systems is provided by a series laglead element, represented by the time constants TB and TC. An inner loop field voltage regulator is comprised of the gains KA and KG and the time constant TA. Rectifier loading and commutation effects are accounted for as discussed in Appendix E. The EFD MAX limit is established by the saturation level of power components. Systems of this type include the General Electric GENERREX and ShuntThyristor excitation systems.
REFERENCES
[3]
[4]
[5]
[6]
Fig. 13 shows the generalized form of such a Power System Stabilizer. Some common stabilizer input signals VSI are: accelerating power, speed, frequency and terminal voltage.
[7]
[8]
some input signals is incorporated by constants, A1 through A6. This, however, does not suggest the use of results from these models beyond 3 Hz. The filter representation is included only in order to properly account for gain and phase contributions in the frequency range up to 3 Hz. Programs should allow for zero input of parameters A1 to A6 (or for bypassing of this block).
Provision
for
modelling
of
high
frequency
[9]
Electric Power Systems', New York IEEE Press, 1974. "Computer Representation of Excitation Systems", IEEE Committee Report, IEEE Transactions on Power Apparatus and Systems, Vol PAS87, June, 1968, pp 14601464. A.S. Rubenstein, W.W. Walkey, "Control of Reactive KVA with Modern Amplidyne Voltage Regulators", AIEE Transactions, Part III, Power Apparatus and Systems, 1957, pp 961970. R.W. H. Ferguson, Herbst, R.W. Miller, "Analytical Studies of the Brushless Excitation System", AIEE Transactions, Part III, Vol 79, pp 18151821, February, 1960. "Transfer Characteristics H.W. of Gayek, Brushless Aircraft Generator Systems", IEEE Transactions on Aerospace, Vol 2, No 2, pp 913928, April, 1964. "IEEE Guide for Identification, Testing and Evaluation of the Dynamic Performance of Excitation Control Systems", IEEE Std. 421A1978, IEEE, New York, NY. "Criterion and Definitions for Excitation Systems for Synchronous Machines", IEEE New York, NY, IEEE Standard 42172, December, 1972. "Practices and Requirements for SemiConductor Power Rectifiers", ANSI Standard C34.21968 (R1973), ANSI, New York, NY. R.L. Witzke, J.V. Kresser and J.K. Dillard, "Influence of AC Reactance on Voltage Regulation of 6  Phase Rectifiers", AIEE Transactions, Vol 72, pp 244253, July, 1953.
[10] "Excitation System Dynamic Characteristics", IEEE Committee Report, IEEE Transactions on Power Apparatus and Systems, Vol. PAS92, pp 6475, January/February, 1973.
The next two blocks allow two stages of leadlag compensation. Stabilizer gain is set by the term Ks and signal washout is set by the time constant T5.
Stabilizer output can be limited in various ways, not all of which are shown in Fig. 13. This model shows simple output limits. For some systems the
stabilizer output is removed if the generator terminal voltage deviates outside a chosen band (not shown). In other systems (not modelled here) the stabilizer output is limited as a function of generator terminal voltage.
NOMENCLATURE
Maximum
shown
explicitly in the nomenclature but are represented by the appropriate subscript (max or min)
on the variable.

and minimum
limits
The stabilizer output VS, is added to the terminal voltage error signal at the location shown in the various excitation system diagrams.
Vs
AlA6
EFD EFDN
PSS high frequency filter constants exciter output voltage f ield voltage)
(generator
A
VSMAX
 value of

changes
FEX
IFD
IN KA
VSMI N
Fig. 13.
KB
502
KC
KD
KE
to
VI
VLR
voltage
VLV
VN
KFsKN
gains
VR
VREF
KG
KH
KL
gain of exciter field current limit current circuit gain coeff icient first stage regulator gain gain of the exciter low voltage limit signal
Vs
VSI
KI
Kj
KLV
VT, IT
VT, IT
generator terminal voltage and current respectively generator terminal voltage and current respectively
Kp
KR
 phasor
constant associated with regulator and alternator field power supply system stabilizer gain
XC
KS
 power

XL
Op
KV
RC
fast
raise/lower
contact
setting
source
potential
 resistive component
compensation
of load
SE
TE
TF
TR
 defined as (A)(B)
A B
.p L
4C
Gate
TRH
T1, T3
T2, T4
 PSS lead compensating time constants  PSS lag compensating time constants
 PSS washout time constant

if
A>B, C = B
T5
A B
_ _P.H

Gt
VA VB
VC
A<B, C
VE
VERR
VF
VFE
APPENDIX A
VG
VH SVL
503
Base exciter resistance Rgb = Rg
Exb/Rgb
Fig. A. 1.
If =
Ex + Rgb SeEX
(A.5)
Es
If
Rf + Lf
dt
(A. 1)
Es
In the above equation, Lf represents the incremental inductance of the field circuit. The, exciter may be driven from the generator shaft or may have a separate drive; however, the effects of speed variation are considered negligible and constant speed is assumed in either case.
If gb
+ J
RKgb
dlf Tt
(A. 6)
Ex is a nonlinear function of If as shown by the exciter load saturation curve in Fig. A.2 and the relationship can be expressed as follows:
If =
Ex

SE
AIf
x
Rgb Se
dEx
(A. 7)
Rg
AIf
(A.2)
dt
dlf
dIf
dEX dt
(A.8)
(A.7) and (A.8) into
Es
Ex E+
[Rgb]
dE
(A. 9)
Let
LfMu Lf
=
Exo
(A. 10)
at
Fig. A.2.
As shown in Fig. A. 2, Rg is the slope of the exciter load saturation curve near Ex = 0. AIf represents the additional current required due to saturation and is related to the saturation function Se (in amps/volt) as follows:
gbj
SE
Ex
Lfu Rgb
dEx
dt
(A. 11)
AIf
SeEX
into (A.2),
(A.3)
Es(A.4)
per
0. E x
Rg
+ SeEX
In order to express the above equations in the following base quantities are defined:
unit,
Base exciter voltage Exb = exciter voltage which gives rated open circuit generator voltage air gap line
on
Fig. A. 3.
Separately Excited DC
Exciter
504
This can be reduced to the following form used in the models for Type DC1 and DC2 exciters:
_f
Rf
E
IEx
Ea

Fig. A.5.
Fig. A.4.
DC Exciter Block Diagram
SelfExcited DC Exciter
where
KE
Rf/Rgb
Assuming that Ea represents the voltage of an amplifier in series with the exciter shunt field or an equivalent series voltage from a multiwinding type exciter,
= Ex Esscx
TE
=
=
Lfu/Rgb
+ Ea
a
(A.17)
SO
VR Efd
Rf SE/b
=
in
block
diagram
form:
Es
Ex
It is seen that adjustment of Rf affects the feedback path including the saturation function SE. Where Rf /Rgb is not close to unity, the term SE should be used in place of SE in the appropriate mOdels. The constant TE in the forward loop remains unaffected by changes in Rf. One possible function is:
functional
form
of
the
saturation
Fig. A.6.
(A. 13)
KE
where A and B are constants describing the saturation function and are defined by specifying the value of SE for two points on the saturation curve. Another possible functional form of the saturation function is:
E SE
I
Rf/Rgb
Lfu/Rgb
RfSERgb
Ea
Ex
TE
SE
(A. 18)
VR 5
Efd =
BEX
xSE Ex
=
(A.14)
(A.ll5)
,BEx
A
e
and
considering
small
EFD(s)
VR( 5)
+
I/KE
SE
1 + sTE/KE
(A.16)
Since the selfexcited case is derived directly from the separatelyexcited case, it follows that the above representation adequately accounts for changes in exciter field resistance in this instance as well*. The effect of periodic adjustment of the rheostat set as to zero the voltage regulator output point (VR) is' to' change Rf and hence to change KE and the' saturation' function SE. The initial 'value of the saturation function (SEO) is related to the initial field voltage as given by equation (A.13) or (A.14). The value of KE required to produce an initial steadystate value of Ea (or VR) equal to
so
zero is:
The
effective time constant is TE/KE Lf uRf Both exciter gain and the effective time constant change with Rf
SelfExcited Exciter
excited dc exciter.
KE
SEO
(A. 19)
schematic diagram of
self
In the above analysis, the excitation system parameters have been normalized by choosing one per unit exciter field current as that which produces rated generator open circuit voltage on the linear portion of the loaded exciter saturation curve. Alternately,
equally
justifiable normalizations can be developed by the choice of different base parameters or values.
505
APPENDIX B.
PER UNIT SYSTEM
Generator currents and voltages in system studies represented by per .unit variables. They are generally derived using the nonreciprocal per unit system in which one per unit generator terminal voltage is defined to be rated voltage, and one per unit stator current is rated current, one per unit generator field current is that current required to produce rated generator terminal voltage on the generator air gap line[7], and one per unit generator field voltage is the corresponding field voltage.
are
Dif ferent computer programs have represented exciter saturation with different mathematical expressions. In general, the saturation function can.be defined adequately by two points. To be consistent, the procedure suggested is to establish two v.oltages at which to specify SE and then use these as data for The form of the saturation function computer input. is not defined here, but rather considered to be a part of the particular computer program used.
O~
ID
Excitation system models must interface with generator models at both the stator and field terminals. Signals which are summed with the per unit generator terminal voltage at the input to the voltage regulator must of necessity be compatible per unit, variables. The exciter output current must be in per unit on the generator field current base, and exciter output voltage must be in per unit on the generator field
>
LO
PK
voltage base.
In the past, several different bases have been used to normalize regulator output voltage. Similar excitation systems having essentially the same performance characteristics can have quite different parameters depending on the choice of the base for regulator output. The per unit system for the exciter described in Appendix ,A will have the advantage of maintaining the physical significance of the exciter time constant.
Fig. C. 1.
APPENDIX C
SATURATION FUNCTION
The exciter saturation function SE is defined as a multiplier of per unit exciter output voltage to in increase the e,xciter excitation represent Fig. C.1 requirements because of 'saturation.e, illustrates the calculation of a particular value of At a given exciter output voltage, the SE. quantities A, B, and C are defined as. the exciter excitation required to produce .,that output voltage on the constantresistanc.eload saturation, curve, on the air gap line, and on the n,o load saturation curve respectively. The constantresistanceload saturation curve is used in defining SE for DCcommutator exciters and SE is given by
SE1
SE2
'EFD2
Since saturation effects are most significant at higher voltages, the values of SE will be specified at or near the ceiling exciter voltage and at a lower value, commonly at 75% of that level.
In some cases, for example a selfexcited dc exciter, the ceiling voltage may not ,be precisely known In such. cases, SE1 corbecause i.t depends on KE. responds to a specified value of exciter voltage near its expected maximum value. In providing saturation data, 5E1'5E2' and the corresponding exciter voltages should be specified.
SE
AB
Note that when exciter field resistance is significantly different from exciter base resistance an
APPENDIX D
SE
should
be
used
as
Et ITATION CONTROL SYSTEM STABILIZATION AND TRANSIENT GAIN REDUCTION "Excitation Control System Stabilizer" is a term used to describe the principle exciter feedback circuit. The feedback signal, normally derived from generator field voltage 'or its equivalent (Fig. 3), is used to partially compensate for the exciter time constant and allow stable operation of the excitation control system with higher steady state regulator gain settings[1O].. This feedback is required for stable offline operation for many of the excitation systems which have significant time constants. While historically employed for off line stabilization, this stabilizer can be used to modify the on line performance of the generato'r.
SE
is given by
The noload saturation curve is used in defining for alternatorrectifier exciters and here SE
SE
CB B
The reason for using the noload saturation curve for alternatorrectifier exciters is because exciter regulation effects are accounted for by inclusion of synchronous reactance and commutating reactance voltage drops in the model.
506
With high initial response excitation systems, stabilizing is not normally required with the generator offline. For online operation it may be desirable to reduce the voltage regulator loop gain at high frequencies to minimize the negative contribution of the regulator to power system damping. This is particularly applicable where damping is not enhanced by the use of a power system stabilizer.
Mode III operation, the inherent delay angle (e) is fixed at 30' and the commutating angle (A) varies from 600 to 1200[9].
Such a "Transient Gain Reduction" can be represented in the models by suitable choice of laglead time constants in the forward path of the excitation regulator (TB and TC in Fig. 10) or equivalently by choice of suitable parameters (KF and TF) in the exciter stabilizing path.
APPENDIX E
The equations characterizing these three modes of operation define the rectifier regulation as a function of commutation voltage drop as rectifier load current is varied from no load to the short circuit value[9]. The rectifier regulation characteristics defined by these equations can be plotted as a smooth curve as shown in figure E. 1.
The quantities, EFD, IFD' VE and KC are all in per unit on the ac generator field base. For computer simulation purposes, the curve of figure E.1 is defined by three segments. The equations are derived in [11].
RECTIFIER REGULATION
All ac sources which supply rectif ier circuits, with either controlled or noncontrolled rectifiers, have an internal impedance which is predominantly inductive. The effects of this impedance are to alter the process of commutation of current between conducting rectifiers and produce a decrease in rectifier average output voltage as rectifier load current increases. The source reactance from phase to neutral
which opposes the transfer of current between rectifiers is defined as the commutating reactance[8].
1.0
0.8
Commutating reactance has the effect of preventing instantaneous transfer of current from one rectifier to another. Thus, commutation requires a finite time which is defined by the commutating or "overlap" angle (uL) and also, depending on the magnitude of load current produces an inherent delay defined by an angle (ar) (note that this angle is not the delay angle associated with controlled rectifiers) in the start of the commutation process. Both of these angles are expressed in electrical degrees[8].
Most rectifier circuits applied in excitation systems are six phase doubleway circuits, also referred to as three phase full wave bridge circuits [8]. It has been shown that this type of circuit may operate in one of three different modes as the load current of the rectifier is varied from no load to the short circuit value [9]. The mode of operation is dependent upon the product of commutating reactance and load current; i.e., commutation voltage drop.
0.6
EFD VE
0.4
0.2
0.2
0.4
0.6
0.8
1.0
In Mode I operation, the commutating angle (A) increases from 0 to 600 and the inherent delay angle (a) is zero with increasing load current. In Mode II operation, the commutating angle (Cu) is fixed at 60' but the start of commutation is delayed by the inherent delay angle (ca) which varies from 0 to 300. In
KC IFD !N  VE
Fig. E.1.
Rectifier Regulation Characteristic
Fig. E.2.
507
Discussion APPENDIX F REPRESENTATION OF LIMITS
The models described in the paper differentiate The functions between windup and nonwindup limits. of these two types of limits are illustrated in Figs. F.1 and F.2, which show the effects of the limits on the output of a block whose transfer function has a single time constant. It should be noted that in the case of a windup limit, the variable y is not limited. Therefore, the output variable x, when it hits a limit, cannot come off the limit until y comes within In the case of a nonwindup limit, the the limits. variable y is limited. To be at a limit y=A or y=B, With this implies input u>A or u< B respectively. limiter, the output comes off the limit as soon as the input u reenters the range within the limits defined by B<u<A.
u1
Fig. F. 1.
Windup Limiter
System Equations:
dy/dt
(uy)/T
H. H. Chen (Westinghouse Electric Corporation, East Pittsburgh, PA): The Committee is to be commended for its comprehensive compilation and reporting on the many new concepts of excitation system modeling. This report is especially appropriate during a period when greater emphasis is being placed throughout the industry on the performance evaluation of excitation systems through the use of digital simulations. Formerly used to model the brushless excitation system, the IEEE Type I Model lacks the representation of rectifier regulation, ac exciter armature reaction, and auctioneering logic. With the inclusion of these representations in the new IEEE Type ACI and AC2 Models, brushless excitation system operation under transient conditions can be more accurately simulated. In the IEEE Type ACI and AC2 models, the exciter saturation is defmed by the noload saturation function, and the exciter internal voltage VE is the noload voltage as determined by the saturation function. The brushless exciter output voltage, EFD, is simulated as the noload voltage reduced by the armature reaction KD IFD and the rectifier regulation, FEX. Since the exciter armature reaction and the rectifier regulation are a function of the generator field current, system disturbances causing changes in the field current will produce corresponding changes in the armature reaction and rectifier regulation. Due to these operational characteristics of the ACI and AC2 Models, the exciter output voltage will differ from that obtained if the Type 1 Model is used. The difference will depend on the contribution of the generating unit to the system disturbance, and a + 20% difference in the exciter output voltage is not unlikely, if each model was used in separate studies with the same disturbance. In our experience in modeling the ACI and AC2 excitation systems, we have found there is a potential difficulty when simulating the rectifier regulation. For the initialization of the simulation, values of IFD (0) and EFD (0) are known, while the unknown (Figure E.i) is VE. A suggested procedure is to define the slope of a straight line passing through the origin equal to the ratio of EFD (0) and IFD (0). The initial operating point is determined as the intersection of the straight line and the curve. With a value for the ratio of EFD/VE or (KC IFD)/VE, the value of VE is directly obtained. Subsequent calculations of EFD, given VE nad IFD, can be made in a straightforward manner by applying the equations of Figure E.2. The complexities incorporated in the Type ACI and AC2 Models may be of special interest to only certain prospective users or planners.
A
1. T b1
TR
KA
TA
=OSec 400
II
= 0.02 Sec
0.0 P.u.
Fig. F.2.
1.
System Equations:
f =
TF
set to 0
KF
= 0.03
1.0 Sec
1800 Rpm 8.2
7.4
*
(uy)/T
TR = 0 Sec KA = 400 TA= 0.02 Sec KE= 1. KF = .03 TF = 1.0 Sec TB = TC = 0.0 Sec KC= 0.2 KD= 0.38
3600 Rpm
VR Max VR Mn SE Max
=
If y=A and f>O, then dy/dt is If y=B and f<O, then dy/dt is otherwise dy/dt = f B < y < A
3600
set
to 0
*
Rpm
7.3
1800 Rpm
8.2 7.4 .12 .03
1.3
In the models considered in the paper, some of the integrator blocks also have limits. These are of the nonwindup type and their function is similar to that illustrated in Fig. F.2. In this case, however, the expression for the function f would correspond to that of an integrator block.
= = = = =
7.3
6.6
66
= = = .1
0.86
1.1
SE.75
.5
.5
.8
1.3
TE
=
SE.75
.03
.8
508
For those who wish to relate simulated variable values of the excitation system with measured values, the Type ACI and AC2 Models should clearly be used. Also, those who wish to evaluate the performance of one type of excitation system with another, or one response ratio with another, will fmd the increased accuracy of the new models desirable and to be of sufficient advantage to warrant the effort needed to modify existing computer codes. Furthermore, when conducting transient dynamic stability studies where high response excitation systems are implemented, the use of the new models will show improved unit stability performance than when the Type 1 Model is employed. In the majority of excitation system simulations, as in the analyses of units with different 0.5 response ratio excitation systems, or for large multimachine power system planning studies, the Type I Model is probably of sufficient accuracy. In anticipation of requests by utilities and analytical groups for typical brushless excitation system parameter values for the Type ACI Model, tabulations of the typical values for the Type ACI and Type 1 Models are shown in the accompanying table. It is not practical to tabulate typical values of parameters for the Type AC2 system because of the wide variation in the parameter values. Data for specific applications will have to be developed on an individual basis.
R. T. Byerly and F. W. Keay (Westinghouse Electric Corporation, East Pittsburgh, PA): The Committee Report correctly states in Appendix C that several different bases have historically been used to normalize regulator output. Model parameters in the regulator and exciter depend on the choice of this base, and simple comparisons of models employing different bases can be misleading. To emphasize this point, the two sets of parameters in Table A, though apparently quite different, are in fact identical descriptions of the same voltage regulator and exciter. The left hand column of Table A corresponds to the middle column of Table I in reference 2. The equivalence of these two sets of parameters can be verified using the voltage regulator and exciter equations for the type DCI model of Figure 3. For simplicity, ignore the stabilizer signal and assume that TB and TC are zero. Then
KA
TA VRMAX VRMIN TE
SEI
SE2
KE
EFDMAX
Normalized Model Parameters 400.0 114.3 0.050 0.050 3.50 1.00 3.50 1.00 0.950 0.271 0.220 0.063 0.950 0.271 0.170 0.049 4.50 4.50
It should also be noted that the models for rotating d.c. exciters with selfexcitation are incrementally unstable for sufficiently small values of EFD. This may be significant in adapting them for use in linear analysis. Possible problems can be avoided by selecting KE in accordance with equation (A. 14) of Appendix A.
Manuscript received February 20, 1980.
D. B. Seely and M. L Spence (U.S. Army Corps of Engineers, Portland, OR): It has been evident to those involved in system studies that as the studies became more sophisticated, adequacy of some of the presently used excitation models was questionable. It appears that the committeeproposed models go a long way towards addressing these inadequacies. It is unfortunate that underexcitation limiters and other similar excitation system "accessories" were not addressed in the report, since these components are frequently part of the modeling problem. Nonetheless, the report is a commendable undertaking and undoubtedly some topics left untouched in this report will be addressed by the industry, as the need becomes more apparent. We would be interested if there are plans to develop guides to verify rapidly through field measurements, the necessary parameters for the models utilized to represent inservice excitation systems. The discussor's own organization has a number of excitation systems currently represented by calculated data. Our estimate of the manpower effort to verify model data of these systems with techniques described in current industry guides is discouragingly high. We suspect that excitation models and field verification guides may go through several more evolutions as more sophisticated test equipment becomes available in conjunction with "high power" analytical techniques.
(1 + STA) VR
"'A
(VERR
VF)
(1)
R (2) (SE + K + STE) EFD Assume that these are normalized equations utilizing a base for the voltage regulator output corresponding to the left hand column of parameters in Table A. The regulator output base for parameters in the right hand column is 3.5 times the base applicable to the left hand column since VRMjAX and VRMIN, which represent the same physical variable in both columns, differ by that ratio. Normalized equations corresponding to the right hand column can be obtained by dividing VR in equations (1) and (2) by 3.5, observing that bases for VERR, VF, and EFD are not to be altered. Thus
(1 + STA) (VR/3.5)
=
(KA/3.5)
(VLM
 VF)
(3)
manufactured.
(VR/3.5)
[(SE
+ KE +
STE)/3.51 EED
(4)
changing the voltage regulator output base changes model parameters as indicated, and the equivalence of the two data sets is apparent. Note that the parameters of the right hand column of Table A are identical to those in the left hand column of Table I of reference 2. Equation (2) indicates that there is a unique regulator output base which, together with the conventional generator field voltage base, will preserve the real physical parameters of the exciter in the normalized model. Any other regulator output base will result in model parameters different from those of the real exciter. The excitation system models corresponding to the first and second columns of Table I in reference 2 have essentially the same dynamic performance characteristics, even though the model parameters are quite different.
TABLE A
Equivalent sets of model parameters for a particular voltage regulator and exciter. Differences in parameters are attributable to different voltage regulator output bases. Values of saturation are at seventyfive percent of ceiling and at ceiling for the specified value of KE.
509
For a selfexcited DC exciter, the relationship shown in equation (A. 18) will then become: KE = 1.0 R gb/Rf
a
J
Efd = Ex
(A. 18)
KI X
T6
Note that the constant TE is equal to Lfu divided by Rf which is the physical time constant for the exciter. Note also that the gain Rgb/ Rf must be included with the regulator parameters. Unless this gain is provided for otherwise, the adjustment of the field resistance Rf will effect the constant TE, the regulator gain and the regulator limits. The method of normalizing the data presented in this discussion has the advantage of maintaining the physical significance of the exciter parameters but at the sacrifice of losing the physical significance of the regulator parameters as the conversion constant Rgb/Rf is not provided for directly in most stability program models.
Manuscript received February 28, 1980.
REFERENCE
[1]
E. J. Warchol, F. R. Schleif, W. B. Gish, and J. R. Church, "Alinement and Modeling of Hanford Excitation Control," IEEE Transactions on Power Apparatus and Systems, vol. PAS90, No. 2, MarchApril 1971, pp. 714724.
Manuscriptreceived February 25, 1980.
K. C. Bess (Western Systems Coordinating Council, Salt Lake City, Utah): Appendix B of the Committee Report states in the last sentence "The per unit system for the exciter described in Appendix A will have the advantage of maintaining the physical significance of the exciter time constant." This statement is incorrect. In order to make this a correct statement both sides of equation (A. 1 1) should be multiplied by Rgb/Rf:
R gb
M. L. Crenshaw: The interest expressed by the discussors is gratifying and contributes greatly to the usefulness of this report. As the Working Group developed these new, more complex models, it was apparent that neither existing computer programs nor data banks would be converted immediately. Thus a different identification system was adopted from that used in original report [2]. For many studies, the models in reference [2] and the corresponding data will suffice. Even the + 20% difference between old Type I and new Type ACl and AC2 models quoted by Mr. Chen, will in most cases be satisfactory. The Working Group could not reach a consensus regarding the inclusion of "typical" data for the various excitation systems. Some challenged the concept of such a tabulation in view of normal design variations and on going changes in design criteria. Other members felt that derived data from field measurements should ultimately be included in data banks. The comment by Mr. Bess is indeed correct, when TE represents the time constant of the entire exciter field circuit. The situation is further confused by the exciter designer, who most frequently defines the exciter time constant as the ratio of field inductance to field resistance, excluding external resistance. Messrs. Schleif and Schurz present an especially useful derivation for relating model parameters to measurable characteristics at various operating points. They correctly observe that the simpler expression, KF, adopted by the Working Group presents a compromise. Greater clarity could have been obtained from a more control design oriented term, KDTF, as the feedback numerator Messrs. Byerly and Keay address a problem that has troubled analysts for some time, essentially one of rapid data verification. Choosing base quantities as outlined in Appendix A will at least allow this checking of parameters on a consistent basis. Messrs. Seely and Spence address some rather key issues. The general consensus of the Working Group was to restrict the models to the voltage regulating and power system stabilizer control functions. Long term disturbances and severe offfrequency operation do indeed require either special models or adjusted data for satisfactory answers to be ob tained. Their comments on manpower required for field testing of excitation systems are valid. The impact of control system performance on overall utility system design and performance must provide the economic justification for such efforts. It is the intent of this Working Group to assess user experience with these new models after several years. A decision will then be made as to the development of an IEEE Standard which would receive periodic review for additions and modifications. There is strong interest in IEC for the development of similar models on a broader base.
Manuscript received October 30, 1980.
term.
Lfu
sEX+SEEX
dfEdt
Rf
Rf
(Al)1
For a separately excited DC exciter, the relationship between the parameters in Figure A.4 and the parameters in equation (A. 11) as given in equation (A. 12) will then become:
KE
TE
=
1.0
Lfu/Rf
SE
R
(A. 12)
/R
E fd
Ex