Engineering and Research Center

Bureau of Reclamation

October 1978
MS-230 (1-76)
Bureau of Reclamation TECHNICAL REPORT STANDARD TITLE PAGE

1
1
October 1978
AUTOMATIC GENERATION CONTROL - NOTES AND 6. PERFORMING ORGANIZATION CODE
OBSERVATIONS

7. AUTHOR^) 8. PERFORMING ORGANIZATION

R E P O R T NO.

W. B. Gish REC-ERC-78-6

9. P E R F O R M I N G O R G A N I Z A T I O N NAME A N D ADDRESS 1 0 . WORK U N I T N O .

Bureau of Reclamation 11. C O N T R A C T OR G R A N T NO.

Engineering and Research Center

-
Same

IS. S U P P L E M E N T A R Y N O T E S
0
I
14. S P O N S O R I N G A G E N C Y C O D E

-116. ABSTRACT

The concept of automatic generation cbntrol is often considered complex because of the
freedom each utility has i n choosing individual characteristics within a basic control
philosophy. This report endeavors to separate the basic philosophy from the individual
characteristics to allow a clearer understanding of the control philosophy. Discussions of
individual characteristics used by the Bureau of Reclamation are also presented.

-7 .
1 K E Y WORDS A N D D O C U M E N T A N A L Y S I S

0. DESCRIPTORS-/ +load-frequency control/ 'interconnected systems/

*power system operations/ 'power interchange/ algorithms/ automatic
control/ control systems/ governors/ power dispatching/ baseloads/
generating capacity/ hydroelectric power/ electric generators/ computer
applications
15. IDENTIFIERS-/ automatic generation control

09B 0902
-c . COSATI Field/Group
18. D I S T R I B U T I O N S T A T E M E N T
COWRR:
19. S E C U R I T Y C L A S S 2 1 . NO. O F PAGE
(THIS REPORT)
I4 v a i l a b l e from t h e N a t i o n a l T e c h n i c a l Information S e r v i c e , o p e r a t i o n s
1I i v i s i o n . S p r i n g f i e l d , V i r g i n i a 2 2 / 5 1 .
UNCLASSIFIED 25
20. S E C U R I T Y CLASS 2 2 . PRICE
ITHIS P A G E )
-
AUTOMATIC GENERATION
CONTROL- NOTES AND
OBSERVATIONS

by
William B. Gish

October 1978

Electric Power Branch

Division o f Research
Engineering and Research Center
Denver, Colorado

UNITED STATES DEPARTMENT OF THE INTERIOR * BUREAUOF RECLAMATION

 ACKNOWLEDGMENTS

The contributions of Mr. Brich of the Watertown Operations

Office, Western Area Power Administration, and Messrs.
Knutson and Realson of the Phoenix District Office, Western
Area Power Administration, in reviewing the text and
suggesting modifications are greatly appreciated.
 1111111

CONTENTS

Page

Abbreviations
andsymbols. . . . . . . . . . . . . . . . . . . . . . . iii
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . 1
Summary. . . . . . . . . . . . . . . . . . . . . . . . 1
BasicAGCcontrol concepts . . . . . . . . . . . . . . . . . 1
Areacontrolerrorformation. . . . . . . . . . . . . . . . . . . . 2

Interchangepower. . . . . . . . . . . . . . . . . . . 4
Frequency and time signals . . . . . . . . . . . . . . . . 9

Area control error disposition . . . . . . . . . . . . . 10

Modifyingthe ACE. . . . . . . . . . . . . . . . . . . . . . 12
The allocator. . . . . . . . . . . . . . . . . . . . . . . . 14

Quality of the AGC . . . . . . . . . . . . . . . . . 21

Bibliography. . . . . . . . . . . . . . . . . . . . . . 22

TABLE

Table

AGC minimum operating criteria . . . . . . . . . . . . . . . . . 21

FIGURES

Figure

1 Simplifieddiagramof the systemsynchronizingloop. . . . . . . . . . 2

2 SimplifiedAGCcontroller. . . . . . . . . . . . . . . . 3
3 TypicalPIDcontroller. . . . . . . . . . . . . . . . . . . . . 3
4 Dynamicscheduleof a joint-ownedgenerator. . . . . . . . . 6
5 Componentsof ACE. . . . . . . . . . . . . . . . . . . . . 11
6 Example allocator with variable-gain, permissive-control with
area requirement and plant requirement. . . . . . . . . . . . . . 16
7 Example allocator with mandatory control with area
requirement and plant requirement. . . . . . . . . . . . . . . 17
8 Example allocator with permissive control and participation
factors basedon plant limits. . . . . . . . . . . . . . . . . . 19
 ABBREVIATIONS AND SYMBOLS

ACE area control error P participation factor for a plant

AGC automatic generation control PA actual total interchange

Pc contracted power schedule
AR area regulation between utilities
B frequency bias in megawatts
per 0.1 hertz dynamic schedule
PDY
BGi base point for generating
plant "i" PGi power generated by generating
plant "i"
FA actual power system
frequency PLL lower limit power of a plant

FS scheduled frequency Pm measured intertie power

the number of the generating Pnm intertie power not measured

plant
Ps total interchange schedule
II inadvertent interchange

k the index for summing plants PUL upper limit power for a plant

KD derivative gain PID proporti ona I-integ ra I-derivative

KF total frequency bias PR plant requirement or station

requirement
KI integral gain
s LaPlace variable
Kp proportional gain
SHADE measurement error
Ks system synchronizing coefficient correction

KSL speed-level motor gain SLM speed-level motor

KT time error sensitivity TA actual or clock time measured

measured from power
MVA megavolt ampere rating of system frequency
equipment
Tg governor equivalent
MWS megawatt second inertia constant ti me consta nt
N the total number of generating plants
in a control area Tm mechanical time constant
(related to inertia)
NAPSIC North American Power Systems
Interconnection Committee Ts standard time from WWV

iii
,w.".
 INTRODUCTION BASIC AGC CONTROL CONCEPTS
The electric power systems of the United States The actual operation of the power system
and Canada contain control systems to regulate requires a form of speed control. The basic
and control energy flow of immense proportions. controller for speed is the governor which
These control systems work together to maintain maintains control of the "prime mover" energy of
clocks to within 4 seconds of WWV standard time a generator in response to generator speed. All
while simultaneously controlling the weekly governors on all the generators within a power
generation and distribution of 40 to 50 system work in parallel because all generators
terawatthours of energy (145 x 1015 to 180 x are locked together in synchronism. Application
1015 joules). This very significant engineering of this concept alone could adequately control a
accomplishment is often overlooked. power system provided all the tielines could
withstand the resultant power flows. However,
The control system for this major task uses over several improvements to system operations are
100 separate primary control units operating in implemented when a master control supervises
parallel to command thousands of generator the governor settings. These improvements
speed controllers (governors). Each of the include:
primary control units has different character-
istics and control philosophies, and only the most . Abilityto determine the individual system
basic concepts are similar. These basic concepts loads and allow each system to carry its
for control are contained in the NAPSIC operating own load.
manual [9].1
. Maintaining tieline flows to interconnected
The primary controller is the AGC (automatic utilities to allow accounting of power sales
generation control) system. The AGC system between utilities.
concepts have been described and discussed in a
multitude of books and technical articles. A very . Maintaining system frequency at 60 Hz
basic text for help in understanding AGC system which permits the widespread use of elec-
concepts is Control of Generation and Power tric clocks.
Flow on Interconnected Systems by Mr. Cohn
[12]. The comments and notes on AGC contained Note that the AGC is a supervisor of the normal
in this report were developed on the basic governor. The AGC should never attempt to
principles given by Mr. Cohn. This report is a overpower the governor during system tran-
series of discussions of techniques used in AGC sients. Although a control to dampen intertie
that are relative to applications within the Burea u swings could be developed for AGC [29], the
of Reclamation and the Western Area Power basic concept of AGC is to allow such swings to
Administration. be controlled by governors, voltage regulators,
power system stabilizers, and doc interties.
SUMMARY Power systems operating personnel view AGC
The AGe concept is not a complicated control systems differently than control engineers.
philosophy. Complications arise as the require- Operating personnel associate the AGC directly
ments of energy marketing and resource with interchange schedules, frequency, and time
management are factored into the control error. The control engineer views these as results
philosophy. The philosophy becomes more of the AGC. The control engineering concepts
involved as the opinions of good performance presented in the next few paragraphs is an effort
criteria are included. Although the basic AGC to provide as broad an understanding as possible.
concept is used throughout the power industry, The remainder of the report does not require
each utility will interpret the energy marketing, these concepts for a basic level of understanding.
resource management, and performance criteria
differently, and an individualized control concept The basic system response which the AGC
will be developed. This report is to help separate attempts to control is "accelerating power."
the basic control concepts from the individualized Accelerating power is the difference between all
characteristics and promote understanding of system generation power and the total system
the automatic gene'ration control. load including losses. If the power system is
connected to a neighboring utility, the power on
1
Numbers in brackets identify selected references listed in the tieline flowing out of the system is considered
the Bibliography. a load.
The frequency deviation is a measure of the Since the accelerating power is constantly
integral of the total accelerating power on the disturbed by random load changes and power
entire power system. Because of the synchron- system disturbances, the AGC attempts to keep
izing effect of all interconnections, and because the first and second integral of accelerating
the AGC should ignore power changes due to power (frequency and time) at zero. The
synchronizing effects, the AGC frequency input proportional and derivative of accelerating power
is assumed the integral of accelerating power. are controlled by the synchronizing and damping
The tieline flow error from the scheduled flow(or forces of the synchronizing loop. The first integral
interchange power deviation) is proportional to of accelerating power is also controlled by the
accelerating power. In figure 1, Tm is the governor in parallel with AGC. However, the
mechanical time-constant related to inertia and droop of the governor forces accelerating power
Ks is the synchronizing coefficient related to to zero after a frequency shift occurs (propor-
tieline strength [66); system damping is not tional control of accelerating power). Again, the
shown for simplicity. The rema.inder of the basic proportional-integral-derivative control from the
AGC system is shown in figure 2. The constant KF synchronizing loop and the governors must never
is normally known as the AGC bias term, 1OB. The be overpowered by the first and second integral
constants KSL and Tg are related to speed-level control from the AGC unless all governor and
motor gain and governor response time, res- system characteristics are completely known.
pectively. The LaPlace variable, s, is used in the
denominators to indicate integration. This type of AREA CONTROL ERROR FORMATION
control system is similar to a PID (Proportional-
Integral-Derivative) control system shown in The discussion of AGC control usually falls into
figure 3, except the derivative gain, KD.is zero. two categories. The first category includes the
The input reference to the AGC is zero level formation of the ACE (area control error) and the
accelerating power which always remains the second is the disposition of the ACE or the
reference. The interchange schedule and fre- allocation process. The formation of the ACE is
quency schedule are used to facilitatethe energy discussed in this section. The ACE is an error
and time accounting and are not the actual signal which, when positive, indicates excessive
reference to the control system. generation. A positive ACE always requires a

Signal Approximately Frequency Deviati on Signal Approximating

Proportional to of Nearby Systems Integral of
Accelerating Power Accelerating Power

Tieline +
Power
( Interchange
Power)
System
- Accelerating System
+ Power I Fre quenc
TmS Deviation
Sys t em System
Generation Inertia
Sys tem
Load

Figure 1. Simplified diagram of the system synchronizing loop.

2
 lilllll

Integral
(-ACE) Path
KF
A/locations Attempt
to Hold Constant
Gain Proportional
+ Path
Reference

Equivalent Interchange
System Tieline Power
~5
Governor
KSL
I + -L
1+li s System T s I
Equivalent Generation I
Speed -level I
I I
Motor I System I
L~roo~ ~oa~ J
Figure 2.-Simplified AGC controller.

Feedback

Kp

Output
Integral
Error K05

Derivative

Figure 3.- Typical PID controller.

3
reduction in generation. The ACE signal is always way of compensating for these errors. One way is
opposite in polarity to the normal error signal to adjust the schedule to account for the errors.
used in most other types of control systems. This adjustment is called "SHADE."
Normally, the ACE has the units of megawatts.
SHADE is defined as the integral of the errors and
The equation assumed by NAPSIC for all AGC is added to the schedule. Thus:
controllers is
ACE = (PA - PSI - 10B (FA - FS)
SHADE = OLD SHADE - ADJUSTED ACE
where PAis the actual total interchange power in
, This is calculated every hour or every day as the
megawatts with positive power flowing out of the system requires. The ADJUSTEDACEis the error
system; Ps is the total interchange schedule for calculation and is divided into several com-
the system; FA is the actual frequency of the ponents;
system in hertz; FS is the scheduled frequency;
and 10B is the system frequency bias in ADJUSTED ACE = inadvertent interchange,
megawatts per hertz (B is always negative). The adjusted time,
subtraction of reference from the actual quant- adjusted frequency,
ities is the reverse from normal control system adjusted ramps, and
procedures. scheduled inadvertent
interchange.
Interchange Power
If the components are broken up, SHADE
Although the interchange power deviation is a calculated for 1 hour can be shown to equal the
measure of accelerating power, the primary quantities: (SHADE from last hour) - (Metered
purpose of the interchange power is for energy sum from all boundary ties over last hour)
accounting. The AGC is primarily used to + (The scheduled power over the last hour times 1
maintain interchange schedules and it ac- hour) + ~ (Bias setting for frequency bias in
complishes this task by holding the accelerating megawatls per 0.1 hertz (Base reference fre-
power to zero, using the generation, rather than 9uency setting for last hour minus 60.00)} +
the tieline, to adjust accelerating power. Because 1(Scheduled power at the start of the hour
power generated in the system will use the path minus scheduled power at the end of the last
of least impedance to seek a load, there is no way hour minus scheduled power at the start of the
to constantly determine the exact power flow last hour plus scheduled power at the end of
path between several interconnected utilities. An the hour before last)/48f + (Total scheduled
energy meter on a specific intertie cannot easily inadvertent interchange desired by the dis-
determine the energy flow for the several utilities patcher).
which may have power contributions on the
intertie. Thus, the schedules and related rates for
the buying and selling of power between utilities The dynamic schedule.- The dynamic schedule
is recorded as contracted schedules and the total concept helps the smaller utilities which own or
schedule of all interchanges is developed. If the share in ownership of large generators. The
actual total interchange deviates from this total smaller utility is not required to regulate the large
schedule, it is not usually possible to properly generator alone but can share regulation over
adjust any individual schedule or to determine several areas more compatible with the gen-
the correct charges. The total interchange erator size.
deviation from scheduled becomes "inadvertent
interchange," and special arrangements must be Before dynamic scheduling, it was customary to
made to pay for the energy. Normally, agree- develop a schedule for 1 hour of operation and
ments exist between utilities to allow payback of leave the schedule constant. Many occasions
inadvertent interchange at agreeable times and required schedule changes in the middle of the
thus reduce the inadvertent interchange to zero. hour or at other times. A dynamic schedule
system was developed for use with certain large
generators having multiple ownership to allow
Measurement errors and SHAD£.- The accurate continual changes in schedules.
measurement of power (and frequency) and the
accurate setting of schedules is important; The generator must never be considered joint
however, errors in measurement and setting do operated, but rather joint owned. The operating
exist [11]. Each utility creates their own special utility where the generator is physically located

4
 Ijl111

must be in control of the unit. Many various seems to be in the control area of the utility
techniques are used to generate and account for owning the load. The schedules used for
the dynamic schedule. One method suggests contracti ng for the transfer of energy throughout
that: the control area surrounding the load is
calculated after the fact by averaging the actual
. Each owner calculates (or allocates) the load energy used over the hour.
desired power requirement or schedule for
the owner's share of the generation. This Metering the interties and accuracy.- The
does not necessarily have to be constant, metering of the interties between a system and
but may be allocated dynamically. This neighboring systems may be divided into two
allocation can be identical to the normal areas, the accuracy of the signal and the
allocation of any generation within the detection of errors.
owners' system.
The accuracy of the signal is again of more
. All the owners' power requirements are concern to the accounti ng procedures than to the
transmitted to the operating owner's system control system. The accuracy of transducers
where they are summed. range from 1 to 0.1 percent error at full scale. The
metering potential and current transformers
. The operating owner controls the gen- have an accuracy between 0.1 and 0.3 percent
erator power output with a closed loop power error at rated voltage or current. With scaling
controller. resistors, temperature coefficients, and tele-
metry electronics, and accuracy of 0.5 percent of
. The generator power output is continual- the reading is normally obtained at the location of
ly divided into the percentages for each the AGC equipment. If the energy (MW'h)
owner. This percentage is determined by monitoring equipment is assumed errorless (as is
the percentage of the owner's power re- normal for accounting), the transducer accuracy
quirement to the total power requirement. may be somewhat improved by a method similar
to SHADE. The energy is read every hour and the
. The owners' share of power output is power output read from the transducer is
dynamically (continually) added to the continuously integrated over the hour. By
owners' system boundary. comparing the actual and calculated energy
used, an error correction may be applied to the
. The energy reading for each owner, next hour of power data. Thus:
except the operating owner, is the owner's
share of generator power output integrated Calculated power (MW) =
over each hour. The operating owner sub-
tracts the total energy for all other owners Actual power reading (MW)
from the actual energy reading and uses the
result as his energy usage. Energy (MW'h) last hour
X
Calculated energy (MW.h) last hour
The operating owner must treat all other owners'
power output as power out of the system rather This concept works well if the power readings do
than input. See figure 4. not cha nge often and are not below 10 percent of
the transducer full scale. An alternative method
Another use for dynamic schedule involves the uses the same actual energy value divided by the
accommodation of another utility's loads (or calculated energy, and then a table of values is
generation) completely enclosed by the control formed for each range of average power readings
area. In this situation, there is no way of over the hour. This table is stored and used to
predicting the actual load ahead of time. Thus, multiply the actual power reading for each
there can be no scheduling of the load. If the load reading. In other words, a semidynamic calibra-
power requirement is used as a dynamic tion curve is used. This concept is rather
schedule for the control area surrounding the complicated to implement and update, and does
load, that control area can eliminate power flow not work well unless the transducer is frequently
to the isolated loads from their interchange operated over its range.
readings. Likewise, the neighboring utility which
owns the load uses the same dynamic schedule Analog telemetry channels usually involve some
to adjust his interchange readings. The load then delay or filtering of the power reading. In digital

5
 OPERATING
OWNER

Power
Output ~
0

Figure 4.-Dynamic schedule of a joint-owned generator.

6
 LlllIl

systems, the sampling rate forms a delay and systems, an analog filter before the analog to
often several computers must transmit the data digital converter is absolutely necessary. The
to the AGC computer, with a delay in the data at time constant of the filter should be no higher
each computer. In the Bureau of Reclamation and than one-forth the sampling frequency or 0.7
Western Area Power Administration systems, second for a 2-second sampling period. Good
the sampling rate is normally 2 seconds with results with aliasing reduction utilizes 3-second
transport delays often as long as 6 seconds. Thus, filters with a 2-second sample period. The gain is
the coherency of the various power readings then down for the high frequencies without
becomes an important question. If the primary undue phase shift at a 1 cycle per minute control
concept of AGC is not to control intertie swings, frequency.
then AGC response to signals faster than 0.016
Hz ( 1 cycle per minute) is really not necessary. It Errors in metering the interties.-A very serious
should be realized that with a frequency problem to the total summation of the intertie
response of 1 cycle per minute, the transient power is the loss of an intertie reading. If the
response of the AGC is not delayed by 1 minute transducer or telemetering equipment fails, large
but rather responds slowly at first to an error. changes in ACE will occur falsely causing
There are no intentional delays of the ACE signal. unnecessary control, inadvertent interchange,
In addition, the governors which the AGC and time error. Since most of the failures occur in
controls usually cannot function effectively the transmission link, the easiest failure to detect
above this frequency. Thus, filters with time is loss of carrier frequencies. Normally, contacts
constants above 0.16 Hz (10 times the 0.016-Hz are provided on telemetering or microwave
frequency) will cause very little phase shift or equipment which close on loss of carrier signals.
gain reduction. Sampling theory indicates that The AGC equipment can monitor these contacts
the samples should be at least 10 times faster than and should one close, the AGC is tripped or
the highest frequency of the controller (0.16 Hz). "suspended" until the trouble can be remedied.
Thus, a sampling frequency faster than 0.16 Hz With digital computers, the last good reading can
(or a sampling period of 6.25 seconds) should be be saved and a timer started, but the AGC
used. It is always best to sample as often as continues to operate. Ifthe timer, usually 20to 30
possible. Randomness in the sampling time of 6 seconds, times out without the carrier signal
seconds can be tolerated (the maximum sam- returning, then the AGC is suspended. If the
pling period) without seriously affecting the con- carrier returns a second timer, 10 to 20 seconds
trol. Thus, a sampling period of 2 seconds with may be used to insure the signal is steady before
a maximum of 4 seconds of random transport the reading is again updated. Should the second
delay would not seriously affect the AGC con- timer, or the "thawing" timer, not completely
trol. If frequencies above 1 cycle per minute cycle, the first timer is not reset but continues.
are required to be controlled, the sampling per- Thus, bouncing contacts will cause a time-out.
iods and filter frequencies must also be raised. Once the AGC is suspended, the operator must
The governors seldom respond faster than on remove the tie from monitoring status on AGC
a 10-second time constant which has a filter and adjust the schedule, usually with a manual
frequency of 0.16 second. Many hydroelectric substitution interchange value, before allowing
generator governors have a 30- to 40-second the AGC to become active again.
ti me constant or more with response frequencies
of 0.5 to 0.4 Hz. Normally, good control of the Failure of a transducer may be detected by
governors can be achieved at one-fifth the time- monitoring the rate of change of the metered
constant frequency or 0.3 to 0.08 Hz (the gain power, and if it is large and the final power
and phase are seriously dropping by this time). reading is near zero, an alarm may be sounded.
This becomes the fastest an AGC can be expected Unfortunately, the same effect may be produced
to control. Fortunately, these fast response times by transmission line tripping or intertie oscilla-
are not usually required and tuning of the AGC is tions and care must be taken not to "save the old
not as difficult. Actual AGC systems usually have value" unless the transducer is the device known
equivalent time constants of less than 0.0003 Hz to have failed.
or 0.018 cycle per minute. At 1 cycle per minute,
the gain may be down more than 40 decibels. A help in increasing the security of the intertie
readings is to have alternate transmission paths.
Since aliasing (or sampling frequencies higher These alternate paths are then used when the
than one-half the sample frequency) is normally primary path has a carrier signal failure.
expected from intertie oscillations between However, the secondary source must have the

7
same error detection as the primary and should Inadvertent interchange.-Inadvertent inter-
be constantly checked to ensure the channel is change is a concept developed for accounting
working when the secondary channel is needed. and is not directly used by the AGC control.
Inadvertent interchange (II) is defined as
Another source of error of the intertie power
readings is channel noise. These usually are II = fPsdt - 2; MW.h
most serious in the form of noise spikes caused 1
by electromagnetic interference. One method of over 1 hour
reducing the effect of noise spikes is to digitally
filter the input reading. Single time constant and is calculated when the energy (MW'h)
filters (recursive filters) with approximately readings are available, normally once an hour.
4-second time constants work well in 2-second The Ps is usually calculated from the start of the
sampling systems. Averaging filters which sum last hour to the start of the present hour. This
the last one or two readings with the present calculation does not include the ramping of the
reading and divide by the number of readings will schedule from normally 5 minutes before the
also help with noise spikes. The averaging hour to 5 minutes after the hour which is
process should not have more than three stages reflected in the energy readings. Thus, a ramp
in order to preserve the phase shift of the signal correction must be made to the integral of the
at a frequency one-tenth the sampling frequency. schedule.
The AGC control will automatically filter these
noise spikes because of the integrator (speed- Integration of intertie power deviation from
level motor) in the control path. The noise spikes power measurements also generates inadvert-
do increase the activity of the ACE. ent interchange as a continuous value. Although
this value is not based on the energy readings
If PMSC (programmable master supervisory (which are assumed perfect by the accountants),
control) systems are used to obtain the intertie the value can be used to aid the dispatcher in
readings, it is not always possible to determine determining the ability for the AGC systems to
the loss of a reading because the reading may maintain the interchange schedule. It should be
have been skipped when the equipment may remembered that inadvertent interchange is an
have been busy transmitting status events. A energy error between systems and may be
method of flags and timers is sometimes used to caused by many circumstances outside a
ensure that a reading is indeed transmitted. particular AGC system. Therefore, inadvertent
Another method uses a separate square wave interchange cannot be used alone to determine
voltage signal applied to an AID input at the the quality of AGC control of a particular system.
same remote terminal unit metering the intertie. This calculated interchange may be used in a
The square wave has a period of 4 to 6 seconds. If coordinated inadvertent interchange payback
this change is not detected by the AGC computer, method proposed by Mr. Cohn [11]. It also can be
the signal is assumed to have failed. The best used as a component of the integral of ACE
method depends on the communication pro- discussed in succeeding material.
cedures of the particular PMSC.
Reducing the inadvertent interchange.-Care
The summation of power or the intertie power must be taken when reducing the inadvertent
deviation.-When all intertie readings have been interchange since reduction of inadvertent
taken, and the schedule entered, including interchange for one system may increase the
manual entry of unmonitored interties, the inadvertent interchange in a neighboring system.
schedule is subtracted from the total intertie The payback of inadvertent interchange should
readings and the "intertie power deviation" is be negotiated between two systems so that the
formed. This quantity is used in the formation of payback will help both systems. This is called
ACE and the determination of inadvertent "bilateral payback." Ifthis coordinated payback is
interchange. not possible, then a "unilateral payback" may be
made but only if time error is decreased. Another
If the total intertie reading is subtracted from the concept would be to institute a "synchronized" or
total generation of the system, the result is an coordinated payback system using time error as
approximate value of the system load. If the well as IIwhere all members of a large power pool
frequency is varying, some of the power would payback II according to a signal from a
calculated as load is actually accelerating power. "master utility." [11] This does require that all
The tieline losses are also included in the value utilities use the same metering methods and the
of load. same methods for calculating II.

8
 Ijlllll

Frequency and Time Signals shows that the frequency signal is essentially the
integral of the accelerating power. Because the
The ACE is formed by two types of signals - the accelerating power is the variable to be
interchange deviation power and the frequency controlled, the integral of accelerating power can
bias. If only interchange power is used, the be used as a valuable signal for the controller.
control is considered "flat tieline" control. The However, the frequency is the integral of
utility must then depend on other utilities for accelerating power times Tm-1 where Tm is the
frequency and time correction and the utility will mechanical time constant of the system within
unnecessarily participate in disturbances within the AGC control boundaries including the inertia
neighboring utilities. Thus, this control is used of the lo!,!ds. To obtain the pure integral of
only when frequency is temporarily not available. accelerating power when frequency is summed
If only frequency bias is used, the control is into the ACE signal, the frequency bias, 10B,
known as "flat frequency controL" This control should be equal to Tm, where
forces a utility to provide power for all
disturbances on the entire system and does not
allow for reasonable accounting procedures. Flat MWS
Tm!::::!2~ rated
frequency control is seldom used. "Tieline bias
control" exists when both frequency bias and L.J MVA
rated
tieline deviation is used and is the normal,
recommended method of control. Each utility all
then shares equitably in a system disturbance.
equipment
Western utilities also add time deviation signals
to automatically correct time. Eastern utilities where the MWS is the inertia constant for the
correct time using the frequency reference offset equipment in megawatt seconds and MVA is the
method. Western utilities will also correct large megavolt-ampere rating of each piece of
time errors with the frequency reference offset equipment. Because this quantity is continually
method. changing as different generators and different
loads are connected to the system, the bias
Frequency signal accuracy.- The frequency should be continually changing. If the AGC
signal should have accuracy to 0.001 Hz or better control were required to perform excellent
and have near perfect long term stability. control for tieline oscillations, then the bias
Fortunately the frequency signal need not be fast
should theoretically be less than this value
since the AGC need not attempt to correct intertie (usually indicated as (3 in many references).
oscillations. Therefore, a filtering system or a However, the AGC is not normally used for tieline
method of measurement can have as much delay
oscillations and a bias equal to or greater than the
as a simple filter with a 3-second time constant. Tm or B is usually recommended. The method of
The phase lag should be maintained as low as determining the bias, 10B, is described in the
possible in the 1 cycle per minute range. Also,
NAPSIC manual [9].
since the AGC is to ignore intertie swings,
frequency is assumed the same over the entire
system and the frequency is monitored at the In practice, the power system is far too
nearest power system bus to the AGC equip- complicated to adequately measure Tm of a
ment. Failure of the signal can usually be specific control area, and the bias is based on the
detected by the signal going out of limits. These control area peak generation during the year. The
limits are usually set at 59.8 and 60.2 Hz. If the value of B may be 1 to 2 percent of the peak
system frequency is outside these limits for any generation.
reason, it is customary to suspend AGC
operation. The power system is in serious trouble Frequency Schedules and Use in Time Correc-
if frequencies of these magnitudes are ex- tion.-Since the inability to maintain the intertie
perienced, and the AGC can do little to help. Ifthe schedule results in movement of accelerating
frequency is telemetered, the loss of carrier must power, a frequency error is introduced. The
suspend control. A similar timing system to the frequency error then results in a time error that
interchange power monitoring system may be can be seen on electric clocks connected to the
used and alternative sources of frequency should system. Therefore, the time is normally moni-
be provided. tored by one utility in the system and compared to
an accurate time standard such as WWV or
Frequency bias.- The setting of the frequency WWVB radio stations. When ti me corrections are
bias is a very controversial concept. Figure 1 needed (when the error is 3 to 6 seconds) an

9
order is sent from the "time" utility to all other where Pm is the measured tieline power flows in
utilities requesting a time correction. All the megawatts and is positive for power flow out of
utilities change their frequency schedule (usually the control area. The Pnm is an entry from the
by 0.02 Hz) for a period of several hours. This is dispatcher to account for interties not measured
the same as changing the schedule interchange because of equipment failure or other technical
if the bias, 10B, is included; however, it is more or economic considerations. The scheduled
convenient to change the frequency reference power Ps = (~ Pd + PDY + SHADE where Pc is the
the same for all AGC controls working within the contracted power sold or bought between
overall system. utilities, PDY is the dynamic schedule from
joint-operated generators or loads, and SHADE is
The time error may not be corrected during peak the measurement error correction factor. The
system loads in some situations because the values of Pc, PDY, and SHADE are in megawatts
added energy for time correction utilizes higher and are positive for power or energy leaving the
cost fuel. The time may be "overcorrected" or system (or energy sold). The frequency bias, B, is
advanced using lower cost fuel in the early the bias value which converts a frequency
mornings in anticipation of the heavy system deviation into an equivalent power change; it is in
loads. megawatts per 0.1 hertz and it is always a
negative number. The measured or actual system
Time Deviation Signals. -It is now possible for all frequency, FA, is measured on the power system
utilities within the overall system to monitor near the AGC equipment and is in hertz. The
accurate time from WWV at a reasonable cost. schedule frequency is the frequency desired by
If all the utilities monitor time and bias ACE for the dispatcher; it may be different than 60 hertz
time correction, then time can be continuously for time error correction. The constant KT is the
corrected without resulting inadvertent inter- time sensitivity factor and is related to the total
change (from time correction). This is the very time desired for correction. The units are hertz
same concept as changing the frequency per second. All utilities with automatic time
schedule except the schedule is changed deviation correction must use the same KT'
continuously and automatically. To be success- Sometimes the time sensitivity is reported as
ful, all utilit.ies must use the same amount of 1OBKT and is given in megawatts per second. The
offset for the' same amount of time deviation. time, TA, is measured by integrating system
This quantity is usually called "sensitivity" of frequency and has the units of seconds. The time,
the time deviation correction system. In the T is the standard time from WWV.
s'
Western systems, a sensitivity of 0.02 hertz per
second has been used. In reference 11, a system
is presented that uses a signal from the "time- AREA CONTROL ERROR DISPOSI-
keeping" utility to all utilities to synchronize TION
both time and inadvertent interchange. Un-
fortunately, all utilities must participate to allow Once the ACE has been formed, the signal must
accurate corrections. be routed to generator governors so that the
generation of the system can be changed. The
principal function of this part of the AGC control
Usually the time deviation signal is limited to a
is to establish control loop gain. Before the ACE is
preset value so the ACE is not greatly biased
formed, the control gain could not be easily
for large time errors. For minimum inadvertent
implemented since all signals contributing to
interchange, all utilities must use the same limit
value as they must use the same sensitivity. The
ACE would have to be multiplied by the same
factor. After ACE is formed, a multiplying factor
time deviation error must not have much gain (or
can be applied to a single signal. After the ACE is
time must not be corrected quickly) since the time
formed, the gain should be as high as possible
deviation is the second integral of accelerating
without instability. There are two major concerns.
power control.
First, the gain of the speed-level motors differ
from unit to unit and more than one SLM
The Final Form of ACE.-The final formation of (speed-level motor) may respond. There is always
ACE is shown in figure 5. The basic equation is: the problem of deadband and hysteresis in the
SLMS and governors. Second, the control must
activate the SLM, governor, and prime mover
ACE = (P A - PsI - 10B { (FA - FS) + KT(T A -T s)} controls such as throttles or wicket gates. These
devices are mechanical in nature and wear out
and PA = (2; Pm) + Pnm with constant use. Unnecessary movement of

10
 I IIj~I

Standard
Time from
WWV

I Measured
s Time + Time
TA Error Time Sensitivity
Error
Limit Bias B is
lOB Negative

Freq uen cy Frequency Bias

Schedule Fs Deviati n
lOB
from
Operator B is Negative Intert~ange
Deviation
Ps
Measured Schedu e
Dynamic
Frequency from
Schedule
Operator
Shade
PA

Power Not
Measured from
Operato r +
+
+ of
+ +
, v
/

Mea sured
Interchange
Power

Figure 5.-Components of ACE.

11
these devices creates additional maintenance It is generally acknowledged that deadbands and
costs. Thus, a strong incentive is present to limits can be applied elsewhere in the control
reduce the number of movements or control system with perhaps better results and that little
operations if such a reduction does not greatly is accomplished by modifying ACE itself.
increase the inadvertent interchange. This
reduction of control activity is implemented by Smoothing Functions.-Smoothing filters can be
modifying the ACE and by the design of the applied to modify ACE. These filters are normally
allocator. simple, single-pole time constant filters which
work to reduce the gain of the "noise" and allow
Modifying the ACE the slow changes to have full gain. A time
constant of 120 seconds has been used to reduce
Modification of ACE is usually an attempt to system "noise." [33] This filter does reduce
reduce the random noise inherent in the ACE "noise" and allow correction of slow changes,
signal. The power system load is constantly but the rapid, large system disturbances are not
changing in a somewhat random manner and the responded to.
power system often restores itself by random
A possible remedy for the rapid, large system
disturbances which cancel the previous change. disturbances is to switch to an "emergency
Thus, the accelerating power is constantly in
assist" allocator using unmodified ACE when the
motion and, for a large percentage of the time, is unmodified ACE reaches a present level. The
constantly fluctuating about zero. The AGC
filtered ACE is then used when the ACE returns
control cannot remove this "noise" from the to the normal operating levels.
power system with the slow speed-level motors
and governors. Further, to control such self- The Probability Filter.-Many AGC systems have
correcting variations causes unnecessary wear adopted some form of a probability filter. The
and tear on the mechanical parts. Therefore, two basic filter design was presented by Dr. Ross [33]
criteria are usually used to decide when ACE is using the designation "Error Adaptive Control
valid. First, slow drifts within the ACE must be Computer." The adoption of the filter to
corrected to minimize the inadvertent inter- Bonneville Power Administration is reported by
change and the time error. These slow drifts, due Taylor and Cresap [42].
to slow load changes, are of the range of 5- or
1O-minute duration or more. Second, large rapid The description of the filter is not overly
changes of ACE indicate the loss of generation or complicated, but the description of the adjust-
load. A reasonably rapid response of the AGC will ment of the filter parameters is very difficult. The
allow a control area to rebalance generation to adjustment depends on the operating character-
load without unnecessary or lengthy assistance istics of the power system which are difficult to
from neighboring utilities. It should be noted that measure and interpret. The following paragraphs
many AGC systems do not modify ACE and do attempt to describe the needs for the filter, the
work satisfactorily. However, the use of some basic design of the filter, and a basic concept of
ACE modifications has significantly reduced filter adjustment.
activity of certain AGC systems.
The "Error Adaptive Control Computer" des-
Limiting and Deadband. -The most si mple cribed by Ross is based on the need for
modification to ACE is to provide a deadband discrimination of three different types of signals
sized to remove some of the random "noise." The found in power system control. These are
deadband works well to reduce activity and to "deterministic," "probabilistic," and "sustained"
allow rapid recovery of system disturbances. error signals.
However, the deadband may mask a small, slow
change or offset in the ACE. Another modifica- . Deterministic error signals have character-
tion to ACE is to provide a limit. The limit is istics which are known although their occur-
especially useful in systems where a larger ACE rence may be random. These deterministic
could not influence more generation anyway errors may be caused by intertie swings, periodic
and, thus, the dynamic range of the ACE is not load fluctuation (i.e., rolling mills or arcfurnaces)
excessive for the hardware. Often, ACE levels or large, sudden changes in load or generation.
cause a switch of gain on the ACE signal by
changing the type of allocator used. Such a . Probabilisticerror signals are random in nature
switch may be called "emergency assist" mode. but may have some definable amplitude level and

12
 IJHI

frequency band. These errors may be caused by Ross implies the opposite concept. When the
the normal, random changing of load on the zero crossing occurs, a separate timer is reset
power system. to zero, and if no other filter elements are
active, the gain is set to zero. This does not
. Sustained error signals are usually caused by necessarily filter the high frequencies because
the control system errors from either man or other filters may be active but does provide
equipment. These errors develop into inadvertent strong control of "sustained" ACE. Ross also
interchange and/or time deviation. suggests two different timers to provide even
higher gain for "sustained" ACE that do not
The filter attempts to discriminate the signals of seem to respond to control.
each type into errors which need correction (or
more accurately, which will respond to cor- The weighted error signal and the integral of
rection and benefit the system in an economical the error signal exceeding a preset limit for a
sense) and errors which can be safely ignored. preset length of time.- This filter element
The result of the discrimination process is an works on the concept that if the ACE and the
"amplitude modulation" of the ACE. The ampli- integral of ACE are not sustained for a given
tude modulation is usually in discrete steps, length of time, control is "probably" not re-
such as multiplying the ACE by zero, one, or quired. The difficulty lies in the word "probably."
two. The ACE is never "phase shifted," because To one observer, the control is too active, and
an amplitude modulation is used, however, the to another, the same control is very quiet. The
error is used to control governors and power measurement of "probably" is usually control
system inertia, and the time constants of these activity. There is further discussion of the
devices use the overall effect of the modulated "activity" in succeeding material.
wave and effective phase shifting does occur.
Both Ross and Taylor describe a probability
The process is very similar to the firing of
filter with two of these filter elements. How-
SCR's (silicon controlled rectifiers) into an in- ever, the method of determining the integral
ductive load. The individual pulses of voltage
used by these elements differs. Ross indicates
are smoothed into a current and the overall
that the integrator should be periodically reset
effect is very nearly the same as if th~ voltage
to the inadvertent interchange and time devia-
were amplitude controlled and phase shifted
tion as calculated from sources other than ACE
(instead of being switched).
(such as watthour totalizers, etc.). This implies
that the integral stage can have a substantial,
As described, the output of the filter is switched sustained, but limited value. Taylor uses an
between two or three gains. This switching integral with very low limits and thus provides
process is basically caused by the ACE crossing a hysteresis effect on the weighted ACE.
zero, and the weighted ACE (a constant times
ACE) plus the integral of the ACE exceeding a The weighted ACE signal is essentially an
preset value for a preset time. Usually several artificial error energy signal in the same
filter elements are used in a specific filter manner that the integral of ACE is the actual
application. error energy signal. (Another way to look at the
signals is to realize that a power error, ACE,
The error signal crossing zero.-When the ACE applied to the system inertia procedures fre-
crosses zero, an assumption is made that the quency error.) The artificial error energy
power system is restoring itself and no con- indicates what energy is in the ACE signal if it
trol action is required. Each time the ACE had continued at the present level for a preset
crosses zero, the gain applied to the ACE is (or weighted) time. This can be considered a
set to zero (this is the concept described by method of predicting for a short period of time
Taylor [42]). The gain remains at zero until a what the error energy will be. The sum of the
timer times out or ACE again crosses zero. This actual error energy and the "predicted" or
provides a very effective "filter" of all fre- artificial error energy give a total error energy
quencies having periods of less than twice the or a total expected energy that indicates how
ti mer setting. Taylor sets the timer to 3 seconds serious the error is. If the error energy exceeds
and all frequencies above approximately 0.16 a preset level, the error is serious and may
Hz receives no control. It also should be noted require control. A timer is included to ensure
that this type of zero crossing filter will allow that the serious error energy remains serious
slow or sustained ACE of about one-half the long enough to "probably" require control (or
random noise signal amplitude to continue that the expected time error is enough to
without control action. "probably" require control).

13
 If the error energy builds to a high level in a errors decreases a nd the dispatcher becomes
short ti me, the error is considered serious. But, aware of circumstances when the control is too
if the error energy builds to a medium level for quiet. Reasonable settings for the filter elements
a longer period of time, the error is more must be determined despite the conflicting
serious and the gain is then doubled. This long condition.
duration may indicate that the error is not
responding to control action. The result of the One method for adjusting the filter is to
filter element is that a certain area of frequen- precalculate the settings based on some simple
cies include the higher frequencies. Also, a data measurements and then tune the system
deadband effect occurs because the filter will while in service. Ross discusses this approach.
not respond to low level signals. The "optimum" tuning can be done by a very
knowledgeable installer who fully understands
It is important as noted by Ross thatthe integral the filter. Unfortunately, few people fully
output does not ever exceed the preset level understand the filter and often less knowledge-
for switching gain of the sum of the integral able people will later alter the tuning to suit their
plus the weighted ACE. Otherwise, the ACE "feel" of the system.
would become active (have a gain other than
zero) when excessive inadvertent interchange A second method of adjustment is by simulation
is present and the control would become ex- as described by Taylor. This method usually
tremely active (but miscontrol would not allows a parameter study to be made so that the
result). For this reason, a hard limit on integral change in operation from the change of any
of ACE is required. parameter can be understood. Usually, most of
the parameters are then defined as "best," and
There are numerous ways to adjust the probabil- one or two parameters are left for the tuning
ity filter elements. The adjustment is primarily process during normal operation. System simu-
based on activity and speed of system recovery; lation is not an easy task and very often
neither of these concepts are well defined. Sys- parameters found from simulation neglect some
tems have operated (and are still operating) characteristics of the system which the filter
without the filter; however, on these systems, must either accept or reject.
there have been some complaints about the
control being too active or not fast enough. It is
Biasing the ACE with the Integral of ACE.-Some
known that if activity is reduced, the speed of
AGC systems use the integral of ACE to bias the
recovery or response can be increased. Because
ACE; this helps correct slow drifts in ACE. The
increased response usually implies lower levels
principal effect is the reduction of opposite
of inadvertent interchange, speed of recovery
polarity pulses to the generators resulting in
may become important. The response speed also
reduced activity. As an example, during an
depends on otherfactors such as the number and
extended loading period (such as the morning
size of the generators responding and the re- loading), the integral of ACE will restrict "lower"
sponse of the powerplant controllers, and be-
generator pulses and favor "raise" pulses. The
cause the inadvertent interchange depends on
activity is reduced because a lower pulse would
many different parameters, the only quickly
require a raise pulse in a short time. Improved
observable characteristics of an AGC system is
results with the integral of ACE can be obtained
activity. Therefore, a "feel" for systems activity is
by also using a filtered ACE (as previously
usually developed by anyone who has respon-
described). The reduction in activity will be
sibility for an AGC system. Activity can usually be
between the filtered ACE alone and a well tuned
mentally equated to "wear and tear" of
probability filter.
equipment. The speed-level motor may wear out
faster when the AGC is active, but it is debatable
whether the governor suffers from added "wear The Allocator
and tear" since there is always a little governor
activity due to the power system frequency. The a lIocator is the most controversia I and
Because inadvertent interchange is more "im- complex part of the AGC system. The control
portant" than speed-level motors, the normal system requires that the ACE is delivered to a
association should be "less activity to allow for generator to alter the generation and thus the
more response to reduce inadvertent inter- accelerating power. The principal question is
change." Unfortunately, as the activity de- which generator must be controlled. The first area
creases, the ability to respond to certain types of of concern is the range between "permissive"

14
and "mandatory" control concepts. These con- mandatory operation. Some examples include (1 )
cepts are based on the requirement for base load placing baseload on mandatory control with
setpoints. basepoints and the AGC generators on permis-
sive control without basepoints; (2) executing all
Base/oad Setpoints.- The generators used in a baseload changes with all generators in man-
power system have many constraints. Some datory and then operating the rest of the time in
generators may be required to run at a fixed load permissive mode; and (3) not allowing baseload
for mechanical reasons. Other generators have changes to oppose the ACE (but if ACE is zero, the
auxiliary equipment which runs best between mandatory movement is made). There are
two specific (and usually close together) load numerous variations. However, the basic con-
levels. Other generators may have reservoir level cept is that the permissive control provides good
or river flow constraints. Thus, when the control with m in imum activity a nd the ma ndatory
dispatcher selects generators to supply the load, control provides good capability to honor
many generators must be maintained at a generator or powerplant constraints.
basepoint generation level.
Participation Factors.-Participation factors are
If the basepoints are allowed to change, often the the crux of the allocating system. Unfortunately,
movement of the generator to the new basepoint they are also difficult concepts to define. The
must be done as quickly as practical to avoid normal definition of participation factor is the
turbine rough zones or switching of auxiliary fractional part of the ACE that is allocated to a
systems. Thus, some generators must be ramped specific generator or plant. In most systems, the
to the new basepoint without regard to the participation factor is only a small part of the
immediate need of the power on the system. determination of the amount of ACE that is
allocated to a plant. A few examples (not an
The same concepts of baseloading and ramping exhaustive summary) of actual systems will
also occur for entire powerplants. Hydroelectric demonstrate the various uses of participation
installations may be required to pass large factors.
volumes of water or must be run to maintain a
certain river flow, and the requirements of the Permissive control with plant and area
immediate system are less important. requirements.-Figure 6 shows the basic
block diagram for the allocator using station
Permissive or Mandatory Contro/.-A permissive requirement, area regulation, and variable-
allocator can be defined as an alloca'tor which gain, permissive control. The effect of variable
never causes a control output that either perturbs gain is present because the switch to send
or conflicts with ACE (i.e., the immediate system ACE to a plant may be closed for more than one
energy balance is most important). This is the plant at a time and there are a variable number
best controller from the control theory stand- of plants or generators responding to the ACE.
point. The ACE is always in control and no The permissive concept is that only ACE
unnecessary control actions occur. The probabil- signals are sent to the plant, and even though
ity fiIter used to modify ACE works especia lIywell the "station requirement" has a value, no
on a permissive control since the closed filter (or control is present until an ACE is present.
zero ACE) will not cause any control which will The AR (area regulation) is:
attempt to open or activate the filter. All filter
N
activations are caused by the power system loads
and frequency shifts.

A mandatory allocator attempts to hold all plants

at a basepoint. The plants move their power in
AR = ACE +
~i=1
(BGi - PGi)

response to the ACE but when the ACE is zero, where PGi is the monitored generator power
the plants return toward the basepoint. This for plant i and BGi is the basepoint setting that
system works well when generators have severe the dispatcher has selected for plant i. The
constraints. The ACE is always perturbed as soon total number of plants in the control area is
as it is zero unless the unlikely situation occurs N. The station requirement (or plant require-
where all the plant basepoints match the ment) is
required system generation. N

Actual allocators of the various AGC controllers PR = PG - BG + p[ ACE +

2.: (BGi -PGi)]
usually use a combination of permissive and i=1

15
 ACE to
Plant

Close Only
ACE When ACE
and Station
Requirements
are Some
Participation Sign
Foetor from
Sum of Operator
Plant
Errors
Bose Setting
from Plant +
Operator Error Station
+ Requirement
Actual
Generation
from Plant

TYPICAL PLANT
Figure 6.-Example allocator with variable-gain, permissive control with area requirement and
plant requirement.

where PG is the monitored plant generation, The plant requirement is:

BG is the basepoint for the plant and P is the
N
participation factor selected by the dispatcher.
The sum of all the participation factors for all
the plants should be 100 percent.
PR = PG - BG + P [ACE + L (BGi - PGi) ]
i=1
For control analysis, the "participation" of the
plant is not the participation factor alone but where the variables are the same as in the
the fact that ACE and PR are the same sign. previous example. If all the plant requirements
The participation factor actually acts as a bias of this example are summed together,
to determine which plant will be favored to N N N
receive ACE. However, a plant may receive ACE
even if the participation factor is zero. Note that
LPR =L: PGi - L:BGi
the difference between plants on "automatic"
and plants on "baseload" is that the plants on i;t ACE~=t (B~:1-PGk)]
baseload havea zero participation factor. Also, +t
positive ACE, AR, and PR indicate the genera- ;=1 k=1
tion should be reduced. N

Mandatory control with plant requirements

and area regulations.-Figure 7 shows the
where Li=1
P is always 100 percent. Thus

basic block diagram for the allocator. The N

concept looks very much the same as figure 6 LPR = ACE
as far as the form of the equations. However,
i=1
a more careful search' will show that ACE is
not ever sent directly to the plant. Rather the and the system is "mandatory" in the form
plant requirement is sent to the plant and ACE that control signals will be sent even if ACE is
is only one of the components. zero, but since the sum of all control signals

16
 IJI111

ACE
+

Participation
P Factor

P= I
Number of
Plants on Control
Sum of
Base PIant Errors
Settin Plant
from Error ( PIant
Operator +
Actual Generation Control
from Plant Error)

TYPICAL PLANT
Figure 7.-Example allocator with mandatory control with area requirement and plant
requirement.

is ACE, the system is somewhat like "per- this is done, the concept of the total PR being
missive" control. In theory, the control is very equal to ACE is lost. Ways of providing more
good. In practice, the response of the plants permissive control are:
to the control signals will not be proportional
to ACE because each plant may have a dif- . Limit the PR if it opposes the ACE. If ACE is
ferent gain to the plant requirement. The zero, or not in opposition, the PR is not limited.
concept will work but perturbations of ACE
will continually occur as the plants seek a . Set PR to zero if the PR opposes the ACE. If
balance point among themselves. This control ACE is zero or not in opposition, send the PRo
exibits mo.r:eactivity than the purely permissive
controls. . Set the PR to zero if the ACE is zero or
opposes the PRo If ACE does not oppose the
In this example, the baseload plants simply have PR, send the PRo
zero participation factor (in practice they also
have a frequency regulation term while in
baseload). Also, the participation factor is . Set the participation factor to zero and PR to
simply 100 percent divided by the number of zero if ACE is zero or opposes the PRoThen the
plants on automatic control. This implies that participation factor is 100 percent divided by
all plants are approximately the same size and the number of plants on control and have a PR
will respond nearly the same for a plant re- that aids ACE.
quirement signal.
The progression from the first to the third
This type of control can be made more permis- concept tends to reduce activity but also allows
sive by limiting the plant requirement. When the gain on ACE to become more variable. The

17
last concept is nearly the permissive con- generation is excessive and must be reduced.
trol of the first example except the gain is more The actual plant power is subtracted from the
consta nt. The mandatory changes are removed lower power limit of the plant (determined by
and a plant will respond only to signals in the dispatcher) and divided by the summation
phase with the ACE. of all the differences of power and lower limit.
Thus, for positive ACE,
Permissive control with participation factors PG - PLL
based on plant limits.-Figure 8 shows the Pi = N
signals used in a permissive automatic control - PLLi
with participation factors based on plant limits. L:IPGi I
It is noteworthy that the basepoint allocator is i=1
mandatory control (with a frequency regula- where PLL is the lower power limit of the plant.
tion term). This system has basically the same For negative ACE,
equations as the first and second example. A
major difference is that baseload plant errors PG - PUL
p. =N
are not summed into the area regulation but I
rather are used to directly bias the ACE for - PULi
ra mp cha nges. Seca use the participation ~IPGi
I
factors are dynamic (they change with gen- 1=1
erator loading), the allocator can be reduced
to a "floating" basepoint controller where no Where PUL is the upper power limit of the
base setting is required for plant on automatic plant. This procedure works well since the
control. plant farthest away from the limit will have the
highest participation and a plant at the limit
If will have no participation. Two problem areas
N
PR = PG - Ps +P ~ (PSi - PGi) + ACE
exist however. If the PG - PLL or PG - PUL
terms become negative. the control would
reverse the plant requirement and the control
i=1 would become mandatory. Thus. negative
in the sum of all the participation factors in terms in the numerator of Pi must be set to
subsection 1, and if PSi - PGi are summed only zero. Also, the absolute value is not necessary
for units on control, then since the negative terms in the summation
N should also be set to zero before summing.

i=1
~
PRi = ACE The second problem occurs when all plants are
at or beyond the limit. Then the denominator
becomes zero and Pi must be limited to a
However, if the basepoint (PS) is eliminated, number below overflow within the computer.
the same effect can be obtained from
The actual participation is:
PR = P(ACE) and
N Pi
P=
PRi = ACE ~ Pi
L:
Ii=1 This relationship provides that the sum of P will
N be one. However, as the sum of Pi becomes
because
Li=1
P = 100 percent zero because all plants are at zero, a zero over
zero condition occurs which should be avoided.
This relationship is not necessary if negative
The floating basepoint is superior to the fixed terms in the numerator and within the absolute
basepoint system in practice although they value of the denominator are set to zero. The
are the same theoretically. The fixed basepoint possibility of having no participation is present
will cause ACE perturbation because the gains and should be brought to the operators atten-
of the various plants are not equal. Whereas, tion.
the "floating" basepoint concept does not
cause such perturbation. An additional approach of changing all plant
limits to emergency limits when all participa-
The participation factor itself is calculated tion factors are zero would provide automatic
using the polarity of ACE. If ACE is positive. the emergency allocation.

18
 IJIJ.!!I

ACE

p. Participation
L
LP' Factor

Pi = Px- PG
L I Px - PGI

Base Setting Sum of PIants

from on Control Plant
+
Operator Plant + Requirement
Error
+
Actual
PG Generation
from Plant TYPICAL PLANT
Figure B.-Example allocator with permissive control and participation factors based on plant
limits. (Px = Upper plant limit for - ACE and lower plant limit for + ACE. Limits are entered by
operator. P G =Actual generation from plant.)

The mandatory baseload system is not really except small allocations during each pass (which
compatible with the above concept. An attempt may be within a deadband in the generator
to overcome baseload changes by biasing ACE controller) are avoided. Therefore, a response to
with baseload changes helps overcome some the loading signal is usually more precise, the
of the problem, but still causes perturbations gain of the system is held more constant, and the
to ACE which the floating base point system baseload units can be used if the rate of system
for automatic generation attempted to elimin- response becomes too low.
ate. If a plant is required to have specific
ramp rates and mandatory settings, the plant Participation factors are also used in economic
error (i.e., PG - PS) should be subtracted from dispatch. The concept of economic dispatch is
ACE rather than the change of Ps only, and outside the scope of this report.
ACE will perturb less (theoretically, not at all).
However, if the loading of a plant can be per- Emergency Assist and Standby Generation.-
missive, the upper and lower limit could be set When the ACE becomes very large, most AGC
to the desired basepoint. Then loading would systems have provision for providing very high
be done according to the needs of ACE. Also gain and, thus, quick generation response.
changing limits in an emercency would auto- Usually, the resulting gain is actually larger than
matically include the plant baseload in normal control would allow. If the gain is active
the allocator. for more than the minimum time to reverse the
ACE trend, the AGC will tend to become unstable.
There are many forms of participation factors The emergency assist may take many forms.
besides the "coefficient-of-ACE" factor de-
scribed in the preceeding three examples. . All plants connected to the AGC system
Circular allocators will load plants according to including baseload plants are sent the ACE when
the size of the plant similar to the third example, the ACE exceeds a present limit. There is no

19
participation and all plants receive the ACE, thus Pia nt control error. -The pia nt control error
multiplying the effect of the ACE many times. is sent directly to the plant by digital or analog
When the ACE recrosses a preset value (a little transmission. This system provides an ac-
hysteresis is used), normal allocation procedures curate power controller around the plant and
are then used. On special occasions, the ACE has a very definite maximum gain for the SLM
may oscillate several times but will stabilize. integrator. If the gain becomes too high,
. A baseload assist system is used when the ACE oscillations develop. One method of controlling
becomes large or when the ACE cannot be the gain is to allocate the error at the plant in
allocated to control plants (because of Iimits). The the form of pulses to each generator using a
gain is changed by providing a second allocator to sophisticated plant allocator. This control
allocate to the baseload plants in the same system must also be rate limited to ensure
manner that the main allocator used. Therefore, that system damping is not impaired.
the gain is approximately doubled (unless the
main allocator cannot allocate due to limits). If Power setpoint signa I.-Another method for
the ACE is large, plants on "standby" are controlling the plant is the use of a power set-
included in the allocation. These "standby" point transmitted either with analog or digital
plants have their generation controlled at the equipment. The allocated error must be inte-
plant. and in "emergency" conditions, the ACE is grated in the AGC controller (replacing the
used by the plant. SLM integration). Then the powerplant may
use an allocator and closed-loop power con-
The emergency assist is actually an attempt to troller. Th is places the closed-loop controller at
reduce the system droop. The speed-level motors the individual generator and also allows the
move the speed reference of the governor so that use of local constraints (breakers, gate limit,
more generation is changed than would have rough zone, efficiency, temperature, etc.) to
been possible with the droop alone. Care must be modify the individual controllers. Usually a
taken not to cause a larger disturbance than the rate-of-power or a predictor closed-loop con-
original disturbance. troller is used [35, 65]. The rate-of-power con-
troller usually has a rate-limit or loading rate
Signal to the Plant.-Once the allocator has included. The predictor controller does not
calculated the plant requirement. the signal must require a rate-limit system and may respond at
be transmitted to the plant. This communication nearly the maximum linear rate of the governor
can take place in several ways. (if the windings, fuel handling equipment, or
river level can tolerate the loading speeds).
Pulse systems.- The size of the ACE is This speed of response is not usually needed by
periodically converted to pulse width modula- by the AGC, however.
tion and sent to the plant. There is a definite
period to the pulse (2 to 6 seconds) and the Plant Regulation Included in the Signal.-
width of the pulse hasa maximum value. Since Since all AGC allocators, with the possible
the speed-level motor will integrate these exception of the floating basepoint concept, use a
pulses (after being allocated to the generators form of closed-loop power control. the long term
at the plant), the system is considered a "rate- effect of the droop of the governor (or the speed
limited" control system. The generation can regulation of the governor) will be negated. The
move only at a certain maximum rate limited basic concept of the AGC system is to use the
by the maximum speed of the motor and the droop characteristic of the system as a basic
maximum pulse width. Thus, the gain change assumption and reregulate the system around
during emergency assists actually increases the droop characteristic. The bias setting on the
frequency component of ACE includes the
the rate limit by using more than one plant.
system droop characteristic. Thus, removal ofthe
The plant is in a "rate-limited" power feedback droop by a constant power controller opposes the
loop if measured plant power is used in the basic AGC concepts and opposes the natural
AGC allocator. Since the rate-limited control system response.
automatically reduces gain when the speed of
the disturbance is rapid, system damping is not It is preferable to have every generator controller
seriously impaired. Further power-rate feed- which is not being directly influenced by ACE to
back is usually not provided and the control have some type of frequency bias allowing the
gain around the plant power control is droop to work. Even plants directly affected by
usually low. ACE should have some type of frequency bias to

20
aid in system load stability although the ACE may The regulation margins.- The difference be-
overcome the frequency bias over the long term. tween the present generation and the maxi-
This concept becomes more important as the mum capability (or limit of generation) for each
strength of the closed-loop power controller plant on automatic should be calculated. Also,
around the plant increases. the difference between the minimum plant
power limit and the present power output for
Another problem in plant control is the governor all plants on automatic should be calculated.
with the dashpot bypassed, or governors that These two calculations will provide alarms
have been tuned to give very quick responses. when plants can be put on or taken off auto-
These governors usually depend on the power matic to ensure an adequate operating range
system for control loop damping and they will not for the AGC system.
operate with an isolated load. The power system
can support many governors of this nature but Spinning reserve.- The spinning reserve is
there is a maximum limit. Unfortunately, when used as an indicator of system reliability and
the limit is reached, system oscillations may is usually the difference between the total
begin without any obvious cause or warning. maximum limit of generation for every gen-
Thus, the number of "fast" governors should be erator on-line and the present power output
kept to a minimum. of every generator on line.

Calculations Useful for Plant Allocation.- QUALITY OF THE AGC

Several calculations and tabulations are useful
for operation of the AGC and, in particular, for The NAPSIC operating guide [9] defines a mini-
determining the capability of the syst ~m in the mum operating criteria. This criteria is sum-
event of a disturbance. marized in table 1.

Plant response errors.-A powerplant may not Table 1.-AGC minimum operating criteria
respond to AGC commands for a variety of
reasons including all generators at limits or For normal system conditions:
loss of communications. This condition may be
detected by finding the plant power outside a . Area control error (ACE)must equal zero
deadband from the base generation plus at least one time in all10-minute periods.
plant requirement. Another method is to
monitor the rate of change of plant power to . The average deviation of ACEfrom zero
determine if the direction of the plant require- for all 10-minute periods must be within a
ment and plant power correspond. Once the specified limit, Ld' based on system genera-
slow response has been detected, a signal tion. The value of Ld can be found from
may be sent to the plant to determine if any calculations in the NAPSIC manual and is
response is possible. If no response is found based on maximum rate of change of load.
the plant should be removed from allocation
and an alarm generated. For disturbance conditions:

Disturbances outside the AGC system.- . ACE must be returned to zero within 10
Often, severe disturbances outside the AGC minutes.
area may require action by the dispatcher or
system operator. Such disturbances can be . Corrective action must be observable in
alarmed to the operator by using the tieline the ACE within 1 minute of the disturbance.
power error and the frequency bias terms
usually added for ACE. However, the two are This criteria is easy to use since it depends only
subtracted to give the "ACE" for the neigh on measurements taken from chart recordings of
boring systems. If the disturbance is within the ACE.
local system, the ACE will become large. When
the disturbance is outside the system, the ACE Theoretically, each AGC system is "indepen-
should remain small, but the "reverse ACE" dent" of the neighboring AGC systems since the
will become large. If the equivalent external addition of frequency bias to the tieline error
system bias can be found, the signal may con- results in an ACE that should respond only to
tain the actual disturbance size information. disturbances internal to the system and ignore

21
disturbances in neighboring systems. In practice, 2 W. R. Brownlee, "How to Account for Energy
the setting of the frequency bias is not equivalent in Complex Interconnections," Electrical
to the system characteristics, and interaction World, October 23, 1961.
does occur.
3 H. J. Fiedler, "Power System Operation,"
Thus, the ACE of a given system cannot be relied Indian Journal of Power and River Valley
upon for fine tuning of an AGC system. The ACE Development," vol. 21, No.7, July 1971 .
will indicate course adjustments improvement.
but the ACE changes due to activity in a
neighboring system will mask fine adjustments 4 C. J. Frank, "Generation Control Research
and improvements will not be conclusively Opportunities," Power System Control
indicated. Workshop, San Louis Obispo, CA, April
1977.
Another problem lies in the difficulty of the lack of
definition of good AGC operation. The definition 5 H. J. Holley and J. C. Hung, "Load-Frequency
seems to cover at least three areas. First, the Control of an n-Area Power System Inter-
integral of ACE should be maintained as close to connected by AC Lines and DC Links,"
zero as possible. Second, the AGC should International Electrical, Electronics Con-
respond to transients and disturbances, and, ference, Toronto, Ontario, October 1971.
third, the AGC must not create unnecessary
operation of the mechanical equipment in the
system. Unfortunately, only the first area can be 6 Mochon, "Power System Planning and
analytically measured. The second and third Operations: Future Problems and Research
areas are defined more by opinion than by Needs," EPRI EL-377-SR, February 1977,
analytical testing. Henniker Conference 1976.

A possible solution would be to develop an

"index" of performance which, over a 10-minute 7 R. Pod more, "Recommendations for Control
period, would measure the activity of the ACE, of Jointly Owned Generating Units,"
the integral of ACE, and the activity ohommands System Control, Inc., September 1975.
to the speed-level motor. A single number of
"goodness" would be generated. However, B R. O. Usry, "Inadvertent Energy Interchange-
interpretation of this number may be difficult and Causes, Remedies, and Balancing:' IEEE
involve opinions and may cause more problems PAS, vol. PAS-87, No.2, February 1968.
than it solves.

Another possible solution is to use a fast Fourier [9] "NAPSIC Operating Man-
transform calculated every 10 minutes for ACE, ual:' North American Power Systems Inter-
speed-level motor activity, and the integral of connection Committee, April 1976.
ACE. Then comparisons of the amplitude of the
frequency components could be made. Because 10 L. A. Mollman and T. Kennedy, "Interrela-
the response of various frequencies may not be tionship ofTime Error, Frequency Deviation,
as dependent on neighboring AGC systems (each and Inadvertent Flow on an Interconnected
system will have characteristic frequencies of System," IEEE PAS, vol. PAS 87, No.2,
activity), the effects of fine tuning of filters and February 1968.
other constants may be easier to see. However, in
the final analysis, the improvement will still be a
matter of opinion. AGC THEORY

[11] N. Cohn, "Energy Balancing On Inter-

BIBLIOGRAPHY connected Systems," Proceedings of the
Americal Power Conference, vol. 35, 1973.
GENERAL AGC OPERATION

1 A. R. Benson, "Control of Generation in the [12] N. Cohn, "Control of Generation and Power
U.S. Columbia River Power System," Flows on Interconnected Systems:' John
Bonneville Power Administration, 1960. Wiley and Sons, 1966.

22
13 N. Cohn, "A Step-by-Step Analysis of Load- 23 J. F. Engle, "Effect of Grand Coulee Deriva-
Frequency Control Showing the System tive Type Governor on Automatic Genera-
Regulating Response Associated with Fre- tion Control," Bonneville Power Admin-
quency Bias," presented at the Inter- istration, July 1974.
connected Systems Committee Meeting,
Des Moines, lA, 1956. 24 D. N. Ewart, "Automatic Generation Con-
trol - Performance Under Normal Condi-
14 N. Cohn, "Techniques for Improving the tions," Henniker Conference 1975.
Control of Bulk Power Transfers on Inter-
connected Systems," IEEE PAS, vol. PAS 25 C. W. Ross and K. W. Goff, "An Exper-
90, No.6, Nov/Dec 1971. imental Corr-elation Analyzer for Measuring
System Dynamics," IEEE Transactions on
15 N. Cohn, "The Use of Synchronized Time and Communication and Electronics, vol. 82,
Frequency Standards to Improve Control of 1963.
Interconnected Electric Power Systems,"
presented to North American Power Sys- 26 , "IEEE Standard Definition of
tem Interconnection Committee, Atlanta, Terms for Automatic Generation Control,"
GA. February 1969. IEEE Standard No. 94, IEEE PAS, July/
August 1970.
16 N. Cohn, "Considerations in the Regulation
of Interconnected Areas," IEEE PAS, vol. 27 F. P. deMello, R. J. Mills, and W. F. B'Rells,
PAS 86, No. 12, December 1967. "Automatic Generation Control Part II -
Digital Control Techniques," IEEE PAS,
March/April 1973.
17 N. Cohn, "Research Opportunities in the
Control of Bulk Power and EnergyTransfers 28 F.P. deMello, R. J. Mills, and W. F. B'Rells,
on the Interconnected Systems," EPRI "Automatic Generation Control Part I -
EL-377-SR, February 1977, pp 6-1 through Process Modeling," IEEE PAS, March/April
6-30, Henniker Conference, 1976. 1973.

18 N. Cohn, "Some New Thoughts on Energy [29] C. E. Fosha, Jr., and O. I. Elgerd, "The Mega-
Balancing and Time Correction on Inter- watt-Frequency Control Problem: A New
connected Systems," Power System Con- Approach Via Optimal Control Theory,"
trol workshop, San Louis Obispo, CA, IEEE PAS, vol. PAS 89, No.4, April 1970.
April 1977.

19 N. Cohn, "Area Wide Generation Control - [30] O. I. Elgerd and C. E. Fosha, Jr., "Optimum
A New Method for Interconnected Systems," Megawatt-Frequency Control of Multiarea
Electric Light and Power, June, July, Electric Energy Systems," IEEE PAS, vol.
August 1953. PAS 89, No.4, April 1970.

20 N. Cohn, "Some Aspects of Tieline Bias 31 K. Honda, "A Filtering Method Applied to
Control on Interconnected Power Sys- Load Frequency Control," Electrical Engi-
tems," AlEE Transactions, February 1957. neering in Japan, vol. 93, No.4, 1973.

21 C. Concordia and L. K. Kirdhmayer, "Tieline

Power and Frequency Control of Electric 32 T. W. Jenkens, Jr. and B. Littman, Re-
Power Systems, Part I and II," AlEE Trans- sponse Capability in the Control of Large
actions, June 1953 and April 1954. Generating Units," IEEE Paper 71 CP 73
PWR, IEEE Winter Power Meeting, Jan-
uary 1971.
22 R. L. Cresap, D. N. Scott, and C. W. Taylor,
"Closed Loop Digital Hydroplant Controller 33 C. W. Ross, "Error Adaptive Control Com-
for Automatic Generation Control," Bonne- puter for Interconnected Power Systems,"
ville Power Administration, September 1973. IEEE PAS, vol. PAS 84, No.7, July 1966.

23
34 C. W. Ross, "A Comprehensive Direct 44 K. N. Stanton, "Power System Dynamic
Digital
Load Frequency Controller," IEEE Energy Balance," Proceedings of the Sym-
PICA Conference Record, Pittsburgh, PA, posium on Real-Time Control of Electric
1967. Power Systems, Brown, Boveri, and Com-
pany Limited, Baden, Switzerland, 1976.
[35] D. N. Scott, R. L. Cresap, R. F. Priebe, D. E.
Tebrink, and K. A. Takeuchi, "Closed Loop
Digital Automatic Generation Controller," 45 L. H. Fink, L. M. Smith, and R. P. Schulz,
IEEE Conference paper C 73, July 1973. "Dynamic Mathematical Model of Inter-
connected Electric Power System," Mod-
36 C. Taylor, "Effect of Sampling Rates on eling and Simulation, IEEE/ISA,. vol. 5,
Digital Automatic Generation Control," Pittsburgh, PA., 1974. '
Bonneville Power Administration, Aug-
ust 1974.
46 L. M. Smith, L. H. Fink, and R. P. Schulz, "Use
37 J. Weiss, "Digital Load Frequency Control of Computer Model of Interconnected Power
in Multiarea Power System," Fourth IFAC/ System to Assess Generation Control
IFIP International Conference on Digital Strategies," IEEE PAS, vol. PAS 94, No.5,
Computer Applications to Process Control, Sept/Oct 1975.
Zurich, Switzerland, March 1974.

38 J. L. Willems, "Sensitivity Analysis of the 47 H. G. Kwatny, K. C. Kalnitsky, and A. Bhatt,

Optimum Performance of Conventional "An Optimal Tracking Approach To Load-
Frequency Control," IEEE PAS, vol. PAS 94,
Load-Frequency Control," IEEE PAS, vol.
No.5, Sept/Oct 1975.
PAS 93, 1974.

39 J. Zaborszki, A. K. Subramanian, and K. M. 48 M. Ehsan and M. J. Short, "Digital Simula-

Lu, "Control Interfaces In Generation tion for On-Line Load-Frequency Control
Allocation," Henniker Conference 1975. Studies," IEEE, 1976.

GENERAL AGC PERFORMANCE ADVANCED CONTROL THEORY

40 C. W. Ross and T. A. Green, "Dynamic 49 R. Aggarwal and F. Bergseth, "Large Signal
Performance Evaluation of a Computer Dynamics of Load Frequency Control Sys-
Controlled Electric Power System," IEEE tems and Their Optimization Using Non-
PAS, vol. PAS 91, No.4, Sept - Oct 1972. linear Programming, I and II," IEEE PAS,
vol. PAS 87, No.2, Feb 1968.
41 C. W. Taylor, "Analysis of Dittmer Power
System Frequency Deviation Records," 50 R. K. Cavin, M. C. Budge, Jr., and P.
Bonneville Power Administration, March Rasmussen, "An Optimal Linear Systems
1976. Approach to Load-FrequencyControl," IEEE
PAS, vol. PAS 90, No.6, Nov/Dec 1971.

AGC SIMULATION 51 T. W. Reddoch, P. M. Julich, T. O. Tan, and

E. C. Tacker, "Models and Performance
[42] C. W. Taylor and R. L. Cresap, "Real-
Functionals for LO;;ldFrequency Control in
Time Power System Simulation for Auto- Interconnected Power Systems," IEEE Con-
matic Generation Control," IEEE PAS, vol.
ference on Decision and Control, Miami
PAS-95, No.1, January/February 1976. Beach, FL, December 1971.
43 J. E. Van Ness and R. Rajagopalan, "Effects
of Digital Sampling Rate on System Stabil- 52 J. D. Glover and F. C. Schweppe, "Advanced
ity," PICA Conference Record, Pittsburgh, Load Frequency Control," IEEE PAS, vol.
PA, May 1967. PAS 91, No.5, Sept/Oct 1972.

24
53 S. M. Miniesy and E. V. Bohn, "Optimum and Cybernetics, vol. SMC 3, No.4, July
Load Frequency Continuous Control with 1973.
Unknown Deterministic Power Demand,"
IEEE PAS, vol. PAS 91, No.5, Sept/Oct
1972. 61 T. W. Reddoch, P. M. Julich, G. J. Burchert,
Jr., O. T. Tan, and E. C. Tacker, "On the Use
54 E. V. Bohn and S. M. Miniesy, "Optimum of Advanced Control Theory in the Design of
Load-Frequency Sampled - Data ContrQI Automatic Generation Controllers," IEEE
with Randomly Varying System Disturb- Conference Paper C 73, 101-3, Ja nuary 1973.
ances," IEEE PAS, vol. PAS 91, No.5,
Sept/Oct 1972. 62 V. R. Moorthi and R. P. Aggarwal, "Damping
Effects on Excitation Control in Load-Fre-
55 M. Calovic, "Linear Regulator Design for a quency Control System," Proc. IEEE, vol.
Load and Frequency Control," IEEE PAS 121, No. 11, November 1974.
91, No.6, Nov/Dec 1972.

56 E. C. Tacker, C. C. Lee, T. W. Reddoch, T. O. 63 O. T. Tan, E. C. Tacker, and D. R. Fischer,

Tan, and P. M. Julich, "Optimal Control of "Automatic Generation Control Systems
Interconnected Electric Energy Systems-A Containing Governor Dead Band," IEEE Con-
New Formulation," Proceedings of IEEE, vol ference Paper C 74,127-7, January 1974.
60, No.1 0, October 1972.

57 E. C. Tacker, C. C. Lee, P. M. Julich, and T. O. 64 T. W. Reddoch, "Load Frequency Control

Tan, "Models and Optimal Stochastic Con- Performance Criteria With Reference to the
troller Designs for Interconnected Power Use of Advanced Control Theory," Henniker
Systems, "Region III,IEEE Conference, 1972. Conference, 1975.

58 W. R. Barcelo. "Effect of Power Plant Re- DIGITAL LOAD CONTROL

sponses of Optimum Load Frequency Con-
trol System Design," IEEE PAS, vol. PAS 92,
No.1, Jan/Feb 1973.
[65] W. B. Gish, "Digital Load Control for Hydro-
59 N. G. Malek, O. T. Tan. P. M. Julich:and E. C. electric Powerplants," U. S. Department of
Tacker, "Trajectory-Sensitive Design of the Interior, Bureau of Reclamation, REC-
Load-Frequency Control Systems," Pro- ERC-77-10, August 1977.
ceedings IEEE, vol. 120, No.1 0, October
1973. [66] W. B. Gish, T. R. Whittemore, and U.
Milano, "Grand Coulee Third Interim Con-
60 E. C. Tacker, T. W. Reddoch, O. T. Tan, and troller Prototype Digital Load and Voltage
T. D. Linton, "Automatic Generation Con- Control," U.S. Department of the Interior,
trol of Electric Energy Systems - A Simula- Bureau of Reclamation, REC-ERC-76-8, July
tion Study," IEEE Trans on System, Man, 1976.

25 GPO 848-069
"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .-. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
..
:

...
...
ABSTRACT
.... ABSTRACT

The concept of automatic generation control is often considered .. The concept of automatic generation control is often considered

....
complex because of the"freedom each utility has in choosing indi- complex because of the freedom each utility has in choosing indi-
vidual characteristics within a basic control philosophy. This report vidual characteristics within a basic control philosophy. This report
endeavors to separate the basic philosophy from the individual' endeavors to separate the basic philosophy from the individual
characteristics to allow a clearer understanding
philosophy. Discussions of individual characteristics
of the control
used by the ... characteristics to allow a clearer understanding
philosophy. Discussions of individual characteristics
of the control
used by the
Bureau of Reclamation are also presented. .. Bureau of Reclamation are also presented.
...
.
...
.. .# .:.. ... .....
... ...
...
...
ABSTRACT .... ABSTRACT

The concept of automatic generation control is often considered

complex because of the freedom each utility has in choosing indi-
..... The concept of automatic generation control is often considered

....
complex because of the freedom each utility has in choosing indi-
vidual characteristics within a basic control philosophy. This report vidual characteristics within a basic control philosophy. This report
endeavors to separate the basic philosophy from the individual
characteristics to allow a clearer understanding of the control .... endeavors to separate the basic philosophy from the individual
characteristics to allow a clearer understanding of the control
philosophy. Discussions of individual characteristics used by the .... philosophy. Discussions of individual characteristics used by the

....
Bureau of Reclamation are also presented. Bureau of Reclamation are also presented.

....
...
...
REC-ERC-78-6 REC-ERC-78-6
Gish, W. B. Gish, W. B.
AUTOMATIC GENERATION CONTROL - NOTES AND OBSERVATIONS AUTOMATIC GENERATION CONTROL - NOTES AND OBSERVATIONS
Bur Reclam Rep REC-ERC-78-6, Div Res, Oct 1978. Bureau of Recla- Bur Reclam Rep REC-ERC-78-6, Div Res, Oct 1978. Bureau of Recla-
mation, Denver 25 p., 8 fig, 66 ref mation, Denver 25 p., 8 fig, 66 ref

DESCRIPTORS-/ *Ioad-frequency control/ *interconnected sys- DESCRIPTORS-/ *Ioad-frequency control/ *interconnected sys-
tems/ *power system operations/ *power interchange/ algorithms/ tems/ *power system operations/ *power interchange/ algorithms/
automatic control/ control systems/ governors/ power dispatching/ automatic control/ control systems/ governors/ power dispatching/
baseloads/ generating capacity/ hydroelectric power/electric gen- baseloads/ generating capacity/ hydroelectric power/ electric gen-
erators/ computer applications erators/ computer applications

IDENTIFIERS-/ automatic generation control IDENTIFIERS-/ automatic generation control

COSATI Field/Group 09B COWRR: 0902 COSATI Field/Group 09B COWRR: 0902

REC-ERC-78-6 REC-ERC-78-6
Gish, W. B. Gish, W. B.
AUTOMATIC GENERATION CONTROL - NOTES AND OBSERVATIONS AUTOMATIC GENERATION CONTROL - NOTES AND OBSERVATIONS
Bur Reclam Rep REC-ERC-78-6, Div Res, Oct 1978. Bureau of Recla- Bur Reclam Rep REC-ERC-78-6, Div Res, Oct 1978. Bureau of Recla-
mation, Denver 25 p., 8 fig, 66 ref mation, Denver 25 p., 8 fig, 66 ref

DESCRIPTORS-/ *Ioad-frequency control/ *interconnected sys- DESCRIPTORS-/ *Ioad-frequency control/ *interconnected sys-
tems/ *power system operations/ *power interchange/ algorithms/ tems/ *power system operations/ *power interchange/ algorithms/
automatic control/ control systems/ governors/ power dispatching/ automatic control/ control systems/ governors/ power dispatching/
baseloads/ generating capacity/ hydroelectric power/ electric gen- baseloads/ generating capacity/ hydroelectric power/ electric gen-
erators/ computer applications erators/ computer applications

IDENTIFIERS-/ automatic generation control IDENTIFIERS-/ automatic generation control

COSATI Field/Group 09B COWRR: 0902 COSATI Field/Group 09B COWRR: 0902