POWER SYSTEM OPERATION AND CONTROL

(Core Course Course Code :18EE81)

Typical power system control center (ISO New England Control Room)

Energy management system (EMS) and Business management system (BMS)

 VIII SEMESTER DETAILED SYLLABUS

B. E. ELECTRICAL AND ELECTRONICS ENGINEERING

Choice Based Credit System (CBCS) and Outcome Based Education (OBE)
SEMESTER – VIII

POWER SYSTEM OPERATION AND CONTROL(Core Course)

Course Code 18EE81 CIE Marks 40
Number of Lecture Hours/Week 3:0:0 SEE Marks 60
(L:T:P)
Credits 03 Exam Hours 03
Course Learning Objectives:
• To describe various levels of controls in power systems and the vulnerability of the system.
• To explain components, architecture and configuration of SCADA.
• To explain basic generator control loops, functions of Automatic generation control,
speed governors and mathematical models of Automatic Load Frequency Control
• To explain automatic generation control, voltage and reactive power control in an
interconnected power system.
• To explain reliability and contingency analysis, state estimation and related issues.
Module-1
Introduction: Operating States of Power System, Objectives of Control, Key Concepts of
Reliable Operation, Preventive and Emergency Controls, Energy Management Centers. R1
Supervisory Control and Data acquisition (SCADA): Introduction, components, application in
Power System, basic functions and advantages. Building blocks of SCADA system, components of RTU,
communication subsystem, IED functional block diagram. R2
Classification of SCADA system: Single master–single remote; Single master–multiple RTU; Multiple
master–multiple RTUs; and Single master, multiple submaster, multiple remote. R2
Module-2
Automatic Generation Control (AGC): Introduction, Schematic diagram of load frequency and excitation
voltage regulators of turbo generators, Load frequency control (Single area case), Turbine speed governing
system, Model of speed governing system, Turbine model, Generator load model, Complete block diagram of
representation of load frequency control of an isolated power system, Steady state analysis, Control area
concept, Proportional plus Integral Controller. T1
Module-3
Automatic Generation Control in Interconnected Power system: Two area load frequency control,
Optimal (Two area) load frequency control by state variable, Automatic voltage control, Load
frequency control with generation rate constraints (GRCs), Speed governor dead band and its effect on
AGC, Digital LF Controllers, Decentralized control. T1
Module-4
Control of Voltage and Reactive Power: Introduction, Generation and absorption of reactive power,
Relation between voltage, power and reactive power at a node, Methods of voltage control: i. Injection
of reactive power, Shunt capacitors and reactors, Series capacitors, Synchronous compensators, Series
injection. ii Tap changing transformers. Combined use of tap changing transformers and reactive power
injection, Booster transformers, Phase shift transformers, Voltage collapse. T3
 Module-5
Power System Security: Introduction, Factors affecting power system security, Contingency
Analysis, Linear Sensitivity Factors, AC power flow methods, Contingency Selection and
Ranking. T2
State estimation of Power Systems: Introduction, Linear Least Square Estimation. T2
Course Outcomes: At the end of the course the student will be able to:
• Describe various levels of controls in power systems, architecture and configuration
of SCADA.
• Develop and analyze mathematical models of Automatic Load Frequency Control.
• Develop mathematical model of Automatic Generation Control in Interconnected Power
system
• Discuss the Control of Voltage , Reactive Power and Voltage collapse.
• Explain security, contingency analysis, state estimation of power systems.
Question paper pattern:
• The question paper will have ten questions.
• Each full question is for 20 marks.
• There will be 2 full questions (with a maximum of three sub questions in one full question) from
each module.
• Each full question with sub questions will cover the contents under a module.
• Students will have to answer 5 full questions, selecting one full question from each module.
Text Book

1 Modern Power System Analysis D. P. Kothari McGraw Hill 4th Edition, 2011
2 Power Generation Operation and Allen J Wood etal Wiley 2nd Edition,2003
Control
3 Electric Power Systems B M Weedy, B J Wiley 4th Edition, 2012
Reference Books Cory

1 Computer-Aided Power System G. L. Kusic CRC Press 2nd Edition.2010

Analysis
2 Power System SCADA and Smart Mini S Thom and CRC Press 2015
Grid John D. McDonald
3 Power System Stability and Control Kundur McGraw Hill 8th Reprint, 2009
 ~~

USN I I I II I I I I I ) t8F.E8t

Eighth Semester B.E. Degree Examination. Juty/Auj!USt 2022

Power System Operation and Control

Time: J Ius. Max. MarI<s: 100

Note: Alts .... r ony FIVE full questtons, choosing ONE full qll61;on from toch modult.

Modl11e-l
t a. Explain the operating Slates of power system, with a neat diJIgram. (118M" )
b. Explain the key concepts for reliabk operation of power S)Slem. (06 " )
c. Expllin the major components of cnerg) management center. (06 ,It )

OR
2 a. Explain in brief the component, of remote terminal unit for power system SCAOA.
(118M.rk.)
b. With usual notations, explain following with reference to SCAOA systems. SCADNAGC,
EMS, OMS, LMS, AMR. (06 M.rk.)
c. Explain single master. multipk ,ubma5ler, mukiple remote configuration with a neat
diagT1lm. (06 ",.ru)

\1od"I ....2
J a. Explain in brief the function of load f~uency control and ucitation vohage regulators of
turbo generators with a neat schematic diagram. (06 ",.rk.)
b. Draw the schematic diagram of steam turbine speed governing system and explain the
functions of the various components. (118"'.rk.)
c. Two genennon rated 200MW and 400M W are operating in r3rll1lel. The droop
ehat1lcleristics of their govemors arc 4·. and 5t. mpeeti,el) from no load to full load.
Assuming that the generators are operating .. 50Hz It no load. 110... "ould a load of600MW
be shared between them? What will be the S)Slem frequency at this load? Assume free
governor operation. (06 M.rk,)

OR
4 a. Explain the modeling of
i) Speed go,.mor system model
ii) Turbine mod.l
iii) ~nerator·load model. (Il " )
b. With I neat diagT1lm,explain the proponioll3l plus integral controller. (08 '" '

MOtJu'r:J
S a. Explain state space model of'two area system. (10 M..... )
b. Explain the function of imponant components and their transfer function of automatic
'okage regulator of a generator ...~h a neal diagram. (to M..... I

) of2
 18££81

OR
6 •• Explain the load frequency control with Generution RUle Constrejnts (GRCs) with a neat
diagram. (01Moru)
b Describe the effect of the speed governor dead band on AGe. (01 MOM)
c. Explain the decentralized control in interconnected po we r system. (06 MI'Iu)

~lodult:4
7 • Explain generatIOn and absorplion of reactive po .. er in electrical pow er system. (06 Morb)
b. Derive the equatlons to get the relation bet ween voltage. power ond reactive power at •
node. (06 \toru)
c. Brieflye.plain the different methods of reactive power injection in po .. er system. (08 \1011<.)

OR
8 a. Explain the >arlous methods of 'oltage control using tap changing tran.formers. (06 MOM)
b. What is meant by sub synchronous reactance? Briefly explain. (06 Mortu)
c. Three generating stations are connected to a common bus bat X as sIIo .. n in Fig.Q.8(c). For
a particular S)Slem load. the line ,ohage at the bus bar falls by 2kV Calculate the reatli"
po .. er injeelion required to bring back the vohuge to the original value. All pu values are on
a SOOMV A base. (08 MOM)

iSIU-
:L1-SI'.V r-J ~.~-j

8 .O'If.... X

Fig.Q.8(c)

Modul ....5
9 a. upl.inthe factors affecling po wer ,Y\lem securuy. (001\1" ....)
b. Wilh lhe help of flo .. chan. c'Plain the conlinlOcnt) anal), ts. (08 "ortu)
c. Explain calcul&lion of linear sensili,ity factors .. ith the help of flo .. chan. (08 M,,"')

OR
10 •• Explain major functions involved in sysrem security, (06 Mortu)
b. Explain the formulation and state estimate using linear least square estimation. Also "Kplain
Ihe condition for observabilil) in leas! square estimation. (01 "orb)
c. \\Iilh the help of flow chan. explain lhe AC power no..
5eCurily anal) ••• with conlingency
~se~~. ~Mo~

•••••

20f2
 a
lJS~ L rL1J]]~[J]]
EiJ.!hC" Semester B.E. Dl'J,:n.'c Exumlnatlon, Fcb.l~lar. 2022
Power System Operation and Control

Time: J his. Max. Marks: RO

~otl': Answt'r any FIJ 'Efull questions, choosing 0"'£ full question fram each module,

~Iodlile-I
1 J. Briefly describe the major components of a SCADA system. (08 -"lark')
b. What an: the various transducers used in power system SCAOA? (f}.,t -"brks)
c. Discuss the various options available for communication in SCADA. (0-4 \Iuks)

OR
2 a. Draw the flowchart for the priority list method of unit commitment and explain. (08 \larI..J)
b. Draw and explain the flowchart for the forward dynamic programming algorithm. (08 -"larlu)
"
Module-2
3 a. Explain algorithm for Hydro thermal scheduling using Discrete Time Interval method,
(10 ~larlu)
b. Draw flow chart for 8-A interactions.) (06 ~larI..J)

OR
4 a. What are the functions of AGC?" (04 ~l:arb)
b. Draw the block diagram of steam turbine governing system and explain the functions of the
various components. (08 ~lar~s)
c. What are the two modes of governor operation and explain? ' ..... (04 \tar"',)

Module-3
5 a. Two areas Al and A2 are interconnected by a Tie line T 12. Derive an expression for
frequency change and Tie line power flow when the load in Area 1 changes. (03 "ar~,)
b. Two areas I and 2 are interconnected. The capacity of area 1 is 1500 M\\' and area 2 is
500 MW. The incrementalregulation and damping torque coefficient for each area on its
own base are 0.2 pu and 0.9 pu respectively. Find the steady state frequency and change in
steady-state the line power, for an increase of 60 MW an area 1. Nominal frequency is
50 Hz. (08 \luk,,)

OR
6 a. Prove that by adding a feedback of proportional integral controller to ALFC, the steady state
frequency deviation is zero. (08 \larks)
b. A control area has following data, total generation capacity :::; 2000 MW, normal
load= 1500 MW, II = 4.8s. ():::; 1.2%, f= 50 liz, R:::; 2.5 1l1JpU M\V.
i) Determine primary AI.Fe parameter.
ii) For increase of 0,.02 pu unload, find frequency drop without governor controL
iii) With governor control. (08 Marks)
,; .... ..
~'

. Module-4
7 a. Explain the different methods of voltage control by reactive power injection. (08 Marks)
 15EE81

b. Three general mg ~tallon' are connected to a (0Il11110n hut; bar and as shown in Fig.Q7(b).
For a parncular 'i),tcrn Il)aJ the line voltage at bUi x f.lIl" by 5 KV. Calculate the reactive
power mjccuon required to brmg hack the voltage to the ongmal value. 1\11 pu values arc on
a base of500 ~tVA. (08 :\tark~)
It jo" rIA
2.~~vHt
Fig. Q7(b)

OR
8 a. Explain voltage control using; tap changing transformers, Booster transformers and phase
shifting transformers. (08 'fark,)
b. A 415 V, 50 Hz 3~ system delivers 500 KW at 0.8 p. f. lag. Shunt capacitors arc installed to
improve the p.f to 0.92. Determine the value of capacitors needed if the capacitor bank is
star connected. (08 "arks)

MQdule-5
9 a. With a neat flow chart, explain contingency analysis for generation outage using generation
shift sensitivity factors. (08 'larks)
b. Explain the formulation and state estimate using linear least square estimation. Also explain
the condition for observability in least square estimates. (08 'tar~)

OR
lOa. With a neat flow chart, explain contingency analysis for line outage, usmg line outage
distribution factors. ,(08 'tar"s)
b. Explain I PI Q method for contingency Ranking. Also explain contingency processing using
AC load flow analysis with a flow chart. (08 '18r"\)

fr ••••
USN I '-1 '-1
r-- -r-i I--r---r-L--rI~_J-
__ [r--j-r--_J· ~~I ~7EE81
"
Eighth Semester H.E. Ucgn'c Examlnatlon, Fcb.l.\lar. 2022
Power System Operation and Control
Time: 3 hrs. Max. ~fiHks: J 00
Sate: Answer any FIVE full questions:'~~oos;"g O;\,1::lullquestion from

~ 'Module-l
, "
~QC" mOilu/t'.

1 3. Explain the operating states.of power system with a neat diagram. showing [he tri1:.s l:WO
be~wcen the s~atcs. ". '¥ , .. ' (10If.utt)
b. \\llth a neat diagram, d?c,n~e the major component\of~CADA system. (10\Iul.)
1iI~ , ·io ~. ~

\ It" OR ..
2 a. Discuss about the' commonly used preventive- and emergency control measures in a power
system. I' "/ \ ~ __) (10~br1u)
b.
,

Jf>","tIl-
,

~.;./
.
Explain thevarious constraints to be considered in unit commitment. (10\brir..I)
.fI'

<: '~'"I' 11\1qdule-2 r»

3 a. Explain f!1eshort term hydrothermal scheduling using y-A. itcnltion method. (10 \b~J)
b. Two machines operate in parallel+to supply a load o( 400 MW. The capacities of the
machines are 200 MW and SOO MW. Each' has}a"droop characteristic of 4%. ThetT
Governors are adjusted .$0.
that the frequency {1".1.00% on full load. Calculate the load
supplied by each unit and 'the';'frequencyat this load. The system is a/'50 Hz system.
. ( ) - / (to .\brks)
~ <. li
r\ O¥-)
('
~. ,,,,,J
4 a. With a neat block diagram, explain the basic generator control loops. (10 "a,u)
b. Explain the following terms used in.AGC:' ~
(i) Contrpl area ,(ii)' Tie line (' (iii) Net interchange
(iv) Stati~ control error (v) Unit control error . ) (10~larir.~)
.,' ... ,- 'If ;. 1'~
\ ( t,. .4
.. - "" Module-3, C::" I

5 a. Deriv~ an expression forg~~erator model of ALfC system. _ (to \brb)

b. A 1000 MVA generator operates on full.load 'at the rated frequency of SO lIZ. The I\)-w ts
$

,.( ·'reduced to 800 ¥~:JJhe steam valve haS'tan operating time lag of 0.6 s. If l( 5 sec,

.
!e

determine the chang~m frequency.L. ~ (10\lar\,)

~ ~ \
{t',,, ~,,/
. \. " .J.!..r\ .. " OR
. t' :".
6 a. Denve the state space model of an Isolated AGC system. (to ~l.tfkJ)
b. Two generating areas have capacities of 500 MW and 1000 M\\' f':!'lX"\:tl\dy. They an:
interconnected by a short line,;The percentage speed regulations 1'1'\)(1\ no...k,aJ to full-lead of
the two stations are 3% and 4% respectively. If the 10t,,1 on ench station is ~50 MW. find the
power generation of each station and the tic-line power,
i
.: \
(II) \brir.,)

J '., ...1 ~IO(h' lew"

7 a. Discuss about the issues related to AGe implementation. (10 MarkJ)
b. Two areas of equal capacity of25oo MW, speed regulation R '""'3 Ilz/per MW and II = 5 sec
are connected by a tie-line of capacity 250 MW. operating at 45°. Find the frequency of the
line pow~.roscillations. (10 Marks)
10f2
 .{
171~EHI
_""" .
I! ._.~"
. , ..' OR '(~ -,"
8 t. .
a. Prove that voltage at receiving end is dependent on reactive power 10 power system.
" Jdo." (10 Marks)
b. A 440 V, 3<j)distribution feeder has a load of 75/KW drawing a current of 130 A. A
capacitor rating of 45 KV Ar is connected across th~ load. Determine the'
(~~ pf and reactive load before compensati0l!~' w ~\. .J
(11) pf after compensation. 4 ~ \. (10 Marks)

. ( ~1dule-5 r: .I
9 a. WIth a neat flow chart, explain contingency analysis for .... [ine.Joutagc using line outage
distribution factor. ,.. <...
(10 Marks)
b.
,
List and explain the major functions involved in system security.
("
(10 Marks)

, • OR ~~~
lOa. Discuss the set of state :vanables and measuremen~"ofJa conventional state estimator.
. ( ~ , (10 Marks)
b. LIst and explain. the other issues in state estimation. (10 Marks)
x-
*****

,
~(;
~?-~.

\(; "
.l~. 'I. ~
2of2
_
 "I) --~-' /" // ... ',,_-
/.... ~~~ r11)

/t'·-;~'- ~~'{ .Jf.)

USN i""--i""--i"'--ir--1ir--1i'---'i'---'i'---'I'---'Ir-I1 t';d'>"'!,:;V>' ,/. \/~~~~ "!}~i':1
Eighth Semester B.E. Degree Examinpti~n; July/August 2021
Power System Operat,on ..~nd Control
,~ \
Time: 3 hrs. 'frj Max. Marks: 100
Note: Answer ~nr Flr~ full questions. ~/'..J
1 a. What are the states of power system, explain in brief with a suitable diagram. (06 Marks)
b. With usual notations, explain following with reference to SCADA systems.
SCADNAGC, EMS, DMS, LMS, AMR, I' (06 ~Iarks)
c. Explain the constraints in UC, r It- / (08 Marks)
( l.~
~~ ,
2 a. , t hee mai
E xp 1am I ~0 f Energy Management
major components ,. \Center. (06 Marks)
b. Explain in brief the c~mponents of Remote Temlin~1 Unit for power system SCADA.
(',. 4' , (07 Marks)
c. Explain with Flow Chart the Dynamic Programming Method of unit commitment. (07 Marks)
( \V
3 a. . a no/t'e~'S
W nte on c he d u I'mg 0 fH yd rosystems
4. , h necessary &:
WIt rormu lae
ae imvo 1ve d iinto It.
,

' { ~~h M h . I&:" (I ). D' " . ~,4.1 . hmi ' (05 Mdar~)
b. E xp IaID"".teat ematica 19E1luation, lScnhzatlo~.1 ft gont c steps mvo 1ve m
Discrete Time Interval Method of Hydro Thermal ScHeduling. (10 Marks)
c. Explain the need for Automatic Generation Control (~GC) in power 'system operation and

4 a,
control.
,J.
~( ~ )
-'.
~
~ -J y
• ....,
~.'\
...~~\

.
v~
...
~~
Write a brief note" features of hydlo~w'er plants that p.articipate in Hydrothermal
.
(05 Marks)

,
J '\'~
Scheduling, '\ ...,;) • • '9 ~~.?"'\ (05 Marks)
b. Explain with a suitable Flow chart the/short Term Hydrothermal Scheduling using y - A.
Iterations. , ~ l ' (10 Marks)
c. Explain tfi'eBasic Generator COntrol loops with reference to AGC in PSOC. (05 Marks)
\'~ (~
~i'~~<i ("It V' -;' ,~
5 a. Obtain the Mathematical.Model ALFC components Speed Governor, Turbine. (10 Marks)
b. ~:.-~ain the TransfeJ;, ~1lion of a A<lC~with Integral controller from its relevant block
~1 ".diagram representation of ALFC.
; ~~
~
\ , (10 Marks)

t': i::) A. ( )
6 a. Analyse the eff<}Ctsof changes in loads of two area ALFC system with primary loop.
( '" (10 Marks)
b. Obtain the s ate space Model of an Isolated system. (10 Marks)

A~·.)
7 a. Explain the state space Model for Two-Area ALFC system. (12 Marks)
b. Explain in brief the ~ssues related in AGC implementation. (08 Marks)
-,, .,)
, :,
.l>
 17EE81

8 a. With a suitable assumptions made in T~o-Atea ALFC system, obtain the Tie-line
(10 Marks)
b.
oscillations formula and analyse different ,.damping conditions.
Two Area ALFC control system has follows ~data Area:
i) AreaCD: RI = O.lpu,
.

DI = 0.8pU; .MVAlrated = 1500

v
~~;
~,~

ii) Area®: R2 = 0.098pu, D2('= O.9pu MY A2rated= 500 C ,

In Area - 1 Load increase = 100m . Find steady state fre uetlcy and Tie- line power flow
change. . . ~, (06 Marks)
c. Write an explanatory note on production absorption. of reactive power and listout the
methods of voltage contra in power system operation and controls. (04 Marks)
~
\,...../
~ ~

9 a. Explain the powe(syftem reliability a~d system 'security levels. (10 Marks~
b. Write a note on~iability cost, LOLE, L~~!1:§~LOLF, and LOLD. (10 Marks)

10 a.
~' ~
With ~,~t~ble flow chart explain,the c~ntingency analysis procedure. (10 Marks)
b. What are tire state variables, mea~urements involved in state estimator, explain in brief state
estimaiion problem formulation. "\ ....'Y'. (10 Marks)
, r"
~(;
 ~~~

USN II I I I I I I I I I ;'.
r. 4"- 15EE81
Eighth Semester D.E. Degree Exalnin~~fo~'JuIY/August 2021
Power System Operatlon'and Control
Time: 3 hrs.
I
,/r~ -...,. Max. Marks: t, ... 80
Note: Answer {IllY FIVE /J/I questions.
1 a.
What are the different operating states ofpowcr system? Explain thftransition that can take
place from one state to another state, with a block diagram, ' '.. (08 Marks)
b. What is SCADA? Explain the' components of SCADA zwith a general SCADA
configuration. (~" r C\ ,.
(08 Marks)
l..
V '\ ._,.
~-ta
c
'Jr .:.~

2 a. List the seven key concepts proPosed by NERC. Explain'any two. (08 Marks)
b. What is a unit commitment problem? Draw and explain the flow chart for forward dynamic
programming method. ..J 't ~.- (08 Marks)
~ ; ,
3 a. · th e a Igont
E xp Iam ('~ "h"
' fcor hyd rot h erma I.sc~h(d"
m inz diiscrete time
e u)1'mg usmg " mterva 1 met.h0 d .
/ ... ') .... (10 Marks)
b. Explain the basic control loop of a generator with a schematic diagram. (06 Marks)
-<" V \.; .~
J.~ ~:\ §
4 a. Draw the block diagram of steam turbine govet.~ystem and explain the functions of
various components. I C)
t4;::'}..J .:\ U (08 Marks)
b. Explain the followingterms . •W A .
i) Control Area -~~ Ii 'l~'• ;
ii) Tie - line r: '
iii) Area Con roJ.,Error (ACE). It;. • ~
r~, (; (08 Marks)
~
.,
~
'\(V
- "y
~
.,
5 a. Obtain the transfer function for.the complete ALF}.J~system (08 Marks)
b. A control area has the folloring"'data: ~~
Total generation capaci!~~ rlOOOMW, No~l load = lS00MW, H = 4.8s, D = 1.2%,
f = S.oHz, R = 2.5HZ/p~
I i) Determine the pnmary ALFC parameters
~Y"l r
,Iw\ jl) For an incr~'of O.02pu in tlie,,-ioad, find the frequency drop without and with
, governor ~o1.' (08 Marks)

6 a.
r~.' -,C~
Obtain the ·bloc~diagram representation of a two area system, Wit pnmary contro .
.h . 1
( tJ'~ (08 Marks)
b. Obtain the state space motel or a single area system. (08 Marks)
'\~:V
~~ _
.... I

~
7 a, Explain tie line oscillations in a 2-area system. (10 Mar~s)
b. Explain briefly the components of a power system that can generate and lor absorb reactive
power. '\ ...1 (06 Marks)
........ "'~
 15EE81

8 a. Show that the real power flow between nodes is proportional to the transmission angle
'8' and reactive power flow is proportional to the scalar voltage difference between the two
nodes. < i).;' (08 Marks)
b. Explain the method of voltage control by reactive power injection, (08 Marks)
'.,f- .:)
C~. ..",
..J
9 a. Explain the Security Constrained Optimal Power Flow (SCOPF) function of power system
security with an example. f r" ~.
\.-,~ (08 Marks)
b. Explain contingency analysis using a flow 'chart. (08 Marks)
c f"Y if

V A.'
\0
10 a. What is state estimation in £'ower system? Discuss need and importance of state estimation.
. . #\ ,. .. r _"

(06 Marks)
b. Explain the Weighted Least Square EsttmattonJ\(WtSE) method of power system state
estimation. ~.. '. (10 Marks)
.,
A.()
.... Ill,•...
~~
.,
.:

~(; §~
., c

,
~(; .~
~.
~~

,
~

4(; S·
~ ~(;
.; ~
,
~v
, rlI"
~(J
,
~

~(; A~ ~(J
~ ,
Co ~
~CJ·
,
~

~(;
,
20f2
~(;
 -- ~ w .......

USN I I I I I I I I I I I
•
''.:'~/
.,/. - ~ t-
ISEE8I
EIghth Semester B.E. Degree Exanlination, November 2020
Power System Operation' arid Control
~
F
.. til
'

Time: 3 hrs. ( .j
Max. Marks: 80
.. ' "

Note: Answer any FIVE full questions irrespective of modules.

, .,.,1

Module-I
1 a, W~lat are the objectives of power system control? Explain. It (06 Marks)
b. WIth ~ neat flow-chart, explain forward dynamic programming method of solving unit
commitment problem. . (10 Marks)

2 a. With a neat diagram explain the general configuration and major components of SCADA
#
syste~. :'-r '. (08 ~larks)
b. Explam the key concepts for reliable operation of power system. (08 Marks)
..,J
../
.....
.
. ,/ (}~. Module-2
3 a. Explam t~~.ge~eral algorithm for hydro-thermal scheduling. (08 Marks)
b. Two generators rated 200MW and" 400MW are operating in parallel. The droop
characteristics of their governors ~ 4% and 5% respectively from no-load to full-load. The
speed s~t point are such that th~~en'erators operate at' ~OHz when sharing the full load of
600MW- in proportion to their ratings, i) If the load reduced to 400MW, how is it shared ?
At what frequency will system"operate? ii) If rio~ the speed chanCers. are reset so that the
load of 400MW IS shared.hr 50Hz in proPbrtfon to their rating, what are the no-load
.1> \
frequency now? "".,.~ ".,. • v' .~
(08 Marks)
< •

4 a.
C/ .irV
A steam plant.ilid a hydro plant supply a~load of 500MW'for 12h and 300MW for 12h in a
\('/
day. The ther;;mt~lant characteristics are given by ;-
F(PG;) = O.06'i'P~T+ 40PGT + 100 u~itlcoslh . (' ,
The hyd~o plant characteristic i~given by .\ ....
~
. \ 2 3/ ~I
Q = O.003POH+ O.5PGHm s : 0-'"
~""t:". ~< ... : .t 2 ./",...
r,The loss is given by PLdss,,=O.OOlPoH "...'. .)
• ~',1" .
.,' "The value of y is 80. Find the scheduling of power and the total discharge. Also determine
, the daily operating cost of thermal plant and the water used daily by the hydro plant. Obtain
the schedule: i) Neglecting lossesr" ii) Considering losses. (10 Marks)
b. Explain d~ef'e_'f modes of go~ef~r Jperation. (06 Marks)

\\J .......
"Module-3
5 a. Derive the state space m?ge10f an iso lated AGC system. (08 Mark~)
h. A 1OOOMYA generator· operates on full load at the rated frequency of 501Iz. The load IS
reduced to 800MW. "lhe steam valve has an operating time lag of 0.65. If II = 5sec,
determine the change in the frequency. (08 Marks)
~ J
'.
lof2
 ."
6 a. Derive an expression for tic-line power and frequency deviation for two area system.
• I (10 Marks)
b. A system consists of four identical I OOMVAgenerators feeding a total load of250MW. The
inertia constant II = 5 for each machine on its own base. The load varies by 1.2% for a 1%
change in frequency. If there is a drop of 10MW of load, determine the speed deviation and
plot it. (06 Marks)

Moclulc-4
7 a. Explain state space model for two area system. . (08 Marks)
b. Three generating stations are connected to a common bus bar X, as show in Fig.Q.7(b). For
a particular system load, the line voltage at the bus bar falls" by 2kY. Calculate the reactive
power injection required 2kY.(All pu values are on 500MVA base to bring back the voltage
to the original value. {.,'" (08 Marks)

j 0,1 f'"

T .~,1;k"
~.
. I"'\I. -•

Fig.Q..~(b)\ v ,,()
~)'
... . ,
)

~ ~ lfr '" ~
8 a. Explain briefly the various elements of power system that can generate or absorb reactive
power. -\ \ ) . t.-: . (08 Marks) ~/'\··W
b. Two areas 1 and 2 are interconn,ecfed:fThe capacity of area 1 is 1500MW and area 2 is
500MW. The, incremental regulation and damping ..toni'ue coefficient for each area on its
own base are O.2pu and O.9pu ~e~p;ctively. Find.thesteady state frequency and change in'
steady-'statJ tie-line power.rfor an increase of 60MW in area 1. The nominal frequency is
50HZ.·~~ ~.~ ~'"' $ (08 Marks)

•'
._" $
i>.. .,J ·.r
(-, '.J
~'
r .
Module-5'
9 a. \ Explain briefly various\ecurity levels of Energy Management System (EMS). (08 Marks)
b. 'Explain the formulati6n and state e~timate using linear least square estimation. Also explain
the conditio~ for observability in, l~~;; square estimation. (08 Marks)
C: ":J ..t,.;,\
10 a.
.
<, .)
~
Explain major functions involved in system security. (08 Marks)
b. With a neat flow chart, explain contingency analysis for the line outage, using line outage
distribution factor. . '. (08 Marks)

.....
.( l' ) '" '" .........
,\ "

2of2
 Eighth Semester B.E. Degree Examination, Aug./Scpt.2020
Power System Operation and Control

Time: 3 hrs. Max. Marks: 80

Note: i) For Regular Students: Allswer any FI VE f,,11 questions irrespective of modules.
ii) For Arrear Students: Allswer allY FIVE ful! questions, choosing ONE full question
from each module. "..

Module-I
a. Briefly describe the major components ofa SCADA system. (08 \larks,
b. What are the various transducers used in power system SCADA? «()4 ~1arkl)
c. Discuss the various options available for communication in SCADA. (04 \larks,

2 a. Draw the flowchart for the priority list method cfunit commitment and explain. (08 \Iarks)
b. Draw and explain the flowchart for the forward dynamic programming algorithm. (08 )1ar1.S)
-,
,
Module-2
3 a. Explain algorithm for hydro thermal scheduling using Discrete Time Interval method.
~ , " (10 \Ia~s)
b. Draw flow chart for 8-"- interactions.
., (06 ~1a~)

../
4 a. What are the functions of AGe? (04 'lar1.s)
b. Draw the block diagram of steam turbine governing system and explain the functions of the
various components. •_ I
<' (08 'boo)
c. What are the two modes of governor operation and explain. (04 ,Jan., •
•
Module-3
5 a. Derive the transfer function for the complete ALFC block:' (08 ,larks)
b. Two generators rated 1000 MWand 500 MW are operating on parallel with a droop of5e~
and 4% respectively. The frequency in I PU, 50 HZ at no-load. How is a load of 800 MW
shared between them? At what frequency? ~ (08 ~larks)
("I·";~.) r
6 a. Draw the block diagram ofa two area system with primary control loop. (08 'Jan.s)
b. The data of a two area system are as folloWs,
Area 1: PGI = 1000 MW, RI = 0.015, DI;= 0
Area 2: PG2 = 10600 MW, R2 = 0.0015, D2 = 0
An increase of 10 MW takes place in areal. Determine the change in frequency, ACE and
the appropriate control action. -_ . (08 'Ian.~)
:1'
,;'-. ,,Module-4
7 a. Two control areas of capacity 1500 MW and 10000 MW are interconnected through the
tie-line. The parameters of each area on its own capacity are R = 1 Ilz/'PU~tW and
D = 0.02 PUMWlHz. There is an increase of 200 MW. In load of area 2. Determine the
steady state frequency deviation and change in tie-line power. (08 ~lark5)
b. What are the tie-line oscillations? What determines the frequency of these oscillations?
(08 'la .....s)

1 of2
 S II, I~~pl:till I.!I.'IWl'lIliOIlund IIhSllrplinll O"I'l.'lIl'livc p(lwer ill clcclricul power system. (0(, Marks
h. J ~!l'IlCntrillg Slllliolls 111'1.' 1.:1)1IIIl'cIl.'d II) II common bus-bar X, as shown on Fig.QR(b) for
p:lnil.'ular system l..liId. the line \,l)ltag~ at tile hWI bur nllls by 2 K V. Calculate the reactiv
I'll\\l'1' injl''':( ion I\.'qllil'l'd In hrillg hark the voltage to the original value. All PU values arc o
u 500 '" V t\ has ...'.

'=I (' KV

~.
p.> _L.JO'IJU

1.1 ~ ~oJ
Fig.Q8(b) (10 Marks)
\
:~
','
,
.' ." ..
'l" ' .;..
Modulc-5 '"
I,~

9 a, Explain the factors affecting power system security. (06 Marks)

b. With the help oftlow chart, explain .the contingency analysis. (10 Marks)
,,.. , ... ~\

10 a. Explain calculation oflinear sensitivity factor and contingency ranking. (08 Marks)
b, What arc state variables? (02 Marks)
c, Describe the D.C. State estimator. (06 Marks)
A$~~' "
* * * * lie""
.r
.(
\
,',~
"0

•.•
.'
1
&
, ',1.
·V

,"

~,,,,
'\~"}""
.~.~
 , \
f ' \

I I I I lJ~[LIJJ
, #.',

USN 'V~i' I ,,~ J51-:1-:8J

""'1·· I " ./
I {'it"
....:::. ',,~'
I ~ ~- ..'

Eighth Semester B.E. Degree Exarnination/Ucc.20J91.Jan.2020

\II
Power System Operation and Control
)
Time: 3 hrs.
{ .
... ,,)
(
Max .•Marks: 80

Note: Answer any FIVE full questions, choojlng ONE full l/ue.\lion/ro,,( each module.
I ' ~
1''' .
. Modulc-l ,
1 a. With a neat diagram describe the major components ofSC{\6'A·.systcm. (03 ~brkJ)
b. Explain the spinning reserve arnI thermal constraints in u"it'w~mjtment. (04 ~rarkl)
c. Discuss on different eme~cncy control action initiated In a power system to prevent
degradation of system. I " J' • r; (04 ~farkl)
f

f
, OJ{
2 a. With a flow chart explain the priority list method of unit commitment. (08.\lark,)
b. What are the diffe)cnt states in which pow,~ ~stem operates? Explain. (08 ~fark.s)
'....
L: " ..
/' .J Module-2
3 a. Deduce an.expression for gradie!l~tor in hydrother~l,~heduling based on discrete time
interval, ~....\ / ." (08 ~brk~)
b. Two generators rated 200 ~.1\V·and 400 MW." are,Yoperating in paralleL The droop
characteristics of their governor are 4% and 5%).esptctively from r6'load to full ioad, The
speed set points are such-truit the generators o~erate at 50Hz w~; s~ring the full load of
600 MW in proportionto their ratings: (to. C! ...,/ , :.do

(i) If load reduces to 400 MW, who js", thC)]oad shared.jat W"hatfrequency will system
operate. (' , • 'J' <V
(
,
«: )
,
r
«1'
~)
-
(ii) If speed changer are reset so thalJocfd of 400 MW i~,shared at 50 Hz in proportion to
their rating' what is the no load.!~ency now?

OR
,
~~('.
k:">i,,,
'-.
(08 ~1arks)

4 a. Ded6ce . an expression for, hyCIro power generation and thermal power generation m
lambda-gamma technique 'of hydro thermalschtduling. (09 ~lar~s)
b. Two machines operate 111-' paraJJel to suPPly a load of 400 MW. The capacities of the
" >' machines are 200 M~)'and 500 MW. E;c}t has a droop characteristic of 4%. !beir governo~s
are adjusted so that frequency is 100% .on full load. Calculate the load supphcd by each unit
and the frequenc;: ai'this load. Thesystem is a 50 Hz system. (07 ~larks)
,i'\ '\ ..~
" \ ,1,~\
y " "Modulc-3 .
5 a. Two areas AI and A2 are interconnected by a Tie line T12. Derive an expression for
frequency change and JiJ )in~ power flow when the load in Area Ichanges. (08 J\lark~)
b. Two areas 1 and 2 are interconnected. The capacity 0 f area I is 1500 MW and area 2 .IS
500 MW. The incremental regulation and damping torque coefficient for each area on I~S
own base are 0;2~pu and 0.9 pu respectively. Find the steady state frequency and change I,n
steady-state the" Hnc power, for an increase of 60 MW in area 1. Nominal frequency IS
50 Hz. '~ (08 Marks)
~
 I'
Ol{ I ,

6 n. Prove tluu by Hddil\~ II Il'\,'dhur" \It'pI'OPOJ'liolllll illll.'l',I'tll,~oll'l'OlIcr to AlXC, the steady state
frequency lk\'illt i~\I1 it' ~I.'I\\. . (08 Marks)
h. A control urea hlls 1~)II'l\villg dutu, 101111 gcucratlon cupucity - 2000 MW, normal
loud= 1500 ~I\V.II· ·LXs,l> I.~%, r
~O 11/., R .2,511:dpli MW.
(i) Determine priuuiry t\ I.F< plll'lll\l~""1' I

(ii) For increase l)I'O.01 pu unloud, lilld trcqucncy drop without governor control "
<iii) With governor control. ( ) (08 Mark~)
• J

,Mnduk-4 .
7 a. Highlight the event of tie line oscillat ion in inter connected powcr system by deriving
necessary equat ion. (' <'" ~, (08 Mar~s) . I \ "",) •

b. At a 34>, II kV bus, a load dr..lwlng (2 + J I) MVA IS connected. The 11 kV bus IS supplied

from a radial line. Total sy~telilreactance is 0.5 a/phase. Calculate the:
(i) Receiving end CUJ''(CQL ' .I )
(ii) Regulation " ". ~. ~~~ .,-,
(iii) Sending end ~ottngc
(iv) Short circ~lit'capncity of the system ( ,
Assume system to be loss less. '\~~.) (OSMarks)
4' r ~~.,\
¥~ ~()I{
8 a. Prove that voltage at receiving end, is dependent on reactiye power in power system.
...,. \' J .(~. t ' (08 Marks)
b. Two control area of capacity 1500 MW and 10000 MW a're interconnected through the line.
The parameters of each area on its own capacity are R =. 1 Hz/pu MW and
D = 0.02 pu MWlHz. Thcre~is an increase of290 MW in load of area 2. Determine steady
state frequency deviati~. ~nd change in the V~p.pwer.
,,~
.\ "~ (08 Marks)
.
(),;: .,. ~
;J
.
c
:;-

~ Mody!e~5
9 a. With an exarnple[ejplain how secur~~~ constrained OPtio,~~~o~er flow is implemented.
~~ .(~'!i_ioiJ' (OS Marks)
h. Explain syst~ adequacy and syste~ ecurity in reliability analysis of power system.
r: ~~ ~ (04 Marks)
c. Discuss on}najor issues of state estimation. ..I.'~ ~ (07 Marks)
,L-r ~ h,)_, , ..~~
, (!\~ OR.. '
10 a.. "W~h.a flow chart ~xpla!tt.conting~ncy a?atys,i~ for generator outage. (08 Marks)
~. ~\Obtam ~n e~press\o!\ f~r state estimator problem by weighted least square technique in DC
"state esttmation. ~''14 V, (08 Marks)

A
," ,;: ••••
_.\q,,,,,,,,~
~ \ h"o~

.i~,
,:,) I
 ~ // .~~ "~':':':;:':'--"<")"-: \

II' '," . f~r"rCi/,; ...,if

//" .: '\ I" I :
-'I '.:;/ r-,

II IIIIIIII
'II

USN I__ i\~'''.

(:;§~~~4~)',fl5EE
fli. ~ ~"...\. ;,;Y

Eighth Semester D.E. Degree Exami~ation, JI{ner.l~iy2019

Power System Operatlon.ii'nd Control
<\
Time: 3 hrs. ~ Max. Marks: 80

Note: Answer any FIVE full questions, ~ ••ONE full

cbof)slpg r
question f(ollJieach module.
\.:) ~
Wodule-l . . , . ..
1 a. Explain the operating states of PQ\yersystem, with a neatrdiagram showing the transition
between the states. ~~ ..." ..... '\V (08 Mark.)
b. Explain the algorithm Ofpr\O~llst method of unit COIT~lltment. (08 Marks)

" ~, OR (
2 a. With a neat diagram~l'm the general configul!tiin and major components of SCADA
system. , (08 -'IarkJ)
h. Explain the V~~DStraints to be CODSi't)'in unit commitment. (08Mo.ks)

3
L~'
a. With mat~tical
M.t~e-2
model and constraint, explain r-A iterative algorithm for short term
hydrotQeI'lJ)81scheduling. , ) ~ (10 MarkJ)
b. Two S}'hchronous generators OJl~le-'in parallel to sup~~~_!oad of 400 MW. The capacities
of the machines are 200 MW aniSOO MW. Eac.l_b@9i droop characteristics of 4%. Their
governors are adjusted s~ t~t the frequency ~ ,,00910 on full l~) Calculate the load
supplied by each unit :i~quency at this loa~e system is SO~system (06 Marks)

itj. ;
4 a. A two plant Syst~th a hydra! plant ~ thermal plan\~ following characteristics.
The fuel cos~acteristic of the~"'lant is FT =1"OI\rr +O.04P~Rs/hr. The water
dis~harge cwa~teristics of hY~lant is ~ =1SP&H + O.004P~ m3/sec. The constant
which C@Vyrts mcremental ...~ discharge to IDc{_ntal plant cost y is 4.1 x 10-4Rs/m'
I
and ~7(fRslMWhr, B~ ~2S MW . De~e the generation of each plant, the load
on the system and losses. ~, (08 !\Iarks)
b. Expl~in the following ~ used in AGe· C \
,,'i1)Control area ~ ," ~\ 'We line
''iii) Net intercha~'" ~~ Station control error (08 :\larks)

. t:l.! '\C'»odUle-3
5 a. Denve t~1n~rator model, foad model and combined generator load model of ALFC
system. ~ fI; (07 Marks)
b. Two control areas are conpected via a tie line with the following characteristics:
Area 1 : RI = 1%, DI 1.,'b.~g~~aseMVA : SOO
Area 2 : RI = 2%, D = [-:-0, base MV A : SOO
A lo~d change 0B
OO~~ occurs ~nArea 1. Find the new steady state frequency, change in
the lme flow an~~nge In generatton of each area if the nominal frequency is 50 Hz.
~ (09 Marks)

lof2
 15EES)

OR ~
6 A single area consists of two generators with following d,aG}
a.
Gl:200MW R,=4%(onmachinebase) r ~
G2 : 400 MW R, = 5% (on machine base) ~'~ . .
They are connected in parallel and share a load<o\ ~OOMW in proportion to .their ratID?s, at
50 Hz. If 200 MW of load is tripped, what-is the generation by each unit? What IS the
frequency at new load is D = 1.5 pu (on a Q3S of200 MW). Choose a base of 200 M\V.
Als~ find the increase in loa~ due to frequen.v~ (; (08 :\Iarks)
b. Denve the state model of an Isolated A C ystem. ~ (08 :\Iarks)

Module-4 f '
7 a. Explain the different methods 0 v !tage control by reactiv,PQjVer injection. (08 :\Iarks)
b. Three generating stations UJ.~~ected to a common bus.bar and as shown in Fig.Q'Zib).
For a particular system load the line voltage at bus x f~ls by 5 KV. Calculate the reactive
power injection requiredlo Sdng back the voltage Ce original value. All pu values are on
a base of 500 MY A. 'I ~.
, It jo'\ Q).1l- r~ .)
~1pvH
~; '_1~V ~ , ~ t;oA- Y.
B JO~J
~Flg.Q7(b) (08 :\(arks)

8 a. Exp Iam ~r. \'

. vo Itage contro ~1J)g; tap c hangmg
O~ ~ormers,
~ Boos~t~~~.,. (.U...,formersand phase
shifting transformers. ~ • (08 :\l:arks)
b. A 415 V, 50 Hz 34>system delivers 500A~ 0.8 p.f ~ capacitors are installed to
improve the p.f.(J.92. Determme the;v~e of capac Ito ed if the capacitor bank is
star connecte~ S· , (~8 Marks)

. , ~OdUIe-5 r
9 a. With ~~ flow chart, exp!l~ntingency ana~-ffir generation outage using generation
shift sett§itivity factors. ~~ - (08 ~Iarl;
b.
e
Explain the formulati~~ ~tate
Condition for o~
estm:~~t
in least ~timates.
linear least square estimation. Also exp
(OS)Iarb)

~ ~~' OR
lOa. With a neat flo~hart, explain ~~t~gency analysis for line outage using line outage
distri~utiO!~rs. L \ 'Itt.) . - , (08 )(ark5)
b. Explam 1.. ethod for con~ncy Ranking. Also explain contingency processing using
AC load flo analysis wi~ a.flow chart.

~...,,; ..... (08~'larks)

20f2
 Introduction to Power
1 System Operation and
Control

LEARNING OBJECTIVES

After a careful study of the chapter, you would be able to understand:

Basic structure of the power system. .Case studies on some major blackouts.
Reasons for blackouts.
Operating states of the power system.
Objectives of the control. Hierarchy of controls.
NERC guidelines for reliable operation. Indian power sector.
.Threats to security.

1.1 Introduction
residential and
is converted form of energy and is used extensively in industrial, commercial,
Electricity a
distances. This
and transmitted in bulk, economically, over long
transportation sectors. It can be generated involved in it and the
deals with the introduction of power system and control, the various issues
chapter for the
various levels of controls used. AC systems have become
the most popular system for use, over DC,
following reasons:

1. AC generators are simpler than DC generators.

different voltage levels at genera-
2. Transformation of voltage levels is simple, providing great flexibility of
tion, transmission and distribution.
widely used, are simpler and more economical than DC
motors.
3. AC motors,
and structure.

the and vary widely in siZe

interconnected
n
modern power systems, grids are
transmission. Synchronous gen
iowever, they comprise three phase systems, at both generation and
erators (alternators) are used for generation. The prime movers (primarily
turbines) act to covcuse
mechanical energy into electric energy. Thermal plants use coal as
the primary fuel and hydel plants
the turbine. The is transmitted over a wide geographical
area, at voltage
water to run generated power
end, the voltage is again stepped down ana ais
evels higher than the generated voltage. At the consumer
levels. Most of the
different
ributed to various consumers. Consumers of different rvpes need voltages of
 Thermai Hydei
Generation

IHI
Transmissicn
Transmission suDstatio

bransmission

SDStatio

Distribution EREBEBe
Distrbuto
Substadon

Load

Residential Commerciai
inausra

Figure 1.1 Basic szrucTure of a

power systern.

industrial loads are also of three phase. Single-phase commercial and residential loads are diszributei equah
berween the three phases so as to form a baanced system. The structure of the power system ra
in Fig. 1.l.
The transmission system interconnects al major generaring staions. Normaly the generatad voltage s
11 kV or 22 kV. The transmission voltages are 220 kV and above. The voltage ievei is stepped domn at the
distribution substations and transterred to the industral coasumers at voitages berween 4 and 3 5 . i b
secondary distribution feeders supply to the residential and commerial supplicrs ar 230 V. Thus the net
work is really large. consisting of a number of generating stations. several transmission intetconnecots and
the distributíon network. ObviousBy, it is not a sampie ask to run such a mzssve strucTure wthout tahures
and disruptions in service!!

2Operating States of a Power System

DyLiacco and Fink and (Carlson have classified a system operatioa into five sates as show in
Fig. 1.2
The system operation is governcd by equaity and inequaity constraines. The egualsry consrants at
nothing but the power balance berween gerneration and ioad. i he nequalsry constraints set the amits ar
 1.2 Operating States of a Power Systemn
NORMAL STATE E,I
SECURE

Economic dispatch
and load tracking

ALERT STATE E,I

RESTORATIVE STATE E, I|
INSECURE
Preventive control
Resynchronization

EMERGENCYSTATE E,T
IN EXTREMIS STATE E,T
Protect equipment, Emergency control
reduce losses
System still intact
System disintegrated
I= Inequality constraints
E =
Equality constraints
E Equality constraints = Inequality constraints
not satisfied
not satisfied

1.2 Operating states of a power system.

Figure

The system states are classi-

currents, etc.
different operating parameters, such as voltage, generation limits,
fied as follows:
constraints (I) are both
In this
state:state, the equality constraints (E) and inequality
1. Normal operating
to meet the demand,
without any equipment being overloaded.
satisfied. The generation is adequate
for normal stresses.
Further, the reserve margins are sufficient to provide security
constraints are satisfied. However, the
reserve

2. Alert state: In this state also, the equality and inequality

constraints (limits o n
are reduced. Therefore,
there is a possibility that some inequality
margins Preventive control will lead the system
trom
equipment) may be violated in the event of disturbances.
the alert state to the normal state.
state. This could be
3. Emergency state: Due disturbances, the system may enter an emergency
to severe
Ihis could
or at the local level.
at the system level
or imbalance between generation and loads, either
Decause in the system after a fault.
Some strong control meaSures,
also be because of instability due to energy built-up network
or reactor switching,
such as direct or indirect load shedding, generation shedding, shunt capacitor
taken on time, the system
measures are not
alled emergency control measures are to be taken. Ifthese
Splitting, break down and go to the
In Extremis state.
stability may be under threat and the system may eventually viola-
constraints are violated. I he
In this state, both the equality and the inequality
exTremis state: and the load demand do not match.
This

tionof the equality.constraints implies that the generation

be taken to prevent a
total
Emergency measures must
nat.some part.ot the.system load is lost.
cans
grid collapse.. eme Dy the
. Kestoratve state: a
are satisTica
transitional state., where the inequality constraints terms
This is
in of the equality con-
but the has still not come to normalcy
Bency.control actions taken, system
straints. We can have a transition cither to the alert state or to the normal state
3 Objectives of Power System Contro
I C n t a l requirenents of a power system irrespective of its size are the following:

1. The
Ihe system must be able the changing demand in active
should be sufticient
to meet
in
reactive
the
power.
demand.
and
spinning erve resen
maintained to take care of sudden variations
h e power quality (read voltage quality!) should meet certain standards with regard to frequency, amnk
tude and wave shape.
3. The energy should be
supplied
at a minimum
cost.

loachieve the above objectives, we have several levels of controls that are integrated in a complex wa
The controls shown in
are
Fig. 1.3. Some of the controls act exclusively on individual
components a
described below:
1. The
generators are
provided essentially with excitation control, to keep the voltage at the desired levets
and with prime mover control to maintain the frequency at the desired leve.
2. The
prime mover control is concerned with regulation of the speed, and the controls are tor the assoc
atedparameters such as water discharge quantity, boiler pressure, temperature, flows, etc.
3. Power sYstem_stabilizers are used to damp oscillations of the generator following a disturbance. A sta
bilizing signal
is injected into the exciter system to damp the oscillations. Some of the commonly used.
feedback signals are frequency and real power.

Tie-line
Frequency flows Po

Schedulesystem generation control

load frequency control+
economic dispatch
Supplementary
control

Generator
unit
Prime
mover
control

Excitation Generator
control

T
P
Speed

Transmissioncontrots
voltage control, VARR -
Frequency
Tie-line flows
control HVOC, FACTS
"PG
Fiqure 1.3 Various controls in a
power system
 1.4 Key Concepts for Reliable Operation
The svstem generauon
cOnitol maintains the requircd active
Control (AGC) is power balance the wstern
responsible for maintaining this in
the
he frcauency around the balance, wheh
A

nominal valuc. The also AGC, turn

maintains the
in
is
requied to hoid
which are
responsible tor power transter betwcen ditferent cheduled perwer Hoes i
contol
The transnission cOnrols incdude power and
voltage
areas
control devues,
irnes
evels within limits, maintain which help
systcm stability, Potct the systein and maintain the vrsitae
sUstem. The control devices are result n reliabie
tap changing transtormers, Flexible AC oger af 6oh oi the
lers. shunt reactors, shunt capacitorS, phasc-shitung transtormers Transnisien (FACTS) conttesi
and HVDC onttois
Distribution levcl conrols such asApacitors,
WAYC
power to the consumer. Lbese devices maintain the shapng urcuits, ct., are used to ptrmde quaity
system voltage the correct at
tude, and also help in removing harmonics frequerey and anp
injected into the load the systern
or
The control objectives need to be defined
properly. They different under normai
are
disturbances. Under normal operating conditions, we just nced to conditK and
che required generation, to maintain the keep tracking the lvad rmatch with to

frequency around the nominal value and regulate the veátag

However, under a disturbance, we need to take suitable conrol
actions to
of the system. The control action prevent catastrophsc codiapoe
depends on the state of the
system.
The operation and control of the
system should ultimately maintain the following
1. Stability: Continued intact operation of the system, following a disturbance. This depends the
ating condition and the nature of the disturbance. oper on

2. Security: It is the degree of risk in the power system's ability to survive contingencies without interrup
tion to the customer. It is related to the robustness of the system.
3. Reliability: It is the probability of satisfactory operation over a long period. It denotes the ability of the
system to supply adequate service on a nearly continuous basis, with a few intermittent interuptions
over anextended time period.

4 Key Concepts for Reliable Operation

The North American Electric Reliability Corporation (NERC)3 has proposed seven key conceprs tor rels
able operation of the power system. These are discussed in this section.
changng the
the load: The load on the power system is dynamic
and all
1. Balance the generation and
constantly.changang load The
time.
The production by the generators must be scheduled to meet this
match the generation with the demand. The demand, though
predct dynamx,
s

AGCs are used to

able and a load prediction is done, to keep the appropriate generation and reserve on hancd. faure to

the nominal valuc The tre

frequency deviation from

match the generation with the demand will
causce
demand exuerds the generation
exceeds the demand and drops if the
quency increases if the generation cquipment. Over treyuehey ard
under

detrimental the life of the

Large deviations in frequency are to

frequency the deviations cross the preset vatues

frequency relays operate when the chedued
maintan

This balance is required

to

and demand:
n c e reactive power generation capacitor banks.
lhey must
be constantty adjusted
sources are g e n e r a t o r s and
ges. Keactive power
h e gchcrator
mas
to protect
the equpment.
a i n t a i n the voltages at all levels, within
permissible range. controllers are

genetators. Today, FACTS

the voltage level of the
C voltage.regulators control be exceeded;
commonly used for reactive power control ines niust not
heating limits of the ovczhcad which have
not exceeded: The
S u r e thermal limits are
critical blackouts
There are many
below.
into the obiects eiven and ultimately grid colapse
ocrwiseC, the lines will sag relay tuipping
to short
circuits,
t e d u e to sagging oflines, leading
44. Maintain
system stability: The IEEE/CIGRE task force has
defined stability as tollows: "Powerfollows: P'ower system
Stability is the ability of an electric powet svstem, for a given initial operating O i t o n , to regain a state
operating equilibrium afier being uibjected to aphysical disturbance, with most system variables bounded.
that practically the entire system rernains intact". If a system oses stabil1ty, the gricd may face a total collape

taDility limitswill specify the maximum power that can be transterred over the iincs. Angle stability is the
aDility of the generators connected to the grid to remain in synchronism. voltage stability 1s the ability of the
y s c m to maintain steady acceptable voltages at all buses in the system under nornal operatung conditions
and also after disturbance. Both
a
are vital for the health of the system.
veer Nreliability criteria: This means that the system shouldrcmain opsratuonal and sccure cyen
1OSS of the largest gencrator in the system. (N is the number of generators, N-| indicates loss
monnaeonie
Or one
generator!). Further, after contingency, the operators must assess the
a health of the system in the
eventuality of another contingency, and take suitable control actions to maintain system security, if it were
to occur.

6. Plan, design
and maintain to operate
reliably: The planning, design and maintenance should be such
that the system should
be operated reliably and within safe limits at all umes. P'lanning involves both
short-term and long-term planning.
7. Prepare for emergencies: In spite of thorough planning and good design, emergencies such as weather
fluctuations,"operator eror, software tailure, equipment tailure, ctc., can occur. Operators must be
trained to prepare for such emergencies.
1.6 Preventive and Emergency Controls
Preventive control is meant to keep the system
n thcC state or bring it back to the normal
normal
the alert state. Automatic controls are proviaca ror state from

ures commonly used are as follows:

reduency and
voltage control. Preventive control meas-
1. Rescheduling of active power generated by vatious units, to match the changine load
2. Start-up of generation units and providing adequate spinning reserve.
3 Swirching of shunt elements for reactive power conO 1 i s wil help maintain the voltage within desired
limits.
of
Changeof reference points controllable devices such as FACTscocontrollers,
5.
phase-shifting
Change in the voltage reference pointsof generators and voltage control vices.fting transformers,
transformers, etc.
etc.

like bus-bar splitting, etc.

6. Change of substation configuration,
 Emergency control measuresare taken to stop worsening.of the situation. prcVEDL dezadatun h th
System and cascading failure effectsand to bring backthe system to the alert state. Under ireauciay. an
under-voltage load shedding schemes are used. Some of the common emergency unts mEaures ase

follows:

1. Tripping of generators.
2. Load shedding.
3. Fast valving or fast water diversion which leads to a fast reduction in generation.
4. Controlled disconnection of interconnected systems, to prevent spreading of frequency problems
5-¬ontrotectis tandingtorereatetocatgeneration-toad-baanee:
6. Blocking of tap changers of transformers.
7. Fast HVDC power transfer control.
8. Application of braking resistors.
 Energy Management Centres
I Can be seen trom the discussions in the previous sectors that the control of the modern power system Is

ExTemev compiex. Modern energ management centres have embedded a number of the above tunctions
he
eneT management centres comprise both hardware and sottware monitor and control the system
to
ot automated
n moden power SVStems, the monitoring is tullv automated. Controlling is a combination
and control
and manual operations. Sophisticated computing machines have enhanced the system operation
centre can be divided into
a u t e s . A NIETarchIcal strucrure is used tor control. The functions of the energy

free biocks
subSySTem as toiows:
ot unit commitment, economC
e dispatch subsystem: This subsystem would involve the functions
dusarch. automatic generation control and demand torecasting
and processing. I he subsy
subsystemm wil
a
subsystem: ! his subsvstem is essentially tor dati acquisition
e
units ot SADA. state estimation and all the associated alatnas and displays.ta
the poweT SYstem.
i his subsvsten is basically to oversee the
secure operatioil of
* iE: on control actions
decision
analvSIS, and
conungencv
RonitorLIng, restOrative control, etc.
control/
e control/emerge1nY
C sT212 D t e sstern. such as preventive
n the system tor the volt-
support to be provided
aiso take a decision on the VAR
d
 Cont.
ton so t'ower Systenm Operation and nroi
hapestnircd

u E O N A u m e a t , tddng (louger duratiOn).

(duratiOn around 5-10 mini
inas aNANG tdow. inteachange
evaluation

syual pNIwer in seconds).

n e n K gud. estnation (tume
onuod, state
Autewaai gouweaiK &UEud, voltage The results oflevel
mnethods are usedd.
kata and hence probabilistic
seatistal
veetas ttNAs EVNuae
nature.
determn1stic in
ain! i*ved i Funttas. These are mainby
NCN v e ?

Centres
.8.1 Major Components of Energy sometimes
are
cenires (or Systems as they
managenent
RN RAKN Nents ot the energy

control and the data acquisi-

consists of wo subsystems - the supervisory
. A A T h e sADA sYstem
central location, the status of cir-
is responsible for: (a) display at the
tn
The Supeevisorv subsytem as tap changers, capacitor
swithcing. generator voltage regulators:;
u i t dreakers annd orher devicrs such at the
breakers, changing of transtormers, etsThedispatcher
and tb) tacilitatingemte tripping
ot tap data acquisition sub-
circuit breakers, change taps, etc(The
antrol will initiatr actions to switch
centre instrumetnation
Units (RTUs) to intertace the power system
Terminal
swstem coasists ot the Remote communi-
communication channels and
control centre. Modern
ntertace
with the oontrol devurs and Communication (PLCC) was
Power Line Carrier
catiot svstems use wireless c o n u n i c a t i o n systems.

alsoused, available trom ditterent nmanufacturers. However, all

of them collect c o m m o n
A D A quipment is
reactive power tlow meaurements, voltage
system data: Circuit
breaker status, transtormer taps, real and
are processed, analysed to
assess the health and take
measurements. These data
magnitude and phase or emergency control.
control actions which could be preventive
necessary
are having immense capabilities
in ternms of memory and speed. The
2. Computers: Modern computers Since the
centres has with advent of fast computing facilities.
changed
structure of energy management for
is built in the hardware. Ditkerent schemes are available backup.
applications are cruicial, redundancy
facilities at the control centre are as follows:
The main functions of the computing
Real-time monitoring and control
User interface
Operating studies
Maintenance and testing
Simulation studies
The response time is vital in the choice of the topoligy, hardware systems, etc.
The user interface consists of consoles. data
3. User interface (with extensive GUI and display facilities):
with
units and screen projections to alert operators. Since there is extensive interacion
loggers, display
intertaces use techniques of animation and extensive graphics to make it more
human beings, modern
user friendly.
4. Applications software: This section is to implement the various functions discussed, namely, UC, cco
nomic dispatch, state estimation, optimal power tlow, contingency analysis, etc.
actions are automated and some manual. The level
As mentioned earlier, some of the control
are

driven by technology and is also dependent on the expertise available

of automation is primarily
on hand.
4 Power system SCADA and smart grids

SCADA electric utility market. As a result, RTUs used microprocessor-­

based logic to perform expanded functions. The application of micropro-
cessors increased the flexibility of supervisory systems and created new
possibilities in both operation and capabilities.

1.3 Supervisory control and data acquisition (SCADA) systems

Automation is used worldwide in a variety of applications ranging from
the gas and petroleum industry, power system automation, building
automation, to small manufacturing unit automation. The terminology
SCADA is generally used when the process to be controlled is spread over
a wide geographic area, like power systems. SCADA systems, though
used extensively by many industries, are undergoing drastic changes.
The addition of new technologies and devices poses a serious challenge
to educators, researchers, and practicing engineers to catch up with the
latest developments.
SCADA systems are defined as a collection of equipment that will
provide an operator at a remote location with sufficient information to
determine the status of particular equipment or a process and cause
actions to take place regarding that equipment or process without being
physically present.
SCADA implementation thus involves two major activities: data
acquisition (monitoring) of a process or equipment and the supervisory
control of the process, thus leading to complete automation. The complete
automation of a process can be achieved by automating the monitoring and
the control actions.
Automating the monitoring part translates into an operator in a con-
trol room, being able to “see” the remote process on the operator console,
complete with all the information required displayed and updated at the
appropriate time intervals. This will involve the following steps:

• Collect the data from the field.

• Convert the data into transmittable form.
• Bundle the data into packets.
• Transmit the packets of data over the communication media.
• Receive the data at the control center.
• Decode the data.
• Display the data at the appropriate points on the display screens of
the operator.

Automating the control process will ensure that the control command
issued by the system operator gets translated into the appropriate action
in the field and will involve the following steps:
Chapter one: Power system automation 5

Master Station Computer System

Communication Channel

Interface Devices Interface Devices

A/D Converter D/A Converter

Sensor/Transducer
Relays Controller/Actuator

Measuring Controlling
Elements Elements
Power System

Figure 1.1 The monitoring and controlling process.

• The operator initiates the control command.

• Bundle the control command as a data packet.
• Transmit the packet over the communication media.
• The field device receives and decodes the control command.
• Control action is initiated in the field using the appropriate device
actuation.

The set of equipment measuring elements helps in acquiring the data from
the field, and the set of equipment controlling elements implements the con-
trol commands in the field, as shown in Figure 1.1.

1.3.1 Components of SCADA systems

SCADA is an integrated technology composed of the following four major
components:

1. RTU: RTU serves as the eyes, ears, and hands of a SCADA system.
The RTU acquires all the field data from different field devices, as
the human eyes and ears monitor the surroundings, process the
data and transmit the relevant data to the master station. At the
same time, it distributes the control signals received from the mas-
ter station to the field devices, as the human hand executes instruc-
tions from the brain. Today Intelligent Electronic Devices (IEDs)
are replacing RTUs.
6 Power system SCADA and smart grids

Communication
System
Master Station RTU/IED

HMI

Field Equipment

Figure 1.2 Components of SCADA systems.

2. Communication System: This refers to the communication channels

employed between the field equipment and the master station. The
bandwidth of the channel limits the speed of communication.
3. Master Station: This is a collection of computers, peripherals, and
appropriate input and output (I/­O) systems that enable the operators
to monitor the state of the power system (or a process) and control it.
4. Human-­Machine Interface (HMI): HMI refers to the interface required
for the interaction between the master station and the operators or
users of the SCADA system.

Figure 1.2 illustrates the components of a SCADA system.

1.3.2 SCADA applications
SCADA systems are extensively used in a large number of industries,
for their monitoring and control. The oil and gas industry uses SCADA
extensively for the oil fields, refineries, and pumping stations. The large
oil pipelines and gas pipelines running across the oceans and continents
are also monitored by appropriate SCADA systems, where the flow, pres-
sure, temperature, leak, and other essential features are assessed and
controlled. Water treatment, water distribution, and wastewater manage-
ment systems use SCADA to monitor and control tank levels, remote and
lift station pumps, and the chemical processes involved. SCADA systems
control the heating, ventilation, and air conditioning of buildings such as
airports and large communication facilities. Steel, plastic, paper, and other
major manufacturing industries utilize the potential of SCADA systems
to achieve more standardized and quality products. The mining industry
with integrated SCADA for the mining processes, like tunneling, product
flow optimization, material logistics, worker tracking, and security fea-
tures, is the latest addition to the list, making digital mines.
The use of SCADA systems in the power industry is widespread, and
the rest of the discussion in this chapter will focus specifically on the power
sector, including generation, transmission, and distribution of power.
Chapter one: Power system automation 7

1.4 SCADA in power systems

SCADA systems are in use in all spheres of power system operations
starting from generation, to transmission, to distribution, and to utiliza-
tion of electrical energy. The SCADA functions can be classified as basic
and advanced application functions.

1.4.1 SCADA basic functions

The basic SCADA functions include data acquisition, remote control, human-­
machine interface, historical data analysis, and report writing, which are
common to generation, transmission, and distribution systems.
Data acquisition is the function by which all kinds of data—analog,
digital, and pulse—are acquired from the power system. This is accom-
plished by the use of sensors, transducers, and status point information
acquired from the field.
Remote control involves the control of all the required variables by the
operator from the control room. In power systems, the control is mostly of
switch positions; hence, digital control output points are abundant, such as
circuit breaker and isolator positions and equipment on and off positions.
Historical data analysis is an important function performed by the
power system SCADA, where the post-­ event analysis is done using
the data available after the event has happened. An example is the post-­
outage analysis where the data acquired by the SCADA system can pro-
vide insights into such information as the sequence of events during the
outage, malfunctioning of any device in the system, and the action taken
by the operator. This could be a powerful tool for future planning and is
extensively used by power engineering personnel.
Power system SCADA requires a number of reports to be generated
for consumption at different levels of the management and from different
departments of the utility. Hence, report generation is essential as per the
requirements of the parties and departments involved.

1.4.2 SCADA application functions

Figure 1.3 illustrates the use of SCADA in power systems, with the initial
SCADA block depicting the basic functions, as discussed in Section 1.4.1.
The right section of the figure illustrates the generation SCADA, repre-
sented by SCADA/­AGC (Automatic Generation Control), implemented in
the generation control centers across the world. Further, the transmission
SCADA is shown as SCADA/­EMS (Energy Management Systems) where
the basic functions are supplemented by the energy management sys-
tem functions. This is implemented in the transmission control centers.
8 Power system SCADA and smart grids

Supervisory Control
and Data Acquisition
(SCADA)

Substation
Automation
(SA)

Distribution Automation SCADA/Automation Generation

(DA) Control (SCADA/AGC)

Distribution Management SCADA/Energy Management

System (DMS) System (SCADA/EMS)

Figure 1.3 Use of SCADA in power systems.

The EMS software applications are the most expensive component of the
SCADA/­EMS, mainly due to the complexity of each application. The left
part of the figure shows the distribution functions superimposed on the
basic SCADA functions, beginning at the SCADA/­distribution automa-
tion system and further expanding to the distribution management sys-
tem functions. As one scans the figure from top to bottom, the systems
become more complex and more expensive (i.e., the basic SCADA system
is the simplest and least expensive, the SCADA/­AGC is more involved
and a little more expensive, and the SCADA/­EMS is much more complex
and expensive). The same is true for distribution. The SCADA/­DA is more
involved and more expensive than the basic SCADA system. The SCADA/­
DMS is much more complex and expensive.

1.4.2.1 Generation SCADA application functions

As discussed earlier, generation SCADA, in addition to the basic functions
discussed earlier, will include the following application functions.

• Automatic Generation Control (AGC): a compendium of equipment

and computer programs implementing closed-­loop feedback control
of frequency and net interchange
• Economic Dispatch Calculation (EDC): the scheduling of power from
all available sources in such a way to minimize cost within some
security limit
Chapter one: Power system automation 9

• Interchange Transaction Scheduling (ITS): ensures that sufficient energy and

capacity are available to satisfy load energy and capacity requirements
• Transaction Evaluation (TE): evaluates economy of transactions using
the unit commitment results as the base condition
• Unit Commitment (UC): produces the hourly start-­up and loading
schedule which minimizes the production cost for up to one week
in the future
• Short-­Term Load Forecasting (STLF): produces the hourly system load
for up to one week into the future and is used as input to the unit
commitment program
• Hydrothermal coordination: the scheduling of power from all available
hydro generation in such a way to minimize cost within constraints
(e.g., reservoir levels)

1.4.2.2 Transmission SCADA application functions

The transmission SCADA will include energy management system (EMS)
functions such as

• Network Configuration/­Topology Processor: analyzes the status of cir-

cuit breakers as well as measurements to automatically determine
the current model of the power system
• State Estimation: provides a means of processing a set of redundant
information to obtain an estimate of the state variables of the system
• Contingency Analysis: simulates outages of generating units and
transmission facilities to study their effect on bus voltages, power
flows, and the transient stability of the power system as a whole
• Three-­Phase Balanced Power Flow: obtains complete voltage angle and
magnitude information for each bus in a power system for specified
load and generator real power and voltage conditions
• Optimal Power Flow: optimize some system objective function, such
as production cost, losses, and so on, subject to physical constraints
on facilities and the observation of the network laws

Details of the above functions and additional functions are explained in

Chapter 5.

1.4.2.3 Distribution automation application functions

Distribution automation/­distribution management systems (DA/­ DMS)
include substation automation, feeder automation, and customer auto-
mation. The additional features incorporated in distribution automation
will be

• Fault identification, isolation, and service restoration

• Network reconfiguration
10 Power system SCADA and smart grids

• Load management/­demand response

• Active and reactive power control
• Power factor control
• Short-­term load forecasting
• Three-­phase unbalanced power flow
• Interface to customer information systems (CISs)
• Interface to geographical information systems (GISs)
• Trouble call management and interface to outage management sys-
tems (OMSs)

Details of distribution automation functions are given in Chapter 6.

1.5 Advantages of SCADA in power systems

Automating a system brings many advantages, and the case of power sys-
tems is no different. Some of the advantages are as follows:

• Increased reliability, as the system can be operated with less severe

contingencies and the outages are addressed quickly
• Lower operating costs, as there is less personnel involvement due
to automation
• Faster restoration of power in case of a breakdown, as the faults can
be detected faster and action taken
• Better active and reactive power management, as the values are accu-
rately captured in the automation system and appropriate action can
be taken
• Reduced maintenance cost, as the maintenance can be more effec-
tively done (transition from time-­based to condition-­based mainte-
nance) with continuous monitoring of the equipment
• Reduced human influence and errors, as the values are accessed
automatically, and the meter reading and related errors are avoided
• Faster decision making, as a wealth of information is made available
to the operator about the system conditions to assist the operator in
making accurate and appropriate decisions
• Optimized system operation, as optimization algorithms can be run
and appropriate performance parameters chosen

Some of the additional benefits by SCADA system implementation are as

discussed below.

1.5.1 Deferred capital expenditure

With a real-­time view of loading on various transmission lines, feeders,
transformers, circuit breakers, and other equipment, and the ability to
chapter two

SCADA fundamentals

2.1 Introduction
Supervisory control and data acquisition (SCADA) systems are extensively
used for monitoring and controlling geographically distributed processes
in a variety of industries. However, many of the SCADA-­related prod-
ucts are proprietary, and the knowledge of the components is acquired
by the personnel on the job. Hence, students and new graduates find it
difficult to understand the fundamentals of SCADA systems. An attempt
has been made in this chapter to elaborate on the essential components
of the SCADA systems which will help explain the functioning and the
hierarchy, especially for power systems.

2.2 Open system: Need and advantages

SCADA systems are complex and require a variety of hardware and soft-
ware seamlessly integrated into a system that can perform the monitor-
ing and control operation of the large process involved. Communication
among devices is key to successful SCADA implementation in modern
power systems. Traditionally most vendors in the automation scenario
established their own unique (“proprietary”) way to communicate between
devices. Getting two vendors’ proprietary devices to communicate prop-
erly is a complex and expensive task. The possible solution to the problem
is through two basic approaches:

1. Buy everything from one vendor.

2. Get vendors to agree on a standard communication interface.

The first proposition was widely used as earlier proprietary products were
utilized for SCADA implementations and large turnkey projects were com-
missioned by a single vendor. This created a monopoly of products and
processes, and it became increasingly difficult to maintain or expand the
established SCADA systems.
The latter approach, to get all the vendors to agree on a standard com-
munication interface, is the fundamental objective of the “open systems”
movement. This led to the concept of nonproprietary, open systems, which

21
22 Power system SCADA and smart grids

created a level playing field for all the players in the automation industry.
Interoperable systems are becoming popular due to the huge advantages
they provide for manufacturers, vendors, and end users.
An open system is a computer system that embodies vendor-­
independent standards so that software may be applied on many differ-
ent platforms and can interoperate with other applications on local and
remote systems.
Open systems are thus an evolutionary means for a control system,
based on the use of nonproprietary and standard software and hardware
interfaces, that enables future upgrades to be available from multiple ven-
dors at lowered cost and integrated with relative ease and low risk.
The advantages of open systems are manifold, evolving from the
definition:

• Vendor-­independent platforms for project implementation can be

used, avoiding reliance on a single vendor.
• Interoperable products are used. Turnkey projects where one ven-
dor supplies and implements the complete project are no longer
required, as use of hardware and software from different vendors
is possible.
• Standard software that could be used to program different hard-
ware can be used.
• The de jure (by law) and de facto (in fact or actually) standards can be
used.
• System and intelligent electronic devices (IEDs) from competing
suppliers will have common elements that allow for interchange and
the sharing of information.
• Open systems are upgradable and expandable.
• They have a longer expected system life.
• There are readily available third-­party components.

As the following sections discuss the building blocks of SCADA systems,

it is apparent that all the components discussed use open systems now
and the SCADA implementations are exciting propositions with hard-
ware and software acquired from multiple vendors as per the functional
requirements of each system.

2.3 Building blocks of SCADA systems

The SCADA system has four components, the first being the remote ter-
minal unit (RTU) or data concentrator, which is the link of the control
system to the field, for acquiring the data from the field devices and
Chapter two: SCADA fundamentals 23

passing on the control commands from the control station to the field
devices. Modern-­day SCADA systems are incomplete without the data
concentrators and intelligent electronic devices (IEDs) which are replac-
ing the conventional RTUs with their hardwired input and output (I/­O)
points. In this book, both RTUs and IEDs have been discussed in detail.
Legacy systems with only RTUs, hybrid systems with RTUs and IEDs,
and new systems with only IEDs have to be handled with ease by the
SCADA system designer today. The second component is the communica-
tion system that carries the monitored data from the RTU to the control
center and the control commands from the master station to the RTU or
data concentrator to be conveyed to the field. The communication sys-
tem is of great significance in SCADA generally and in power automa-
tion specifically, as the power system field is widely distributed over the
landscape, and critical information that is time bound is to be commu-
nicated to the master station and control decisions to the field. The third
component of the SCADA system is the master station where the operator
monitors the system and makes control decisions to be conveyed to the
field. The fourth component is the user interface (UI) also referred to as
the human-­machine interface (HMI) which is the interaction between the
operator and the machine. Figure 2.1 gives a pictorial representation of the
components of a SCADA system. All automation systems essentially have
these four components, in varied proportions depending on the process
requirements. Power system SCADA systems are focused on the master
stations and HMI is of great significance, whereas process automation is
focused on controllers, and master station and the HMI has less signifi-
cance. The following sections will elaborate how the components of the
SCADA system work cohesively to accomplish monitoring and control of
the process to achieve optimum performance of the system.

RTU Field
Communication Equipment
Channel
Master
station
IED
Field
IED Equipment
CFE/FEP Data
Concentrator
IED

Figure 2.1 Components of SCADA systems.

24 Power system SCADA and smart grids

2.4 Remote terminal unit (RTU) [1–7,18–19,24]

The RTU is the eyes, ears, and hands of the SCADA system. In older days,
RTU was a slave of the master station, but now RTUs are equipped with
internal computational and optimization facilities. RTU collects data from
the field devices, processes the data, and sends the data to the master sta-
tion through the communication system to assist the monitoring of the
power system as “eyes” and “ears” of the master station. At the same time,
the RTU receives control commands from the master station and trans-
mits these commands to the field devices, thus justifying the comparison
to the “hands” of the master station. Figure 2.1 shows the location of the
RTU and the communication front end/­front-­end processor (CFE/FEP) of
the master station.

2.4.1 Evolution of RTUs
From 1900 to the early 1920s, varieties of remote control systems were
developed by engineers for remotely supervising processes. The systems
could only monitor the process and no control was possible. In 1921 a sys-
tem designed by John B. Harlow could automatically detect a change of
status at a remote station and could report the change to the control center.
In 1923 the remote control system developed by John J. Bellamy and
Rodney G. Richardson employed an equivalent of our modern “check
before-­operate” technique. It ensured the validity of a selected control
point before the actual control was initiated. In 1927 the first logging sys-
tem, designed by Harry E. Hersey, monitored information from a remote
location and printed status change with reported time and date.
Supervisory systems evolved from electromechanical to using solid-­
state components, electronic sensors, and analog-­to-­digital converters.
With the advent of microprocessors, RTU manufacturers merely upgraded
their technology and did not look at alternate ways of performing the
RTU function.
In 1980s, microprocessor-­based logic was incorporated into the RTUs.
This increased the flexibility of supervisory systems and brought in new
capabilities in operation and performance. The development in commu-
nications and faster microprocessor chips brought down the costs and
improved performance.
The new systems had the following advantages:

1. Modular system development capability

2. Largely preprogrammed user interface system that is easy to adapt
to the individual process
3. Preprogrammed menu-­driven software (final programming using a
few buttons on keyboard)
Chapter two: SCADA fundamentals 25

4. Wide selection of control algorithms with preprogrammed menu

5. Data highway with transmission and communication capabilities
between separate units—wideband, redundancy
6. Relatively easy communication with the control room for supervi-
sory control
7. Extensive diagnostic scheme and devices for easy maintenance and
replacement of circuit board (card level)
8. Redundancy at any level to improve the reliability
9. Industry standard communication protocols (IEEE 1815 or DNP3,
IEC 60870-5-101 and 103)

2.4.2 Components of RTU
RTU has the following major components to accomplish the tasks of moni-
toring and controlling the field devices:

1. Communication Subsystem: Communication subsystem is the interface

between the SCADA communication network and the RTU internal
logic. This subsystem receives messages from the master, interprets
the messages, initiates actions within the RTU which in turn initiates
some action in the field. RTU also sends an appropriate message to
the master station on the completion of the task. It also collects data
from the field, and processes and conveys relevant data to the master
station. RTU may report to a single master or multiple masters.
2. Logic Subsystem: The logic subsystem consists of the main processor
and database and handles all major processing—time keeping, and
control sensing. The logic subsystem also handles the analog-­to-­digital
conversions and computational optimization, in most of the cases.
3. Termination Subsystem: The termination subsystem provides the
interface between RTU and external equipment such as the com-
munication lines, primary source, and substation devices. RTU logic
needs to be protected from the harsh environment of the substation.
4. Power Supply Subsystem: The power supply converts primary power,
usually from the substation battery, to the supply requirements of
the other RTU subsystems.
5. Test/­HMI Subsystem: This subsystem covers a variety of components,
built-­in hardware/­firmware tests, and visual indicators, within the
RTU, and built-­in or portable test/­maintenance panels or displays.

Figure 2.2 shows the components of the RTU, and the following sections
will provide details of each of the RTU components. Figure 2.3 presents a
typical RTU in a substation.
26 Power system SCADA and smart grids

Master Station

SCADA
Communication
RTU Network

Communication SS

Test/
PS Logic SS HMI

Termination SS

Power System

Figure 2.2 Components of RTU.

Figure 2.3 Typical RTU in a substation. (Courtesy of GE.)

2.4.3 Communication subsystem
The communication subsystem is the interface between the SCADA com-
munication network and the RTU internal logic. Messages from the master
station are received and interpreted by the communication subsystem, and
the required action is initiated within the RTU. The RTU then initiates the
requisite control action in the field, on the completion of which an appro-
priate message is transmitted to the master station. The communication
Chapter two: SCADA fundamentals 27

subsystem receives data from the field, processes the data, bundles the rel-
evant data in the appropriate protocol, and conveys the data to the master
station, via the SCADA communication network. Hence, it is evident that
the communication subsystem of the RTU is responsible for interpreting
the messages from the master station, as well as formatting the messages
to be transmitted to the master, including the message security. The RTU
communication subsystem handles the following functions

2.4.3.1 Communication protocols
A large variety of communication protocols exist in the power system,
and the RTU communication system is designed to format and interpret
the data in the required protocol. Details of the communication protocol
structure and the protocols used in the power system are discussed in
Chapter 3. SCADA communication protocols generally “report by excep-
tion” or give information on the points that have changed since the last
scan, to reduce the communication system load. For analog points, this
means changing beyond their deadband between scans.

2.4.3.2 Message security
The data handled by the SCADA system are critical, and any corruption
in the data can lead to serious consequences. Parity check is the simplest
method, where a single bit is added to the message so that the sum is
always odd. Cyclic redundancy check (CRC) is another error-­checking
mechanism used, which is more reliable. Here, each block of data is
divided by a 16-degree polynomial; the remainder of the division is added
to the end of the message block. The message will have a fixed length pre-
amble of overhead characters, depending on the protocol used, the station
address, the function code, and other details. CRC code is calculated sepa-
rately for the preamble and the data block.

2.4.3.3 Multi-­port communication
Modern RTUs have to communicate to the higher SCADA hierarchy to
more than one master station, and at the same time, communicate with
peer RTUs and IEDs in a variety of protocols. The communication sub­
system should be designed to handle this capability.

2.4.4 Logic subsystem
The logic subsystem is the central processing and control unit of the RTU.
Modern-­day RTUs perform a number of advanced functions to off-­load
the master station in addition to the two primary functions: data collec-
tion and processing and control point selection and execution.
The primary functions of the RTU are time keeping and data acquisi-
tion and processing, as shown in Figure 2.4.
Chapter two: SCADA fundamentals 35

data which leads to requests to implement advanced processing features

into the RTU. One of these requests is to consider analysis of status point
changes to report only the high-­level action. For example, when a circuit
breaker operates, numerous analog point alarms can be generated (low
voltage and low current for all three phases). The important message to
give the system operator is the fact the circuit breaker operated (primary
alarm), but present systems also provide the analog point alarms (second-
ary alarms), too. The desire is to log all the analog point changes in the
master station for later engineering analysis but report only the breaker
operation to the system operator. Intelligent alarm processing is advanced
data processing that can be installed in the RTU (see Section 4.9.3.2).

2.4.8.6 Other functions
Time tagging of analog and digital values for sequence of events record-
ing is implemented in RTUs. Other functions that can be implemented
in the RTU include distribution automation, volt-­ampere reactive (VAR)
control and fault detection, isolation, and service restoration, as explained
in later chapters.
Thus, it is evident that modern-­day RTUs are power houses that, in
addition to acquiring the data from the field and executing the control
actions, are capable of performing a variety of other functions and are
integral components of the SCADA system.

2.5 Intelligent electronic devices (IEDs)

The industry standard definition of an IED is “Any device incorporating
one or more processors with the capability to receive or send data/­control
from or to an external source (e.g., electronic multifunction meters, digital
relays, and controllers).”
IEDs have been deployed extensively in power automation systems
recently, and the shift from RTUs to IEDs is evident due to the integration
and interoperability features of the IEDs. It is necessary at this point to
discuss the IED functionality in detail to present a holistic view of auto-
mation in power systems.

2.5.1 Evolution of IEDs
IEDs were introduced in the early 1980s with microprocessor-­based con-
trol features. The deployment of IEDs is revolutionizing the protection,
substation and distribution automation, and data capture and analy-
sis functions of an electric utility. The protection relay migrated from
single-­function conventional electromechanical types to multi-­function
36 Power system SCADA and smart grids

microprocessor-­based relays and started incorporating different protec-

tion functions into the same relay, rather than using individual relays for
each application. Considerable savings were achieved in relay panel and
switchgear costs by the adoption of the multi-­function microprocessor-­
based relays.
However, IED revolution started when other functionalities like accu-
rate voltage and current phasor measurement, waveform capture, and
metering were being incorporated into the relays.
The growth in communication infrastructure, standardization of pro-
tocols, and interoperability were major factors that led to the IED explo-
sion. IEDs are now the eyes, ears, and hands of the automation systems in
a power utility. IED packed with full control and monitoring capabilities
and with analysis of fault report data can manage substations without
human intervention. The wrong tripping of circuits can be avoided by
utilizing the IEDs’ capabilities to the full extent.
With highly integrated IEDs, utilities and industrial plants have a
huge potential for cost savings. These savings can be summarized in the
following categories:

1. Lower installation and panel assembly cost

2. Shorter commissioning and maintenance times
3. Shorter system recovery time after a disturbance
4. Less revenue loss due to wrong settings and IED malfunction
5. Higher system reliability due to automation, integration, and adap-
tive settings
6. Better utilization of installed capacity
7. Better justification of new investments
8. Smaller control houses

Integration of IEDs and proper analysis of fault data will lead to very short
system restoration times after a blackout, and revenue losses of utilities
will be minimized.

2.5.2 IED functional block diagram

Figure 2.7 depicts the structural block diagram of a typical intelligent
electronic device. The modern IED architecture ensures that the device is
multipurpose, modular in nature, flexible and adaptable, and has robust
communication capabilities. Communication capabilities include multiple
selectable protocols, multi-­drop facilities with multiple ports, and rapid
response for real-­time data. IEDs also have tremendous data-­processing
capability for a variety of functions, for various applications like protec-
tion and metering. IEDs have event recording capability that can be very
Chapter two: SCADA fundamentals 37

External Data Input/Output

Communication Processing Measurement

Selectable Protocol Protection* Discrete Inputs

Selectable Protocol Metering Analog Inputs*

Rapid Response Event Recording Discrete Outputs*

Real-Time Data* Fault Recording Analog Outputs*

Multiple Ports Application Logic Selectable Ratings

*Old Relay

Figure 2.7 Modern IED with the functional blocks.

Settings
Intelligent Configuration
GPS Electronic Device I/O Programming
Disturbance
Analysis
Event record data
Commissioning
test
Metering Protection
Monitoring

Control

Substation
Server Office/
Home

Figure 2.8 (See color insert.) Functional view of modern IED.

useful for post-­event analysis, for fault waveform recording, and for power
quality measurements. This eliminates additional digital fault recorders
and power quality monitors. IEDs can also accept and send out analog
and digital signals with selectable ratings, thus making the IEDs versatile.
As far as IED deployment in the field is concerned, Figure 2.8 gives
a complete picture of the IED integration with the devices connected and
the functionalities handled by the IEDs.
The IED brings a relay panel with many single-­function electrome-
chanical relays, control switches, extensive wiring, and much more into
38 Power system SCADA and smart grids

a single box. In addition the IED handles additional features like self and
external circuit monitoring, real-­time synchronization of the event moni-
toring, local and substation data access, programmable logic controller
functionality, and an entire range of software tools for commissioning,
testing, event reporting, and fault analysis. Typical relay IEDs are shown
in Figure 2.9.
The following sections will elaborate the IED building blocks in detail.

2.5.3 Hardware and software architecture of the IED

The architecture of an IED should ensure the ease of use of the device in
regard to programming, commissioning, and maintenance. The hardware
should be designed with the future adaptability requirement in mind,
whereas the software structure should ensure the independent protec-
tion, control, metering, and communication functions.
IED hardware design utilizes draw out–­type cards which is a great
advantage, as the replacement can be done easily without disconnecting
the terminal wires and removing the IED from the panel.
The IED software architecture is designed in such a way that the
commissioning engineer can easily evaluate and program the available
functions independently. The required function can be selected, while
the other functions are deactivated and will not be visible to the person-
nel, which helps to save time while commissioning. Each selected function
is an independent embedded unit generally with the IED with dedicated
logical inputs and outputs, setting, and event reporting features
Figure 2.10 illustrates the functional blocks in an IED which demon-
strate the versatility of the device. In addition to the analog, digital inputs
and outputs, the IED has the capability of waveform capture and distur-
bance analysis capability. Metering and demand values recording are
other features, in addition to programmable logic capability of the IEDs
that eliminates an additional PLC usage. Self and external circuit moni-
toring make the device extremely reliable and reduce downtime.

2.5.4 IED communication subsystem

IED communication is of utmost importance; hence, the device provides
flexibility and at the same time major benefits to the utility. The IED
should support different protocols for multi-­port communication and dif-
ferent media and should have flexible and open communication archi-
tecture. HMI interface, remote access port and direct communication to
other IEDs for protection purposes are musts for modern IEDs.
As discussed earlier, open protocols are the norm today, and the IEDs
have plug-­and-­play communication modules that can support a variety of
protocols. The advantage of these modules is that they can be replaced in
Chapter two: SCADA fundamentals 39

Figure 2.9 (See color insert.) Relay IEDs. (Courtesy of GE.)

40 Power system SCADA and smart grids

Protection
Functions RMS Metering/ General Event
ON/OFF Demand values Buffer

Control
Functions
BLOCK Real time Clock Disturbance Event
Programmable Buffer
Logic
CONTROL

Communication Self & External Waveform Capture

Functions Circuit Monitoring Buffer

LCD
Display

Trip and Digital LED Serial Analog

Alarm Inputs Outputs Port Inputs
Contacts

Figure 2.10 Structure of an IED [12].

the field in case of a change in communication requirement, and the IEDs

can be integrated to a separate system at the higher hierarchy with ease.
IEDs are capable of multi-­port communication and can communicate with
substations and other IEDs at the same time through a modem to office/­
home/­service station.
The IEDs use the communication port and optical port for fiber optic
communication or electrical port (RS-232 or RS-485) and will also have a
service port for remote access via a modem.
Figure 2.11 shows the relay IEDs communicating to a computer/­server
at the substation.

2.5.5 IED advanced functionalities [11–15,17]

2.5.5.1 Protection function including phasor estimation
The protection function is the primary function of a relay IED, as IEDs are
primarily the improvement on the microprocessor-­based relays. There are
tremendous improvements in the new IEDs as more accurate measure-
ment principles and less auxiliary equipment are required. Auxiliary CTs
can be eliminated in a transformer differential relay, as the new relay has
Chapter two: SCADA fundamentals 41

Substation Computer

MODBUS DNP3 over MODBUS Proprietary Operational

IEC 61850
TCP/IP ASCII for data
NOD DNP3

IED IED
IED IED IED

Figure 2.11 IED communication options (NOD, nonoperational data).

a CT mismatch correction function, as the original primary transformer

currents are available for further analysis in a modern relay IED. Similarly,
with appropriate techniques, and numerical comparison algorithms, the
CT mismatch, inrush, and CT saturation problems can be solved without
external devices.
McLaren was the first to propose the concept of open system relaying,
where different relay functions can be obtained from the same hardware
just by modifying microprocessor programming. Modern relay technol-
ogy has shown recently a tendency toward this direction. The generalized
numerical relay concept, which is directly derived from open system relay-
ing, consists of a minimum set of hardware modules and functions of mod-
ern digital and numerical relays. With the generalized numerical relay and
with the amount of information commonly available, it is possible to recreate
the majority of modern digital and numerical relay equipment. The follow-
ing data-­processing modules constitute the generalized numerical relay:

1. Isolation and analog signal scaling: Current and voltage waveforms

from instrument transformers are acquired and scaled down to con-
venient voltage levels for use in the digital and numerical relays.
2. Analog anti-­aliasing filtering: Low-­pass filters are used to avoid the phe-
nomenon of aliasing in which the high-­frequency components of the
inputs appear to be parts of the fundamental frequency components.
42 Power system SCADA and smart grids

3. Analog-­to-­digital conversion: Because digital processors can pro-

cess numerical or logical data only, the waveforms of inputs must
be sampled at discrete times. To achieve this, each analog signal is
passed through a sample- and-­hold module, and conveyed, one at a
time, to an analog-­to-­digital converter (ADC) by a multiplexer.
4. Phasor estimation algorithm: A software algorithm implemented in
a microprocessor estimates the amplitude and phase of the wave-
forms provided to the relay. This is of great importance in modern
monitoring systems, as phasor measurements are becoming an inte-
gral part of the system monitoring. This feature of the IED is used
to compute the phasor of the voltage or current with respect to a
reference phasor. (The IEDs are time synchronized from a common
GPS source.) This is termed as a phasor measurement unit (PMU),
and the phasor data are accumulated by a phasor data concentrator
(PDC) at an appropriate location, mostly in the control center. The
phasor concept is explained in detail in Chapter 5.
5. Relay algorithm and trip logic: The equations and parameters specific
to the protection algorithm and the associated trip logic are imple-
mented in the software of the microprocessor used in the relay.
The microprocessor calculates the phasors representing the inputs,
acquires the status of the switches, performs protective relay calcula-
tions, and finally provides outputs for controlling the circuit break-
ers. The processor may also support communications, self-­testing,
target display, time clocks, and other tasks.

2.5.5.2 Programmable logic and breaker control

A modern relay IED eliminates the use of external programmable logic
controllers (PLCs) as the IED can handle logical inputs and outputs of the
protection functions, which can be connected to flip flops and/­or gates of
the IED directly.
Figure 2.12 presents an example of programmable logic tools. In
Figure 2.13, an example of a PC-­configuration tool for user programmable
logic is shown.

2.5.5.3 Metering and power quality analysis

Metering capabilities of the IEDs became acceptable to the power utili-
ties quickly, and major cost saving was achieved by combining the non-­
revenue metering function into the IEDs. It may be noted that the primary
CTs and PTs for protection purposes may not be accurate enough for nor-
mal current measurement for revenue metering. The normal metering
functions include measuring the voltage and current root mean square
(RMS) values and the real and reactive power.
In addition to these basic functions, metering also includes the values
for commissioning and testing, and this feature reduces the commission
Chapter two: SCADA fundamentals 43

Figure 2.12 Example of programmable-­logic tools.

Figure 2.13 Programming of the graphic LCD display for breaker operation logic.
44 Power system SCADA and smart grids

and testing times on the site. The metered values are the positive, negative,
and zero sequence components of voltage and current phase shifts and
the normal RMS values. The phase mismatch, differential, and restraint
values can be computed easily to hasten the commissioning process.
Load profiling is another metering activity that can be achieved using
IEDs where the power factor profile, ampere demand, long-­term RMS
voltage value, and so on, can be monitored and can be used for load pro-
filing for long-­term expansion planning.
Using the PLC and metering functions, many system requirements
can be met without additional effort, like capacitor bank control by the
reactive power data monitoring and control algorithm implemented using
a PLC and programmable output contacts.

2.5.5.4 Self-­monitoring and external circuit monitoring

IEDs have card-­level diagnostic capabilities for internal problems, with
a self-­monitoring software, which can detect up to 98% of the problems,
such as hardware failure, memory failure, and power supply problems.
Modern-­day IEDs, in addition to the internal monitoring, have capabili-
ties for interface monitoring and external circuit monitoring. Interface
monitoring includes checking the inputs to the IEDs and can be verified
by simple methods. For example, the input currents to the relay from the
three phases should add up to three times the neutral current if any. If
there is any deviation, the analog channel of any of the currents could be
faulty. The relay can block the false tripping. External circuit monitoring
will include monitoring of the circuit breaker coil for any interruption in
the trip-­close path and can also indicate an instrument transformer failure.

2.5.5.5 Event reporting and fault diagnosis

Relay IEDs eliminate the digital fault recorders because waveform record-
ing during a fault can be performed by the IEDs, whereas the electrome-
chanical relays did not have such capability. Event reporting can be easily
done by relay IEDs eliminating sequence of events (SOE) recorders. The
relay IEDs save the captured data in nonvolatile memory and disturbance
event reports (pick up, trip, and auto-­reclose), and general event reports like
changes of settings have to be saved and managed separately. Time stamp-
ing of all events is done by the IEDs, and GPS synchronization for this pur-
pose and a battery backup for the real-­time clock are essential. The events,
once time tagged correctly, can be reported in the correct sequence in which
they occurred, eliminating further sequencing at the control room. It is
hence easy to perform fault diagnosis after a fault, as the values will be
saved in the IED and can be retrieved later, even in case of a blackout.
Chapter two: SCADA fundamentals 45

2.5.6 Tools for settings, commissioning, and testing

User-­friendly software tools are key to better planning, programming,
commissioning, and testing of an IED. The elaborate avenues and applica-
tion functionalities of an IED can be handled only by an intuitive, easy-­to-­
use PC program.
IEDs come with menu-­driven intuitive, easy-­to-­use, flexible software
programming tools for settings and configurations during commission-
ing. These user-­friendly programs come with factory presettings that sim-
plify the job of the commissioning personnel.

2.5.7 Programmable LCD display

The programmable LCD display is a great tool in the new generation
of IEDs. This is used for graphical information as well displaying text
and can be switched between graphic and text modes. Figure 2.9 shows
the display of relay IEDs. The topology of the bus and breaker including
isolators, disconnecting switches, and many more configurations can be
programmed using the software tools in the graphic mode. Text mode of
the LCD display is used for settings and detailed display of the metering
values in primary or secondary units. The LCD display can be switched
between text and graphics modes.

2.5.8 Typical IEDs
IEDs, as discussed, are devices that can be connected to a LAN and com-
municate with other devices over the LAN and have processing capa-
bilities. A large number of IEDs are available currently, relay IEDs being
the most commonly used for automation purposes. However, the smart
meter used for home automation is an IED and so is a programmable logic
controller (PLC) used for automation. Digital fault recorders (DFRs) and
remote terminal units (RTUs) are IEDs with digital data transmission and
reception capability with computational facility built in.
Thus IEDs have become the basic building blocks for automation of
power systems. However, it is not easy to replace all the existing RTU
and related equipment with IEDs, and different approaches are used to
integrate the legacy systems with the new systems. Before discussing the
building of different kinds of SCADA systems, it is imperative to touch
upon the data concentrators and merging units which are used along with
RTUs and IEDs for data communication in the SCADA systems.
62 Power system SCADA and smart grids

2.10 Building the SCADA systems, legacy,

hybrid, and new systems
The above sections elaborated the building blocks of SCADA systems,
starting from the RTU, IEDs, communication systems, master stations and
the HMI. Utilities have a variety of options available to mix and match
the elements to build a cost-­ effective, efficient, and operator-­
friendly
SCADA system.
However, automation of the power systems started as early as the
beginning of the twentieth century, and substations and control centers
operate at various stages of automation all over the world. There are leg-
acy systems with RTUs, hardwired communication from the field to the
RTU, and traditional software functionalities in the control room, and it is
not often financially viable to dismantle everything and purchase a com-
pletely new automation system.
Hybrid systems are a viable option, where any automation expan-
sion project can be implemented with new devices, like IEDs, data con-
centrators, and merging units, as shown in Figure 2.14. The new system
will coexist with the legacy RTU-­based systems, and the data integration
and if necessary protocol conversion issues will have to be handled while
commissioning the project.
If a utility decides to purchase a completely modern system, the lat-
est building blocks of the SCADA system, viz., IEDs, merging units, and
fiber optic communication facility with brand new HMI with situational
awareness and analysis tools, can be implemented.
The legacy, hybrid, and new systems for a typical substation automa-
tion implementation are discussed in detail in Section 4.7.

2.11 Classification of SCADA systems

SCADA systems can be classified into four categories depending on the
complexity and the number of RTUs and master stations present in the
system. The classification will also depend on the number of points at
each RTU and the required update rates, location of the RTUs, communi-
cation facilities, and equipment available.

2.11.1 Single master–­single remote

The simplest configuration, the single master–­single remote supervisory
(Figure 2.18), is utilized for simple systems where small numbers of points
are involved, since it requires one master station and one communica-
tion channel per RTU. This one-­on-­one configuration generally has one
indicator or display at the master station for each remote data point. An
Chapter two: SCADA fundamentals 63

Master
Station

RTU

Figure 2.18 Single master–­single remote configuration.

example could be the control center of a generating station with one RTU
to collect data.

2.11.2 Single master–­multiple RTU

In the one master–­ multiple RTU configuration, one master station is
shared by several RTUs. Generally, the master station communicates
in turn to each RTU using serial digital data messages. This configura-
tion has the advantage over the one-­on-­one of sharing the master station
communications logic among a number of RTUs. An example could be a
power distribution system with one master station controlling a number
of substations with RTUs. These are generally “off-­the-­shelf” systems that
can be procured easily, and the number of RTUs is generally restricted
to 25. The communication configuration could be radial or shared line
(party line) as shown in Figures 2.19a and 2.19b.

2.11.3 Multiple master–­multiple RTUs

In multiple master–­multiple RTU configuration, there will be submasters
available with multiple RTUs reporting to each master. These systems
will have a large number of RTUs connected to it, and extensive engi-
neering and customization are required for commissioning of the system.
The multiple master–­multiple remote is also characterized by a sizable
number of application programs. These systems will take a longer time to
execute and implement. Figure 2.20 gives the multiple master–­multiple
RTU concepts. An example is a generation and transmission (G&T) util-
ity with multiple distribution members, where each member has its own
SCADA system. The member’s SCADA masters report upstream to the
G&T master.
64 Power system SCADA and smart grids

Master Master
Station Station

1
2 N

              
(a)             (b)

Figure 2.19 (a) Single master–­multiple RTU (radial). (b) Single master–­multiple
RTU (shared line).

Master Sub
Station Master
Station

Sub Sub Sub Sub

Master Master Master Master
Station Station Station Station

RTUs RTUs

Figure 2.20 Multiple master–­multiple remote configuration.

2.11.4 Single master, multiple submaster, multiple remote

In this system there is a single master, with additional submasters, with
each submaster reporting to the master station. The remote RTUs/­IEDs
will typically be connected to the submasters. A typical system would be
the hierarchical transmission SCADA used. Figure 2.21 gives the practical
system, which is the National Control Center for the transmission SCADA
in India. The multiple submasters represent the five regional control cen-
ters and also 29 state load dispatch centers. The RTUs are located in the
substations around the country.
Chapter two: SCADA fundamentals 65

NLDC National
Level

2 Secs

RLDC RLDC RLDC RLDC RLDC Regional

Level
2 Secs

SLDC SLDC SLDC State

Level
2 Secs
SUB SUB SUB Distribution
LDC LDC LDC Level
12–15 Secs
Plant/Sub
RTU RTU RTU
Station
Level

Figure 2.21 Hierarchical setup of the Indian transmission system. (Courtesy of

POSOCO.)

2.12 SCADA implementation: A laboratory model [2,5,16]

The SCADA components are clearly depicted in a laboratory implemen-
tation where undergraduate and graduate students undergo a complete
commissioning process and gain hands-­on training on a real SCADA sys-
tem, as explained in the following section.

2.12.1 The SCADA laboratory

SCADA systems are used worldwide in a variety of automation appli-
cations in the gas and petroleum industry, power automation, building
automation, and small manufacturing unit automation. SCADA systems,
though used extensively by the industries, are proprietary to each com-
pany, and hence, very few technical details are available to students and
researchers. The setting up of the SCADA laboratory in India was of great
relevance as this provides research facilities in the form of hardware and
software for adaptive and intelligent control of integrated power systems.
SCADA systems have the following four components:

1. Master Station: This is a collection of computers, peripherals, and

appropriate I/­O systems that enable the operators to monitor the
state of the power system (or a process) and control it.
 causedby momentary chargein generaforspeecl,tI'r.r.tnr*,-i;;?t;qffi; ;;
excitationvoltagecontrolsare non-interactive
for smallchangesand can be

R
1.,
modelledandanalysed
F:tcl :tcfinrr
independently.
ri n
Furthermore,
r r rvrvr rhrirnvhr r
excitationvoltageeontrolis
tLhr cr v r- r, rr ci ^r rJ v r f i r r r o
rrlttw

field; while the power frequencycontrol is slow actingwith major time constant
n,rn..r,rhr ^6^,rri-+^-^.1
vrJrrJr-(lrrr urlLUultLtrlcu
:- rL^e ^$rL- -^-^--^-
r5 llla! ul ulc; ggirtcfalor

contributedby the turbine and generatormomentof inertia-this time constant

is much larger than that of the generatortield. Thus,the transientsin excitation
voltage control vanish much faster and do not affect the dynamics of power
frequencycontrol.

8.T
.INTRODUCTION

Powersystemoperationconsidcrcdso far was underconditionsof stcadyload.

However, both active and reactive power demandsare never steady and they
continually change with the rising or falling trend. Steam input to turbo-
generators(or water input to hydro-generators) must, therefore,be continuously
regulatedto match the active power demand,failing which the machinespeed
will vary with consequent change in frequency whieh may be highly
I
P+JQ
undesirable*(maximum permissiblechangein power fiequency is t 0.5 Hz). Fig. 8.1 schematicdiagramof loadfrequencyand excitation
Also the excitation of generatorsmust be continuouslyregulated to match the voltageregulatorsof a turbo-generator
reactive power demand with reuctive generation,otherwise the voltagesat
various system buses may go beyond the prescribed limits. In modern large Change in load demand can be identified as: (i) slow varying changes in
interconnected systems, manual regulation is not feasible and therefore meandemand,and (ii) fast random variations aroundthe mean. The regulators
automaticgenerationand voltage regulation equipment is installed on each mustbe dusignedto be insensitiveto thst randomchanges,otherwisethe system
generator. Figure 8.1 gives the schematic diagram of load frequency and will be prone to hunting resulting in excessivewear and tear of rotatins
excitation voltage regulatorsof a turbo-generator.The controllers are set for a machinesand control equipment.
particularoperatirrgcondition and they take care of small changesin load
denrandwithout fiequency and voltageexceedingthe prescribedlimits. With 8.2 LOAD FREOUENCY CONTROL (STNGLE AREA CASE)
the passage of time, as the change in lcad demand becomes large, the
contrcllersmust be reseteithernianuallyor automatically. Let us considerthe problemof controlling the power output of the generators
It has been shown in previous chaptersthat for small changesactive power of a closely knit electric areaso as to maintz,inthe scheduledfrequency. All the
is dependenton internalmachineangle 6 and is inderrendentof bus voltage: generatorsin such an areaconstitutea coherent group so that all the generators
.
whiie bus voitage is dependenton machine excitation (therefore on reactive speeoiip ancisiow riowii togetiierrnarntarnrng
^ - ^ ^ l - I ^ l - - - - . - l ^ - - - - ^ L - - - . r -
thelr reiarrvepower angies.Such
- ^ . _ _ _ : - , . r

- " - an area is defined as a control area. Tire boundariesof a coqtrol area will
Changein frequency causeschangein speedof the consumers' plant affecting generallycoincide with that of an individual Electricity Board Company.
productionprocesses.Further,it is necessaryto maintainnetworkfrequencyconstant To understandthe load fiequency control problem, let us consider a single
so that the powerstationsrun satisfactorily
in parallel,the variousmotorsoperating turbo-generatorsystem supplying an isolated load.
on the systemrun at the desiredspeed,correcttime is obtainedfrom synchronous
clocksin the system,andthe entertaining devicesfunctionproperly.
 W Modern power system Analys,s

Turbine Speed Governing System

Figure 8.2 shows schematicallythe speedgoverningsystem of a steamturbine. turbine. Its downward movement opens the upper pilot valve so that more steem
The systemconsistsof the following components: is admitted to the turbine under steady conditions (hence more steady power
. The reverse
Steam
Model of Speed Governing System

Assume that the system is initially operating under steady conditions-the

linkage mechanism stationary and pilot valve closed, stearnvalve opened by a
definite magnitude, turbine running at constant speedwith turbin" po*"r output
balancing the generator load. Let the operating conditions be characteizedby
Speed changer
=
"f" systemfrequency (speed)
P'c = generator output = turbine output (neglecting generator loss)

.IE = steam valve setting

We shall obtain a linear incremental model around these operating
conditions.
Pilot Let the point A on the linkage mechanism be moved downwards by a small
amount Aye.It is a commandwhich causesthe turbine power output to change
--t-\

value
and can therefore be written as
High
pressure oil Main Aye= kcAPc (8.1)
piston
where APc is the commanded increase in power. \
A The command signal AP, (i.e. Ayi sets into rnotion a bequenceof events-
I
rHydraulic amplifier the pilot valve moves upwards,high pressureoil flows on to the top of the main
(speed control mechanism) piston moving it downwards; the steam valve opening consequently increases,
the turbine generatorspeedincreases,i.e. the frequencygoes up. Let us model
Fig.8,2 Turbinespeedgoverningsystem these events mathematically.
Reprinted with permission
of McGraw-HiltBookCo., New York,from Olle l. Elgerd: Two factors contribute to the movement of C:
Electric Energy System Theory: An lntroduction, 1g71, p. 322.
(i) Ayecontributer - Aya or - krAyo(i.e. upwards) of - ktKcApc
[?J
\rll
(i) FIy ball speedgovernor: This is the heart of the system which sensesthe
(ii) Increase in frequency ff causes the fly balls to move outwards so that
changein speed(frequency).As the speedincreasesthe fly balls move outwards
B moves downwards by a proportional amount k'z Af. The consequent
and the point B on linkage mechanism moves downwards. The reversehappens
when the speeddecreases.
G) Hydraulic amplifier: It comprises a pilot valve and main piston movement of Cwith A remaining fixed at Ayo - . (+) orO, - + kAf
alrangement.Low power level pilot valve movement is converted into high
(i.e. downwards)
power level piston valve movement. This is necessaryin order to open or close
The net movement of C is therefore
the steamvalve againsthigh pressure steam.
(xl) Lintcage mechanism: ABC is a rigid link pivoted at B and cDE is AYc=- ktkcAPc+kAf (8.2)
anotherrigid link pivoted at D. This link mechanismprovides a movementto The movement of D, Ayp, is the amount by which the pilot valve opens. It is
the control valve in proportion to change in speed.It also provides a feedback contributedby Ayg and AyB and can be written as
the steamvalve movement (link 4).
,,fr9rn
Ayo=(h) Ayc+(;h) *,
 controt E
= ktayc + koAys 1
(g.3) E^,,^ri^- /o o\ :- . r .
riyLr.Lru' \o.o., rs rcpfesenleo ln tne ronn of a block diagram - t
The movement ay.o-d,epending upon its sign opensone of the ports of the pilot in Fig. 9.3.
valve admitting high pressure'o' into
thJ moving the main Ks9
piston and opening the steam valve "ynnJ.ithereby
by ayr. certain justifiable simprifying 1+ fsss 4Y5(s)
assumptions,which ean be rnade at this
.tugl, ur",
(i) Inertial reaction forces of main
pistoi and steam valve are negligible
compared to the forces exerteclon the
iirton by high pressureoil.
(ii) Because of (i) above, the
rate of oil admitted to the cylinder
proportional to port opening Ayo. is
The volume of oil admitted to the cylinder 4F(s)
is thus proportional to the time
integral o,f ayo. The movementay"i.s Flg. 8.3 ,Blockdiagramrepresentation
obtainedby dividing the oil volume by of speed governorsystem
the area of the cross-section of the-piston.Thus
The speed governing system of a hydro-turbine is more
involved. An
Avn= krfoeayrlat additional feedback loop provides temporary droop compensation
(8.4) to prevent
instability. This is necessitatedby the targe inertia or the penstoct
gut" which
It can be verified from the schematicdiagram regulates the rate of water input to the turbine. Modelling of
that a positive movement ayo, a hyjro-turbine
causesnegative (upward)movement ayulccounting regulating system is beyond the scopeof this book.
for the n"gutiu" ,ign used
in Eq. (8.4).
Taking the Laplacetransformof Eqs. (g.2),(g.3) Turbine Model
and (g.4), we ger
AYr(s)=- k&cApc(") + krAF(s) Let us now relate the dynamic responseof a steam turbine in tenns
(8.5) of changes
Ayp(s)= kzAyd,s)+ koAyug) in power ouFut to changesin steamvalve opening ^4yr. Figure g.4a
(8.6) shows a
two stage steam turbine with a reheat unit. The dynamic
*ponr" is targely
influenced by two factors, (i) entrained steambetwein the inlet
a y u ( g = - k s ol r U n (8.7)
stbamvalve and
first stageof the turbine, (ii) the storageaction in the reheaterwhich
causesthe
EliminatingAyr(s) andAyo(s), we can write output of the low pressurestageto lag behind that of the high pressure
'fttus, stage.
the turbine transfer function is characterizedby two time
k'ktk'AP' (s)- k,krAF(s) constants.For
AYu(s)- ') easeof analysisit will be assumedhere that the turbinl can be modelled
to have
(oo '' t Ssingle equivalent time constant.Figure 8.4b shows the transfer function model
"'tr ,/ of a sreamturbine. Typicaly the time constant lies'in the
\ { range o.i ro z.s
sec.
-lor,<,r-*^or",].i#) (8.8) -=-&
Steam valve

where

n= klct_
= speedregulationof the governor
K2

- gainof speed (a) Two-stage

steamturbine
K.,= governor
+y
.r.
, rs = l " = tlme constant of speed AYg(s)-FAPds)
- ;-;
KqkS r-- governor

(b) Turbinetransferfunctionmodel

Flg. 8.4
 #ph-Si po*", s),rt"r An"ly.i,
rrrroarrn
I Automatlc Generationand Voltage Control I
Generator Load Model
The increment
in powerinputto the generatbr-load
systemis =tAP6g)_
aPo(,)r.[#j (s.13)
APG _ APD
whele AP6 = AP,, incremental turbine
incremental loss to be negligible) and App is the load increment. 2H
This increment in power input to the syrtem is accountedfor in two ways: = pow€r systemtime constant
Bf"
(i) Rate of increase of stored kinetic energy in the generator
rotor. At
scheduledfrequency (fo ), the stored energy is
Kp,= =powersystem
gain
Wk, = H x p, kW = sec (kilojoules) +
where P, is the kW rating of the turbo-generator andH is defined as its inertia Equation (8.13) can be representedin block diagram form as in Fig. g.5.
constant.
The kinetic energy being proportional to square of speed(frequency),
the laeo(s)
kinetic energy at a frequency of (f " + Arf ) is given by ^Po(s)
16---ffioro,

Flg.8.5 Blockdiagramrepresentation
of generator-load
model
=nr,(r.T) (8.e)
complete Block Diagrram Representation of Load Frequenry
Rate of change of kinetic energy is therefore Control of an Isolated Power System

=fffrr"n
$rr*"r (8.10)
(ii) As the frequency changes,the motor load changesbeing
sensitive to
speed,the rate of changeof load with respectto frequ"n.y, i.e. arot\ycan
be
regarded as nearly constant for small changes in frequency Af ard can
be
expressedas
AP(s)=trPn15;
@PDl?flAf=BAf (8.11)
wherethe constantB can be determinedempirically, B is positivofor a
predominantly motorload.
AP6(s)
Writing the powerbalanceequation,we have

f . P] *' d (< o f l +B A f
, r^== T- H
APc- aP

Dividingthroughoutbyp, andrearanging,we get

Flg. 8.6 Blockdiagrammodelof loadfrequencycontrol
(isolatedpowersystem)
AP6$u)- AP;q;u)= 1d / A ' ^ +' B(ptt)
(Afi n ' 7 ' - -af
-\
(8.i2)
f dt
Steady States Analysis
Taking the l,aplace transforrn, we can write AF(s) as
AP,G) -4PoG) The model of Fig. 8.6 shows that there are two important incrementalinputs to
4Fis; - the load frequency control system - APc, the change in speedchangersetting;
B*-'- s and APo, the change in load demand. Let us consider,,.4,.simplesituatiqn in
 Modern
which the sneerl .hqnrro' hoo .. g.-.^) ^-.-. .rL^ ^L^--^ I
'(rr cr rr^tr(r ucttrng af = o) and the load r'E .1uuy' cquauon glves tne steadystate changesin frequency
r.han.,oo

:,:il?:
steady
't::;

;3l i:inT:
;-::--::^

:: frequen-cy a2
rreego,,* ;, 2
; ;;;*;:r;;ffi; *' #ff:Tiil:
\7'e' c demand
changes in load demand. Speed regulation R
causedby
is-naturally so adjusted that
change system fora sudd.n.hung", changesin frequency are small (of the order
ffi;ffi"ffi;ti'l; of 5vo from no load to ruu load).
Therefore,the linear incrementalrelation (g.16)ican
anaount
*, be appliedfrom no load to
(, e.Apog):+)is obtainedas follows: full load' with this understanding,Fig. 8.7
shows the linear relationship
betweenfrequency and load for free governor operation
with speedchanger set
to give a scheduledfrequency of r00% at full toao.
aF@)l*,(s):o
: - AP^ The .droop, or slopeof this
( 'l
I
^f
- is -l
relationship
\ B+(t/R) )
Power system parameter B is generaily much smalrer*
than r/R (a typical
value is B = 0.01 pu Mwalz and l/R = U3)
so that B canbe neglected in
comparison.Equation (8.16) then simplifies to

(8.17)
rhedroop fjfli;], curve
isthusmainly determinedby R, the
"r,,fl",
speedgovernorregulation.

ap,=_
K
r^sorr, I ( = 1. I
*"r: (r^;)o",
Decrease
in system
load= BAf= (uffi)*,
It is also rccognized that Ko, = 7 / B , w h e r e B - Y ^
/P' (in PuMWunit change
ai Of course,the contribution of decreasein system
in frequency). Now load is much less than the
increase in generation. For typical values of B and
R quoted earlier
APo = 0.971 APo
4=-(#6)o,. (8.16) Decreasein system load = 0.029 ApD
consider now the steady effect of changing
speed changer setting

(Or"<rl- load demandremainingfixed (i.e. Apo= 0). The

+)with sready
fi roa
(J
state changein frequency is obtained as follows.
L

8.rog
.c
*For
102 at 250 MW machine with an operating load of 125
li\ dA^t | | MW. let the change in load
\r,, ruu-loLoao be i%o for IVo change in frequency (scheduledfrequency =
101 50 Hz). Then
(ii) 60% Load
a-:?:r?: :2.5
100 NNVtHz
0 af 0.s
Percent Load

Flg. 8.7 Steady

governor
"*-l?39-frequency
system
qharacteristicof a speed '=(#)b :
#:
o'ol PuMwgz
 uodernPowersystemAnalysis Autor"tic G"n"r"tionand Volt"g" Conttol
W F
I trt stg t t v
f (t ^ p s AD
AF@lap,{s):o: xu'c (8.18)
( 1 +T , r s ) ( l * 4 sxl-
) * zors)+ KseKtK p,/R s

I
4,flr*uoyro,":_ t
KreKrKp, \
- lI
AP, (8.1e) Two generators rated 200 MW and 400 MW are operating in parallel. The
r l
I AP',:g
( 1 + K .sK tK ps/ R droop characteristicsof their governors are4Vo and 5Vo,respectivelyfrom no
load to full load. Assuming that the generatorsare operating at 50 Hz at no
If load, how would a load of 600 MW be sharedbetweenthem?What will be the
KrrK ,= l system frequency at this load? Assume free governor operation.

Ar=( | \rc" (8.20)

Repeat the problem if both governorshave a droop of 4Vo.
" \ B+llR) Solution Since the generators are in parallel, they will operate at the same
frequency at steadyload.
If the speed changer setting is changedby AP, while the load demand
Let load on generator 1 (200 MW) = x MW
changesby APo, the steadyfrequencychangeis obtainedby superposition,i.e.
and load on generator 2 (400 MW) = (600 - x) MW
- APo) Reduction in frequency = Af
Ar = ( (8.21)
ru) 'o" Now
". frequency changecausedby load demandcan be
According to Eq. (8.2I) the af_ 0.04x 50
(i)
compensated by changingthe settingof the speedchanger,i.e. x 200
APc- APo, for Af = Q af 0.05x 50 (ii)
Figure 8,7 depicts two load frequency plots-one to give scheduled 600-x 400
frequencyat I00Vorated load and the other to give the samefrequencyat 6O7o EquatingAf in (i) and (ii), we get
rated load.
231 MW (loadon generatorr)
v -

-/A trltf /1 ^-l ^-

600-x= JOy lvlw (IUau ull Btrrltrriltur L)

- 0'0-1150
' = 50
Systemfrequency x 231 = 47.69 Hz
200
A 100 MVA synchronousgeneratoroperateson full load at at frequencyof 50
It is observed here that due to difference in droop characteristics of
Hz. The load is suddenly reducedto 50 MW. Due to time lag in governor
governors,generatorI gets overloadedwhile generator2 is underloaded.
system,the steamvalve beginsto closeafter 0.4 seconds.Determinethe change
It easily follows from above that if both governorshave a droop of.4Vo,they
in frequencythat occurs in this time.
will share the load as 200 MW and 400 MW respectively,i.e. they are loaded
Given H = 5 kW-sec/kVA of generatorcapacity.
corresponding to their ratings. This indeed is desirable from operational
Solution Kinetic energy stored in rotating parts of generatorand turbine
considerations.
= 5 x 100 x 1.000= 5 x 105 kW-sec
Dynamic Response
Excesspower input to generatorbefore the steam valve
beginsto close= 50 MW To obtain the dynamic responsegiving the change in frequency as function of
the time for a step changein load, we must obtain the Laplace inverse of Eq.
Excess energy input to rotating parts in 0.4 sec (8.14). The characteristicequationbeing of third order, dynamic responsecan
= 50 x 1,000 x 0.4 = 20,000 kW-sec r'
Onfy
r
Dg
| 1-!-- - I
ODIalneU
f-,-
luf
-
A
-^^^tC: ^
SPtrUfffU
---*^-:^^1
ll|'llll('llua1'I
^^^^
Ua1DE.
tI^.-,^,,^-
II(rwsYsIr
+L^
LfIs
^L^-^^+^--i^+in
r,Il<ll4ivLsllDrlv

Stored kinetic energy oo (frequency)2 equation can be approximated as first order by examining the relative
Frequency at the end of 0.4 sec magnitudesof the time constantsinvolved. Typical valuesof the time constants
of load frequency control system are rdlated as
+ zo,ooo
= 5ox I soo,ooo )t"= 5r rfz
\ 500,000 )
 Trr4T, <To,
Time (sec)------->
Typically* t, = 0.4 sec, Tt = 0.5 sec and -1
tI
Firstorderapproximatiorl
I
o

Flg' 8.8 Firstorderapproximate brockdiagramof road Dynamicresponse_of

changein frequencyfor a stepchangein load
frequencycontrotof an isolatedarea (APo= 0.01pu, 4s = 0.4 sec, | = 0.5 sLc, Io. = 2b sec, (" = 100,
R= 3)
Irning Tro = T, = 0: K*\ =1), the block diagram of Fig.
Iuld 8.6 is
reduced to thlt of F'ig. 8.8, from which we can write The plot of change in frequency versus time for first order approximadon
^rirst
given above and the exact response are shown in Fig. a.g.
order
to, approximation is obviously a poor approximation.
AF(s)l*r(s):o = - .-.APo
(1+ KpslR)+ Zp.s" s
Gontrol Area Concept

- - "o{1:- =xaP, So far we have considered the simplified case of a single turbo-generator
supplying an isolated load. Consider now a practical system with number
, l, + ^ + r o ' 1 e
generating stations aird loads. It is possible to divide an extendedpower
of
L R4,J (say, national grid) into subareas(may be, State Electricity Boards)
system
in which
-ft{' - *,[-,,a[n#)]]*, g 22)
Ar(,)= the generatorsare tightly coupled together so as to form a coherent group,
all the generators respond in unison to changes in load o, ,p"rJ changer
i.e.

settings.Such a coherentareais called a control area in which the frequency

TakingR = 3, Kp, = llB = 100, = 20, Apo = is assumedto be the same throughout in static as well as dynamic conditions.
e, 0.01 pu
For purposes of developing a suitable control strategy, a control area can
be
Af (t) = - 0.029(I - ,-t:tt', (8.23a) reduced to a single speed governor, turbo-generator and load system. All
the
control strategies discussedso far are, therefore, applibable to an independent
Aflrt"udystare= - 0.029 Hz
(8.23b) control area.

Proportional Plus fntegral Control

It is seen from the above discussionthat with the speed governing sysrem
installed on each machine, the steady load frequency charartitirti" fi agiven
speedchanger setting has considerabledroop, e.g. for the system being used
for
the illustration above, the steadystate-droop in fieo=ueneywill be 2.9 Hz
[see
Eq. (8.23b)l from no load to tull load (l pu load). System frequency
specifications are rather stringent and, therefore, so much change in frequency
"For a 250 MW machinequoted
earlier,inertiaconstanrIl = SkW-seclkVA cannot be tolerated. In fact, it is expected that the steady change in frequency
will be zero. While steadystatefrequency can be brought back io the scheduled
,' = 4 :o. 2 * 5
Bf 0.01x ==2osec
50
ffil PowersystemAnalys
Modern AutomaticGenerationand VoltageControl
I
t-
vaiue by adjus'ringspeedchangersetting,the systemcould under go intolerable in ihe above scheme ACE being zero uncier steaciyconditions*, 4 logical
dynamic frequency changes with changes in load. It leads to the natural design criterion is the minimization of II,CZ dr for a step disturbance. This
suggestion that the speed changer setting be adjusted automatically by integral is indeed the time error of a synchronous electric clock run from the
monitoring the frequency changes.For this purpose, a signal from Af is fed power supply. Infact, modern powersystems keep Eaekofintegra+e4tinae errsr
througfan integrator to the s diagram
all the time. A corrective action (manual adjustment apc, the speed changer
configuration shown in Fig. 8.10. The system now modifies to a proportional
setting) is taken by a large (preassigned)station in the area as soon as the time
plus integral controller, which, as is well known from control theory, gives zero
error exceeds a prescribed value.
steady state error, i.e. Af lrt""d",,ut,= 0. The dynamics of the proportional plus integral controller can be studied
numerically only, the systembeing of fourth order-the order of the system has
Integral APe(s) APp(s) increasedby one with the addition of the integral loop. The dynamic response
controller
of the proportional plus integral controller with Ki = 0.09 for a step load

l+t-r8-
I I t - +
AF(s)
disturbance of 0.01 pu obtained through digital computer are plotted in Fig.
8.11. For the sake of comparisonthe dynamic responsewithout integral control
t l action is also plotted on the samefigure.
AP6(s)

sensor
Frequency

-1
plus integralloadfrequencycontrol
Fig. 8.10 Proportional +I
I
t
The signal APr(s) generatedby the integral control must be of oppositesign o
r
x
to /F(s) which accounts for negative sign in the block for integral controller.
Now
Kn,
AF(s1=
Ko,
(r+%"s).
(** +). ( l * f , r s ) ( l + 4 s )
RKo,s(l+{rs)(l+ 4s)
+ {'s)(1 + 4sXl f zo's)R* Ko'(KiR f s) "+ (8.24)
Flg. 8.11 Dynamicresponseof loadfrequencycontroller with and without
integralcontrolaction(APo = 0.01pu, 4s = 0.4 sec, Ir = 0.5
obviousry sec, Ips= 20 sec, Kp.= 100,B - B, Ki= 0.-09)
Af l"t"^dy = so/F(s) : o
state
(8.25)
,
In contrast to Eq. (8.16) we find that the steady state changein frequency 8.3 IOAD FREOUENCY CONTROL AND ECONOMIC
has been reduced to zero by the additio4 of the integral controller. This can be DESPATCH CONTROL
argued out physically as well. Af reaches steady state (a constant value) only
rr.,lrsrr
wlMl uAp^
rc-
- Ap-
HrD =
- .ons-fant
vvuulqr!. Becarrs-e of fhe intes!'atins actiOn Of the Load freouencv
I
__J control
________ with
,.___ inteorel
_---_O eonfrnller qnhierrAe
'vu
?a?^
lvrv
craolrr
otvsuJ
ora+o
Dl4lg

controller, this is only possible if Af = 0. frequencyelTor and a fast dynamic response,but it exercisesno control over the
In central load frequency control of a given control area, the change (error) relative loadings of various generating stations (i.e. economic despatch) of the
in frequencyis known as Area Contol Error (ACE). The additional signal fed control area. For example, if a sudden small increasein load (say, 17o) occurs
back in the modified control schemepresented above is the integral of ACE. 'Such
a control is known as isochronous control, but it has its time (integral of
frequency) error though steady frequency error is zero.
 f
Automatic
area, the road conrior ,changesthe speed changer _T---
1i..l1r_::ltrol
Dcrurgs -frequency command signai generated'oythe centrai economic despatch
or tne governors of all generating units of the area so that, computer.Figure
together,
theseunits match the load and the frequenry returns tp the scheduled 8'12 gives the schematicdiagram of both thesecontrolsior two
value (this typi.ut units of
action takesplace in a few seconds).However, in the,process a control area.The signal to changethe speedchan3ersetting
of this change the is lonstructed in
Ioadings of u@units accordancewith economic despatcherror,
change in a manner independent of [po (desired)- pJactual)]. suitabry
economi@ In fact, some units in the pro""r, may even modified by the signal representingintegral ncg at that instant
of time. The
get overloaded.Some control over loading of individual signal P6 (desired) is computed by the central economic despatch
units cafi be Lxercised computer
by adjustingthe gain factors (K,) includeJin the signal (CEDC) and is transmitted to the local econornic despatch
representingintegral of controller (EDC)
the area cogtrol error as fed to individual unitr. installed at each station. The system thus operateswith economic
However, this is not desfatch error
satisfactory. only for very short periods of time beforJ it is readjusted.

"fnceot 8.4 TWO-AREA LOAD FREOUENCY CONTROL

An extendedpower system can be divided into a number

of load frequency
control areasinterconnectedby meansof tie lines. Without loss
of generality we
shall consider a two-area case connectedby a single tie line
aslilusnated in
Fi g. 8.13 .

Speed

Fig.B.i3 Two interconnected

controrareas(singre
tie rine)
The control objective now is to regulatethe frequency of each
area and to
srnnultaneously regulatethe tie line power as per inter-areapower contracts.
As
in the caseof frequency, proportional plus integral controller
will be installed
so as to give zero steady state error in tie line power flow as
compared to the
contractedpower,
EDC - Economic despatch controller It is convenientlyhssumedthat each control area canbe represented
lr. by an
CEDC - Central economic despatch computer equivalentturbine, generatorand governorsystem.Symbols used
with suffix I
refer to area 7 and those with suffi x 2 refer to area 2.
Flg. 8-12 Control area load frequency and economic despatch In an isolated control area case the incremental power (apc _ apo)
control was
accountedfor by the rate of increaseof stored kinetic energy and increase
in
Reprinted (with modification) with permission of McGraw-Hill areaload causedby increasein fregueircy.since a tie line t *rport, power
Book Company, in
New York from Olle I. Elgerd: Electric Energy SystemsTheory: or out of an area, this fact must be accounted for in the incremental
An Introd.uction, power
I971,p. 345. balanceequation of each area.
Power transported out of area 1 is .eivenbv
''rrl''l
Ptie,r = sin ({ - q (8.26)
X,,
where
q'q - poweranglesof equivalentmachinesof thetwo
areas.
 I Automatic
Generation
and Vortagecontror
308 | Modefn Power SystemAnalysis
I Fil
I APti".r(s)
For incrementalchangesin { and 6r, the incre.mentaltie line power can be
expressed
as
AP,i,,r(pu)= Tp(Afi - 462) (8.27)
where
'Y:t'-Yf
T, = cos(f - E) - synchronizingcoefficient
PrrXrz
Fig. 8.14
Since incrementalpower anglesare integrals of incrementalfrequencies,we
can write Eq. (8.27) as The corresponding block diagram is shown in Fig. g.15.

AP,i,,r = 2*.(l Afrdt- (8.28)

I Urat) +
APti",r(s)
where Afi nd Af,, arc incremental frequency changes of areas 1 and 2,
respectively.
Similarly the incrementaltie line power out of area2 is given by -iE= --n7ri"l
AF1(s)
aPt;", z = 2ilzr([ yrat - ayrat) (8.2e)
[ Fig. 8. 15
where
For the control area 2, Ap6", r(s) is given by tEq. (g.Zg)l
rzr
LL
= tYr:J cos({L - E): [S]ti z: ar2rrz (s.30) - : gr r r ,
Przxzr "
\Prr)
apt i", z( s)= [ AFr ( s)- 4F, ( s) ] ( g: 35)
With referenceto Eq. (8.12), the incrementalpower balance equationfor which is also indicated Uy ,i. block diagramof Fig. 8.15.
area 1 can be written as \
Let us now turn our attentionto ACE (areacontrol error; in the presence
of
a tie line. In the case of an isolated control area, ACE is the change in area
APo,- APor= + *w)+ nrz|r* AP,,",t (8.31)
frequency which when used in integral control loop forced the steady state
Jr" or
frequencyelror to zero. In order that the steadystatetie line power error
It rnay be noted that all quantitiesother than fiequencyare in per unit in in a
two-areacontrol be made zeroanotherintegralcontrol loop (one for each area)
Eq.(8.3l). must be introducedto integratethe incrementaltie line power signal and
Taking the l-aplacetransf'ormof Eq. (8.31) and reorganizing,we get feed
it backto the speedchanger.This is aeeomplished by a singleintegratingbloek
by redefining ACE as a linearcombinationof incrementalfrequenryand tie line
AF (s ) = IA P6 1 G)- A P r,(s) - A P ti " ,,1r;] $.32) power. Thus, fbr control area I
" t$-
I + 4,,t,!
ACEI = APu".r+ brAf, ( 8. 36)
where as definedearlier [seeEq. (8.13)]
where the constant b, is called areafrequency bias.
Kp31= I/81
Equation (8.36) can be expressedin the Laplace transform as
Tpil = LHr/BJ" (8.33) ACEl(s) = APo., r(s) + b1AF1g) (8.37)
Comparedto Eq. (8.13) of the isolated control areacase,the only changeis Similarly, for the control are a 2, ACE2 is expressedas
ol the signal APri"J(s) as shown in Fig. 8.14.
the appearance
'-l'^Li-- fho T -^l-^a fL*lo4nl l oDfrnurrm ^ufr E^ /a ta\ tl h
ACEr(s) = APti".z(s)+ b2AF,(s) ( 8. 38)
I4ArrrS rrrw lsl/l4vv rrr LY. \v.L9), l rav c iro6 nr rosl^
o ,4P /"\
tie.I\.r/
ic nlrfoinerl
",
AS
Combining the basic block diagramsof the two control areascorresponding
to Fig. 8.6, with AP5rg) and Apr2(s) generatedby integrals of respective
=
AP,i.,1(s) - /4 (s)l (8.34) ACEs (obtainedthrough signalsrepresentingchangesin tie line power and local
ffroor(s) frequencybias) and employing the block diagramsof Figs. g.t+ to g.15, we
easily obtain the composite block diagram of Fig. g.16.
 WIU&| ModernPower svstem Analvsis

Let the step changesin loads APo, and APrrbe simultaneouslyappliedin

control areas 1 and 2, respectively.When steady conditions are reached,the
output signals of all integratingblocks will become constantand in order for
this to be so, their input signalsmust become zero. We have, therefore, from
F i e .8 . 1 6
KtL)
APu",, + b rAfr= block-
- O finputof integrating (8.39a)
\ ,r) trJ
o .Y
o
K'z) (U
- o finpot of integratingblock-
APti",, + brAfr= (8.3eb) oi -o
\ r l ra !t
o
I
il
g
Afr - Afz- =o finpurot
\ integrating
block
s -'4'\ ) (8.40) S A oy,
EF
N
o
o-
FromEqs.(8.28)and(8.29) 8 6
a(\ :pE
APn",, I.=constant .g
=-Tr, -. (8.41) q
<.1
o 6
=a c)
AP.i",z, Tzt; ar2 - ( d
E
E 9 o
*li o
HenceEqs.(8.39)- (8.41)aresimultaneously
satisfiedonly for ol. a
v'it
g t u
Q c l o
e.>
lr o
A P r i " , r =A P , : " , 2 = 0 (8.42) (g() o
o o G'
and Afi=Afz=0 F A 6
O E
Thus, under steady condition change in the tie line power and frequency of *5 o*
o
b 6
each area is zero. This has been achieved by integration of ACEs in the o
tr(') E
o
feedbackloops of each area. E g E
Dynamic responseis difficult to obtain by the transfer function approach (as H'.s (D
()
(U
used in the single area case)becauseof the complexity of blocks-and multi- tt)
* 3 CL
.9 5 u .n
input (APop APor) and multi-output(APri",1, Ap6",2, Afr Afr) situation.A d = o o
more organizedand more convenientlycarried out analysisis through the state o
' 8' 9
a 5 o
spaceapproach(a tirne domainapproach).Formulation of the statespacemodel oo. r\
for the two-area system will be illustrated in Sec. 8.5. *,n
EO- ai
o o
The results of the two-areasystem(APri", changein tie line power and,Af, o o cit
changein frequency) obtainedthrough digital computerstudy are shown in the lr
@

form of a dotted line in Figs. 8.18 and 8.19. The two areasare assumedto be d
identical with system parametersgiven by <; t
l! -ld
Trs= 0.4 sec, 7r = 0.5 sec, ?r, = 20 sec l+
5l
K o r =1 0 0 ,R = 3 , b = 0 . 4 2 5 , & = 0 . 0 9 , 2 f l r 2 = 0 . 0 5
u I f|:-
8.5 OPTTMAL (TWO-AREA) LOAD FREOUENCY CONTROL

Modern control theory is applied in this section to design an optimal load

frequencycontroller for a two-a3easystem.In accordancewith modern control
terminology APcr arrdAP62 will be referred to as control inputs q and u2.ln
the conventional approachul anduzwere provided by the integral of ACEs. In
 AutomaticGenerationand Voltagecontrot M&
itZ I rrrrodern
- SystemAnalysis
Power f*
ComparingFigs. 8.16 and 8.17,
moderncontrol theory approach ur and u2 wtll be createdby a linear
For formt'latingthe xt = Aft xq = Af. XS= JACEit
combinationof all the systemstates(full statefeedback).
statevariablernodel for this purpose the feedback
conventional loops are
.r2- AP,;1 x5 = AP52 t, = JACE, dt
resentedbv a se block as shown in
Fig. 8.17. State variables are defined as the outputsof all blocks baving either tt1= APg, u2 = /)Pa
We immediately notice that the systernhas nine
an integratoror artirne,constanf..
w1= AP" w2 = APp,
state variables.
For block 1
-1-+-r+--i, -f-.'f-- x1 + T. r r i, = K ^ t ( x z -
L P h
- w)
+I
\ . 1 Kprt Kprt -, Kprt
I
Optimal case (full state feedback)
, x, -z - ; - x t -;-wt
-
*f (8.43)

I
I
- l h l 4 l

o 'psl t psl t ptl t ptl

o
X

tt-
For block 2
( t'-2
x.2+ Tiliz= xt
-3 '
With integralcontrolaction
or * z =- + - r * * n
1 ,I For block 3
( 8. 445

F i g . 8 .1' 8 change in tie linepowerdue to step load(0.01pu) changein area1

t r + { , s r i ': = R- L r, r + rr , r
-21.--+-_'--';-;;7-1=a.-1-1=--1
r;+-.1
' - I- - ' 16 18 20
/' 8 12 14
* t=-
A
L / Time(sec)----- or r,- (8.45)
N
I ^h t* ,* * ,,
o
I
with integral control action For block 4
x
IL
X +n * Torz*+= Krrz(xs + ar2x7 - wz)

I Knrz at?K or2 Ko*2

Optimalcase (full state feedback) or iq= - T -- - y ' - a - _ - W ^
( 8. 46)
-
'\A'1--.{<
'' ' z
Tprz Tps2 Tps2 Tpsz

For block 5

Fig. 8.19 Change in frequencyof area 1 due to step load (0.01 pu) x s t 7,2i5 - x6
change in a.rea1 l 1
or is= 4
Y
r<
I - V
t 4 l
u ( 8. 47)
Ttz T,z
Before presenting the optimal design, we must formulate the state model.
This is achieved below by writing the differential equations ciescribing each For block 6
individual block of Fig. 8.17 in terms of state variables (note that differential . l
x s * I ,szx6- -; x4 + u2
I\2
equations are written by replacing s UV
' *1.
dt'
or io=-#*o-**u ( 8. 48)
'2t sg2 t sg2
'3i4',"1 ModernPower SystemAnatysis
T
For block 7 '-- co","".t""
it=2iTtzxt-2iTr2xa (8.4e) "
constructed as under from the state variables x, and -rnonly.
For block 8
ut=- Kirxs=- Kir IeCn,Ar
is= brx, + x.i (8.5O-)
For block 9 uz=- Ki{s=- Kiz la.Cerar
i9= b2xa- anxt ( 8 . s1 ) ln the optimal control schemethe control inputs u, anduz aregeneratedby
means of feedbacks from all the nine states with feedback constants to be
The nine equations(8.43) to (8.51) can be organizedin the following vector
determinedin accordancewith an optimality criterion.
matrix form
Examination of Eq. (8.52) reveralsthat our model is not in the standardform
*=Ax+Bu+Fw (8.s2)
where
employed in optimal control theory. The standardform is

x _ l x r x2 ... xg)r = state vector i=Ax+Bu

which does not contain the disturbance term Fw present in Eq. (g.52).
u = f u t u2fT= control vector
Furthermore,a constantdisturbancevector p would drive some of the system
w = l w t w2fT = clisturbancevector statesand the control vector z to constantsteadyvalues;while the cost function
while the matrices A, B and F are defined below: employedin optimal control requiresthat the systemstateand control vectors
have zero steady state values for the cost function to have a minimum.
I 2 3 4 5 7 8 9 For a constant disturbancevector w, the steady state is reachedwhen
Y
'tPsl
I
0 0 0 -bLoo * = 0
Tpst, Tprt Tprt
in Eq. (8.52);whichthengives
0 - 1 1 o o 0 0 0
Tt Ttr 0 = A . r r "+ B u r r + F w (8.s3)
1 - 1 Definingx and z as the sumof transientand steadystateterms,we can write
o o o 0 0 0
Rr4er Trst = ,
x x' * Ir" (8.54)
atzKprz
0 0 0 - 1 K p ' z 0 0 n = ut * z', (8.55)
A _ Tprz Tprz Tprz
Substitutingr and z from Eqs. (8.54) and (8.55) in Eq. (8.52), we have
--1- 1
0 o o i o 0 0 0 i' = A (r/ + x"r) + B(at + usr)+ Fw
Ttz 7,,
I By virtue of relationship(8.53),we get
6 0 o o - 1 0 0 0 0
RzTrsz TreZ *' = Axt + But ( g. 56)
7 2 irrz 0 0 -2ilr2 0 0 0 0 0 This represents system model in terms of excursion of state and conhol
8 bL 0 0 0 0 0 1 0 0 vectors fiom their respectivesteadystate values.
9 0 0 0 b 2 0 0 -atz 0 0 For full state feedback, the control vector z is constructed by a linear
combination of all states.i.e.
[I o o TI o o o 00.l u=- Kx (8.57a)
-ss1 I
Br = | I where K is the feedback matrix.
l o oo o o +
I
L
aco)
'O-
0 0l
J
I
Now
ttt+ Itrr=- l( (r/+ rr")
Kprt For a stable system both r/ and ut go to zero, therefore
- 0
,;T Tprt ur, = _ Kx*
Hence
/= -
tt Ikl (8.s7b)
 Modern Power SystemAnalysis AutomaticGeneration
Examinationof Fig. 8.17 easily reveals the steady state values of state and
control variablesfor constantvaluesof disturbanceinputs w, andwr. These are
b ? o0 0 0 0 4 0 0
0 0 0 0 0 0 0 0 0
Ilrr=X4"r= /7r" = 0
0 0 0 0 0 0 0 0 0
-arzbz 0 0
ulr, = wl
0 0 0 0 0 0 0 0 0
r5rr= x6rr= lv2 (8.s8) 0 0 0 0 0 0 0 0 0
uzr, = wz 4 0 0 - a n b z 0 0 Q+a?)o o
Igr, = COnstant 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 1
I9r, = Constant
= symmetric matrix
The values of xr* and xe* depend upon the feedback constants and can be
determined from the following steady state equations: R - kI = symmetric matrix
utrr= kttxtr, + ... + ftt8r8", * kt*sr, = wl

r,t2ss= k2txlr, + ... + kzgxgr.,

* kz*gr, = wz (8.se)
The feedbackrnatrix K in Eq. (8.57b) is to be determined so that a certain
performance index (PI) is minimized in transferring the system from an
arbitrary initial state x' (0) to origin in infinitie tirne (i.e. x' (-) = 0). A K = R-rBrS (8.63)
convenient PI has the quadratic form The acceptablesolution of K is that for which the systemremainsstable.
SubstitutingEq. (8.57b) in Eq. (8.56), the systemdynamicswith foedback
' Pr= + u'r Ru'dt is
;ll '.'' Qx'
(8.60)
definedbv
The manices Q arrd R are defined for the problem in hand through the i' = (A - BIgx, (g.64)
following designconsiclerations: Fol stability all thc cigenvaluesof the matrix (A - Bn shouldhave
negative
(i) Excursions of ACEs about the steadyvalues (r,t + brx\; - arrxt, + bzx,q) real parts.
are minimized. The steady values of ACEs are of course zero. For illustration we considertwo identical control areaswith the
following
syste|llparameters:
(ii) Excursionsof JnCg dr about the steadyvalues (xts, xte)are nrinimized.
The steaclyvaluesof JeCg dt are,of course,constants. 4r* = 0'4 scc; T'r= 0.5 sec; 7'r* = 20 sec

(iii) Excursionso1'the contt'ol vector (ut1,ut2) about the steadyvalue are /l = 3: (n* = l/lJ = 100
rninirnized.The steadyvalueof the controlvectoris, of course,a constant. b = O . 425;Ki = 0. 09; up = I ; 2iln = 0. 05
' This nrinimizationis intendedto indirectlylimit the controleffbrt within
the physical capability of components.For example, the steam valve
catmot be openedmore than a certain value without causingthe boiler
presisureto drop severely.
With the abovereasoning,we can write the PI as

pr= * fU-+ + h,.r,,)2+ (- tt,2xt,+ brxta)z 0.52tt6 l.l4l9 0.68l3 - 0.0046-0.021| -0.0100-0.7437 0.gggg0.00001
+ (.r,?+ ,,]) ^f, = [
2Jtt'
L-o.tl046-0.o2tl-0.0100 0.5286 t.t4rg 0.6813 0.74370.0000 0.gggsl
+ kfu'l+ u,|11
at (8.61)
From the PI of Eq. (8.51), Q md R can be recognizedas
'*Refer
Nagrath and Gopal [5].
 iiii'f:l Modernpowerrystemin4gs
As the control areasextend over vast geographical regions, there are two
c= vR.f- vr '_-
ways of obtaining full state information in each area for control purposes.
(i) Transport the state information of the distant area over communication The error initiates the corrective action of adjusting
the alternator excitation.
channels.This is, of course,expensive. Error wave form is suppressedcarrier modulated,
tt" carrier frequency being
the system frequency of 50 Hz.

Load change
Change in voltage
caused by load

tG
1+Iers

skrt

8.6 AUTOMATIC VOLTAGE CONTROL Fig. 8.21 Brockdiagramof arternator

vortagereguratorscheme
Figure 8.20 gives the schematic diagram of an automatic voltage regulator Error amplifier: It demodulatesand amplifies the
of error signal. Its gain is Kr.
a generator.It basicallyconsistsof a main exciter which excites the alternator scR power amplffier and exciter
fierd: It provides the n"."rriry power
field to control the output voltage. The exciter field is automaticallycontrolled amplification to the signal for controlling thl exciter
n"ro.- arr*;"g ,rr"
through error e = vr"r - vr, suitably amplified through voltage and power amplifier time constant to be small enoughio be neglected,
the ovelail fansfer
amplifiers. It is a type-0 system which requiresa constant error e for aspecified function of these two is
voltage at generatorterminals. The block diagram of the systemis given
in K,
l* T"rs
where T"yis the exciter field time constant.
Alternator; Its field is excited by the main exciter voltage
vu. Under no road
L it producesa voltage proportional to field current.The
no load transferfunction
o is
A
D Ks
7*T*s
where
Potential
T*= generatorfield time constant.
The load causes a voltage drop which is a complex
function of direct and
Fig. 8.20 Schematicdiagramof alternatorvoltageregulatorscheme quadratureaxis currents.The effect is only schematically
reBresented hv hlock
G.. The exact load model of the alternatoris beyond ,t"
Fig. 8.21. The function of important componentsand their transferfunctions ,iop" ;rhtJ;;:
is stabitizing transformer: T4*d
given below: are large enough time constantsto impair
Potential transformer: It gives a sample of terminal voltage v.. the system's dynamic response.-lq
Itjs weil known that the dynami. r"rpoor" of
Dffirencing device; It gives the actuating error a control system can be improved by the internal derivative feedback
loop. The
derivative feedback in this system is provided by means
of a stabiyzing
transformer excited by the exciter output voltage vE. The
output of the
 320' l
I
Modern Power SystemAnalysis Automatic and VoltageControl
Generation ------_-----l Jffif
I E

transformeris fccl ncgativclyat the input terminalsof thc SCR power

stabiliz,ing The bandedvaluesimposedhy the limitersareselectedto resffictthe generation
amplifier. The transferfunction of the stabilizing transfo"meris derived below. rate by l}Vo per minute.
Since the secondaryis connectedat the input ternfnals of an amplifier, it can I
be assumedto draw zero current. Now
dt
vr = Rr i., + LrJilL
' d t I.g 9t",

u't +/
_+(
'rr= MY A
dt l-
-t*9r"'--l
Taking the Laplace transform, we get

%,(s) _ sM sMlRt
VuG) R, * s,Lt l*Irs

sK", Fig.8.22 Governormodelwith GRC

1 +{ , s The GRCs result in larger deviationsin ACEs as the rate at which generation
Accurate staterrariablemodels of loaded alternator around an operating point can cha-ngein the area is constrainedby the limits imposed. Therefore, the
are available in literature using which optimal voltage regulation schemescan duration for which the power needsto be imported increasesconsiderably as
be devised.This is, of course,beyondthe scopeof this book. cornparedto the case where generationrate is not constrained.With GRCs, R
should be selectedwith care so as to give the best dynamic response.In hydro-
8.7 LOAD FREOUENCY CONTROL WITH GENERATION thennal system,the generationrate in the hydro area norrnallyremainsbelow
RATE CONSTRAINTS (GRCs) the safe limit and therefore GRCs for all the hydro plants can be.ignored.

frcquencycontrolproblcmdiscussedso far doesnot consiclerthe effect

The l<-racl 8.8 SPEED GOVERNOR DEAD-BAND AND ITS EFFECT
of the restrictionson the rate of changeof power generation.In power systems ON AGC
havingsteamplants,powergeneration canchangeonly at a specifiedmaximum
rate. The generationrate (fiom saf'etyconsiderationso1 the equipment)for The eff'ectof the speed governor dead-bandis that for a given position of the
reheat units is quit low. Most of the reheatunits have a generatiol rate around governor control valves, an increase/decrease in speed can occur before the
3%olmin. Some have a generation rate between 5 to 7jo/o/min.If these positionof the valve changes.The governordead-bandcan materiallyaffect the
constraintsarc not consirlcrcd,systerttis likely to c:ha.sc largc tttottrclttrry system response.ln AGC studies, the dead-band eff'ect indeed can be
disturbances,Thrs results in undue wear and tear of the controller. Several significant,sincerelativcly small signalsare under considerations.
methoclshave been proposecl to consider the effect of GRCs for the clesignof TlLespeedgovernorcharacterristic.thoughnon-lirrear, hasbeenapproxinraaed
automatic generation controllers.When GRC is considered, the systeln dynamic by linear characteristicsin earlier analysis. Further, there is another non-
rnodelbecomesnon-linearand linearcontroltechniquescannotbe appliedfor iinearity introducedby the dead-bandin the governor operation.Mechanical
the optimizationof the controllersetting. f'riction and backlashand also valve overlapsin hydraulic relays cause the
If the generationratesdenotedby P", are included in the statevec:tor,the governor dead-band.Dur to this, though the input signal increases,the speed
systermorder will be altered.Insteadof augntentingthem, while solving the governor may not irnmediately react until the input reachesa particular value.
stare equations,it may be verified at each step if the GRCs are viclated. Similar a.ctiontakesplace when the input signal decreases. Thus the governor
Another way of consicieringGRCs for both areas is to arjri iinriiers io ihe dead-bandis defined as the total rnagnitudeof sustainedspeedchangewithin
governors[15, 17] as shown in Fig. 8.22, r.e., the maximum rate of valve which there is no change in valve position. The limiting value of dead-bandis
openingor closingspeedis restrictedby the limiters.Here 2", tr,r,, iS the power specifiedas 0.06Vo.It was shown by Concordia et. al [18] that one of the
rate limit irnposedby valve or gate control. In this model effects of governor dead-bandis to increasethe apparentsteady-statespeed
l A Y E l . - -g u , n r (8.6s) regulation R.
 lFFf ModrrnPo*., svrt.t Analuri,
The effect of the dead-bandmay be included in the speed governor control
loop block diagramas shown in Fig. 8.23.Consideringthe worst caseforthe
dead-band,(i.e., the system starts responding after the whole dead-band is
traversed)and examiningthe dead-bandblock in Fig. 8.23,the following set of
ly define the behaviourolthe dead.band [9]- Discrete-Time Control Model
The continuous-timedynamic systemis describedby a set of linear differential
equations
x=Ax+Bu+ fp (8.67)
where f u, P are state, conhol and disturbance vectors respectively and
A,B
and f are constantmatrices associatedwith the above vectors.
The discrete-timebehaviourof the continuous-timesystemis modelled by the
system of first order linear differenceequations:

Speed governor
x(k+1)=Qx(k)+Vu(k)+jp&) (8.68)
Dead-band where x(k), u(k) and p(k) are the state,control and disturbancevectors and are
specifiedat t= kr, ft = 0, 1,2,... etc.and ris the samplingperiod.
controlloop
in speed-governor
Flg. 8.23 Dead-band 6, tl,nd
7 Te the state, control and disturbance transition matrices and they are
evaluatedusing the following relations.
u(r+1)= 7(r) 1: _ x, 1 dead-band
"(r+1)
- _ dead-band; if x('+l) - ,(r) I g (8.66) d= eAT
"(r+l)
- tf Xr*l _ xt < 0
{=({r_ln-tr
"(r+1). j=(eAr-DA-tf
(r is the step in the computation)
where A, B and,I are the constantmatrices associatedwith r, ,,LO p vectors
Reference[20] considersthe effect of governor dead-bandnonlinearity by using in the conespondingcontinuous-timedynamic system. The matrix y'r can be
the describingfunction approach[11] and including the linearised equationsin evaluatedusing various well-documentedapproacheslike Sylvestor's expansion
the state spacemodel. theorem, series expansion techniqueetc. The optimal digital load frequency
The presenceof governordead-bandmakesthe dynamicresponseoscillatory. controller designproblem is discussedin detail in Ref [7].
It has been seen [9J that the governor dead-banddoes not intluence the
selectionof integral controller gain settingsin the presenceof GRCs. In the 8.10 DECENTRALIZED CONTROL
presenceof GRC and deadband even for small load perturbation,the system
becomeshighly non-linear and hencethe optimization problem becomesrather In view of the large size of a modern power system, it is virtually impossible
complex. to implement either the classicalor the modern LFC algorithm in a centralized
manner.ln Fig. 8.24, a decentralizedcontrol schemeis shown.x, is usedto find
8.9 DIGITAL LF CONTROLLERS out the vector u, while x, aloneis employed to find out u". Thus.

In recent years,increasinglymore attentionis being paid to the questionof

digital implementation of the automatic generationcontrol algorithrns. This is
mainly due to the facts that digital control turns out to be more accurate and
rcliqhlc
r v^rEv^vt
nnrnnaef in qize less censifive to nnise end drift nnd more flexihle Tt

may also be implemented in a time shared fashion by using the computer

systemsin load despatchcentre,if so desired.The ACE, a signal which is used
for AGC is availablein the discreteform, i.e., there occurssampling operation
; betweenthe systemand the controller. Unlike the continuous-timesystem,the
control vector in the discretemode is constrainedto remain constant between
Flg. 8.24 Decentralized
control
i,i2[,:*.1 ModernPowerSystemAnalysis - Automatic and voltageControl
Generation ffi
4
x - (x1 x2)'
- aF(s)' 1' af (t)dr: liq, * '4F(s): hm/F(")]
ut=-ktxt IL n,n,, tf^461dv
JO ,t JO s-0 S s+0

u2-- kzxz
8.4 For the two area load frequencycontrol of Fie. 8.16 assumethat inte
controller blocks are replacedby gain blocks, i.e. ACEI and ACE are fed
been shown possible using the modal control principle. Decentralized or
to the respective speedchangersthrough gains - K, and - Ko. Derive an
hierarchicalimplementationof the optimal LFC algorithmsseemsto have been
expressionfor the steadyvalues of changein frequency and tie line power
studied more widely for the stochasticcase since the real load disturbancesare
for simultaneouslyapplied unit step load disturbanceinputs in the two
truely stochastic.A simple approachis discussedin Ref. [7].
areas.
It may by noted that other techniquesof model simplification are available
in the literatureon alternativetools to decentralizedcontrol. Theseinclude the 8.5 For the two area load frequencycontrol employing integral of area control
method of "aggregation", "singular perturbation", "moment matching" and error in each area (Fig. 8.16), obtain an expressionfor AP6"$) for unit
other techniques[9] for finding lower order models of a given large scale step disturbance in one of the areas.Assume both areas to be identical.
system. Comment upon the stability of the system for parameter values given
below:
4e = 0'4 sec; Z, = 0'5 sec; Zp. = 20 sec
IEI/IS
PROB K p r = 1 0 0 ;R = 3 ; K i = l ; b = 0.425

8 . 1 Two generatorsrated 200 MW and 400 MW are operating in parallel. ar2= I;2tTr, = 0.05
The droop characteristicsof their governors are47o and 5Vorespectively lHint: Apply Routh's stability criterion to the characteristicequation of
from no load to full load. The speedchangersare so setthat the generators the system.l
operate at 50 Hz sharing the full load of 600 MW in the ratio of their
ratings.If the load reducesto 400 MW, how will it be sharedamong the
generatorsand what will the s)/stemfrequency be? Assumefree governor NCES
REFERE
operatlon.
The speedchangersof the governorsare resetso that the load of 400 MW
Books
is sharedamong the generatorsat 50 Hz in the ratio of their ratings. What
are the no load frequencies of the generators? l. Elgcrd, O.1., Elccu'ic Energv.Sv,s/clrT'lrcorv: An ltttnxlut'lion. 2nd cdn. McCraw-
8 . 2 Consider the block diagrammodel of lcad frequencycontrol given in Fig. Hill, New York, 1982.
8.6. Make the following approximatron. 2. Weedy, B.M. and B.J. Cory Electric Pow'er Systems,4th edn, Wiley, New York,
(1 + Z.rs) (1 + Z,s) =- t + (7rg + T,),s= 1 + Z"c.r I998.

Solve for Af (l) with parametersgiveu below. Given AP, - 0.01 pu

a
1 Cohn, N., Control of Generation and Power Flou, on Interconnected Systents,
Wiley, New York, i971.
T"q= 0.4 + 0.5 = 0.9 sec; 70, = 20 sec
4 . Wood, A.J., and B.F. Woolenberg, Power Generation, Operation and Control,2nd
K r r K , = 1 ;K p r = 1 0 0 R
; =3 edn Wiley, New York, 1996.
5 . Nagarth, I.J. and M. Gopal, Control Systems Engineering, 3rd edn. New Delhi,
Coinparewith the exact responsegiven in Fig. 8.9.
2 0 0l .
8 . 3 For the load frequency control with proportional plus integral controller
oc olrn'rn.i- T i iLc6 . e 1n nhfain en AsnrAccinn fnr tha cfenrlrr cfrfp errnr in
6 . Handschin, E. (Ed.), Real Time Control of Electric Power Systems, Elsevier, New
clJ orlvYYll ll( L v. rvt vuLarrr
York 1972.
7 . Mahalanabis, A.K., D.P. Kothari and S.I Ahson, Computer Aided Power Systent
cycles,i.". f'41t)d
t ^ "
r; for a urrit stepAPr. What is the correspondingtime
, 1 , Analysis and Control, Tata McGraw-Hill, New Delhi, 1988.
lirnl*m
8 . Kirclrrnayer, L.K., Economic Control of lnterconnected Systems,Wiley, New York,
error in seconds(with respectto 50 Hz).lCommenton the dependenceof t959.
error in cycles upon the integral controller gain K,. 9 . Jamshidi, M., Inrge Scale System.s:Modelling and Control, North Holland, N.Y.,
1983.
 10. Singh, M.G. and A. Titli, SystemsDecomposition,
Optimization and Control
PergamonPress, Oxford, 197g.
I I' Siljak, D'D., Non-LinearSystems:The Parametcr Analysis
antl Design, Wiley,
N.Y. 1969.

.t,apers

12. Elgerd, o.I. and c.E..Fosha,"The Megawatt Frequency

control problem: A New
Approachvia optimal control Theory", IEEE Trans.,April
1970,No. 4, pAS g9:
556.
13' Bhatti, T'S., C.S Indulkar and D.P. Kothari, "Parameter
optimization of power
Systemsfor StochasticLoad Demands" Proc. IFAC. Bangalore,
December 19g6.
l4' Kothari, M'L., P.S. Satsangiand J. Nanda,"sampled-Data
AutomaticGeneration
Control of InterconnectedReheatThermal Systems Considering
GenerationRate
Constraints",IEEE Trans.,May 19g1, pAS_100;2334.
15' Nanda, J', M.L. Kothari and P.S. Satsangi,"Automatic
GenerationControl of an
InterconnectedHydro-thermalsystem in continuous and
DiscreteModes consid-
ering GenerationRateconstraints'IEE proc., prD, No. l,
January19g3,130 : 17.
16' IEEE committee Report,'DynamicModels for Steamand 9.1 INTRODUCTION
Hydro-turbinesin power
system studies" IEEE Trans., Nov/Dec. rg73, pAS-92, 1904.
l7' Hiyama, T', "Optimization of Discrete-typeLoad Frequency So far we have dealt with the steadystate behaviour of power system under
RegulatorsConsider-
ing Generation-Rateconstraints" proc.lE4 Nov. g2, r2g, pt normal operating conditions and its dynamic behaviour under small scale
c, 2g5.
I8. concordia,c., L.K. Kirchmayerand E.A. Szyonanski,..Effect perturbations.This chapter is devoted to abnormal system behaviour under
of speed Governor
Dead-bandon Tie Line Power and Frequency Control performan conditions of symmetrical short circuit (symmetricalthree-phase.fault*).Such
ce,, AIEE Trans.
A u g . 1 9 5 7 ,7 6 , 4 2 9 . conditions are caused in the system accidentally through insulation failure of
19' Nanda,J', M.L. Kothari and P.S. Satsangi,"Automatic equipment or flashover of lines initiated by a lightning stroke or through
Control of ReheatThermal
SystemConsidering GenerationRate Constraint and Covernor accidentalfaulty operation.The systemmust be protectedagainst flow of heavy
Dead-band,,.J.I.E.
(India),June 1983, 63,245. short circuit currents(which can causepeffnanentdamageto major equipment)
20. Tripathy, s.9., G.s. Hope and o.p. Marik, ,,optimisatiorr by disconnectingthe faulty part of the system by means of circuit breakers
of Load-frcqucncy
C<lntrolParametersfor Power systemswith ReheatSteamTurbines operated by protective relaying. For proper choice of circuit breakers and
and Governor
Dead-bandNonlinearity", proc. IEE, January rgg2, rzg, pt protective relaying, we must estimatethe magnitude of currents that would flow
c, No. r, r0.
21. Kothari,M.L., J. Nanda,D.p. Kothariand D. Das,.,Discrete-mode under short circuit conditions-this is the scope of fault analysis (study).
AGC of a.two_
area Reheat Thermal system with New Area control Error,,, The majority of systemfaults are not three-phasefaults but faults involving
IEEE Trans. on
Power System,Vol. 4, May 19g9, 730 one line to ground or occasionallytwo lines to ground.These are unsymmetrical
22' Daq D. J. Nanda, M.L. Kothari and D.p. Kothari, ,.AGC faults requiring special tools like symmetrical componentsand form the subject
of a Hydro_Thermal
systemwith New ACE consideringGRC", Int. J. EMps,1g, No. of study of the next two chapters.Though the symmetrical faults are rare, the
5, rggo, 46r.
23' Das, D', M.L' Kothari, D.P. Kothari and J. Nanda, "Variable symmetrical fault analysis must be carried out, as this type of fault generally
Structure Control
strategy to AGC of an IntcrconncctcdRcheat Thermal systcm,,, prctc. leads to most severe fault current flow against which the system must be
IEE, r3g,
p t D , 1 9 9 1 ,5 7 9 . protected. Symmetrical fault analysisis, of course, simpler to carry out.
24. Jalleli, Van Slycik et. al.. "lJndersfandingAutonnaticGeneration A power network comprises synchronousgenerators,ffansfonners, lines and
Control,,, IEEE
Trans.on P.S., Vol 07, 3 Aug. 92, 1106_1122. loads. Though the operating conditions at the time of fault are important, the
2 5 . Kothari, M.L., J. Nanda,D.p. Kothari and D. Das, ,,Discrete loads can be neglectedduring fault, as voltages dip very low so that currents
Mode AGC of a two
Area ReheatThermal Systemwith a NACE consideringGRC,,, drawn by loads can be neglectedin comparisonto fault currents.
J.LE. (rndia), vol.
72, Feb. 1992,pp Zg7-303.
2 6 . Bakken, B.H. and e.s. Grande,"AGC in a Deregulatedpower
system,,, IEEE *Symmetrical fault may be a solid three-phase short
Trans.on Power Systems,13, 4, Nov. 199g,pp. 1401_1406. circuit or may involve are
impedance.
5
Control of Voltage and
Reactive Power

5.1 Introduction

The approximate relationship between the magnitude of the voltage difference of

two nodes in a network and the flow of power was shown in Chapter 2 to be

RP þ XQ
DV
DVp ¼ ðfrom 2:15Þ
V

Also it was shown that the transmission angle d is proportional to

XP  RQ
d / DV q ¼ ðfrom 2:16Þ
V

Hence it may be seen that for networks where X  R, that is, most high voltage
power circuits, DV, the voltage difference, is determined mainly by Q while the
angle d is controlled by P.
Consider the simple system linking two generating stations A and B, as shown in
Figure 5.1(a). Initially the system is considered to be only reactive and R is ignored.
The machine at A is in phase advance of that at B and V1 is greater than V2; hence
there is a flow of real power from A to B. This can be seen from the phasor diagram
shown in Figure 5.1(b). It is seen that Id and hence P is determined by ffd and the
value of Iq and hence Q mainly, by V1  V 2 . In this case V 1 > V 2 and reactive power
is transferred from A to B. By varying the generator excitations such that V2 > V1 ,
the direction of the reactive power is reversed, as shown in Figure 5.1(c).
Hence, real power can be sent from A to B or B to A by suitably adjusting the
amount of steam (or water) admitted to the turbine, and reactive power can be sent

Electric Power Systems, Fifth Edition. B.M. Weedy, B.J. Cory, N. Jenkins, J.B. Ekanayake and G. Strbac.
Ó 2012 John Wiley & Sons, Ltd. Published 2012 by John Wiley & Sons, Ltd.
162 Electric Power Systems, Fifth Edition

P1 System P2
I
jX
Q1 Q2 V2
V1
IX

V1 V2 IqX IdX V2
A B IX V
1
I
δ δ
I
Id Iq
GA GB ω
Load Load ω

(a) (b) (c)

Figure 5.1 (a) System of two generators interconnected, (b) Phasor diagram when
V 1 > V 2 . Id and Iq are components of I. (c) Phasor diagram when V 2 > V 1

in either direction by adjusting the voltage magnitudes. These two operations are
approximately independent of each other if X  R, and the flow of reactive power
can be studied almost independently of the real power flow.
The phasor diagrams show that if a scalar voltage difference exists across a largely
reactive link, the reactive power flows towards the node of lower voltage. From
another point of view, if, in a network, there is a deficiency of reactive power at a
point, this has to be supplied from the connecting lines and hence the voltage at that
point falls. Conversely if there is a surplus of reactive power generated (for example,
lightly loaded cables generate positive VArs), then the voltage will rise. This is a
convenient way of expressing the effect of the power factor of the transferred cur-
rent, and although it may seem unfamiliar initially, the ability to think in terms of
VAr flows, instead of exclusively with power factors and phasor diagrams, will
make the study of power networks much easier.
If it can be arranged that Q2 in the system in Figure 5.1(a) is zero, then there will
be no voltage drop between A and B, a very satisfactory state of affairs.
Now assume that the interconnecting system shown in Figure 5.1(a) has some
resistance and that V1 is constant. Consider the effect of keeping V2, and hence the
voltage drop DV, constant. From equation (2.15)

V2 DV  RP2 R
Q2 ¼ ¼ K  P2 ð5:1Þ
X X

where K is a constant and R is the resistance of the system.

If this value of Q2 does not exist naturally in the circuit then it will have to be
obtained by artificial means, such as the connection at B of capacitors or inductors.
Control of Voltage and Reactive Power 163

If the value of the power changes from P2 to P02 and if V2 remains constant, then the
reactive power at B must change to Q02 such that

R
0 
Q02  Q2 ¼ P2  P2
X

that is, an increase in real power causes an increase in the reactive power needed to
maintain V2. The change, however, is proportional to (R/X), which is normally small.
It is seen that voltage can be controlled by the injection into the network of
reactive power of the correct sign. Other methods of a more obvious kind for con-
trolling voltage are the use of tap-changing transformers.

5.2 The Generation and Absorption of Reactive Power

5.2.1 Synchronous Generators

Synchronous generators can be used to generate or absorb reactive power. An
over-excited machine, that is, one with greater than nominal excitation, generates
reactive power whilst an under-excited machine absorbs it. Synchronous generators
are the main source of supply to the power system of both positive and negative
VArs.
The ability to generate or absorb reactive power is shown by the performance
chart of a synchronous generator. Reactive power generation (lagging power factor
operation) is limited by the maximum excitation voltage allowable before the rotor
currents lead to overheating. In Figure 3.12 this is 2.5 p.u.
The ability to absorb reactive power is determined by the short-circuit ratio
(1/synchronous reactance) as the distance between the power axis and the theoreti-
cal stability-limit line in Figure 3.12 is proportional to the short-circuit ratio. In mod-
ern machines the value of the short-circuit ratio is made low for economic reasons,
and hence the inherent ability to operate at leading power factors (absorbing VArs)
is not large. For example, a 200 MW 0.85 p.f. machine with a 10% stability allowance
has a capability of absorbing 45 MVAr at full power output. The VAr absorption
capacity can, however, be increased by the use of continuously acting voltage regu-
lators, as explained in Chapter 3.

5.2.2 Overhead Lines and Transformers

When fully loaded, overhead lines absorb reactive power. With a current I amperes
flowing in a line of reactance per phase X(V) the VArs absorbed are I2X per phase.
On light loads the shunt capacitances of longer lines may become dominant and
high voltage overhead lines then become VAr generators.
Transformers always absorb reactive power. A useful expression for the quantity
may be obtained for a transformer of reactance XT p.u. and a full load rating of
3V f I rated
164 Electric Power Systems, Fifth Edition

The ohmic reactance

Vf
XðVÞ ¼ XT
I rated

Therefore the VArs absorbed

Vf
¼ 3I 2 XT
I rated
V 2f
¼ 3I 2 XT
Vf I rated

2
3IV f
¼ XT
3I rated Vf
ðVA of loadÞ2
¼  XT
rated VA of transformer

5.2.3 Cables
Cables are generators of reactive power owing to their high shunt capacitance.
A 275 kV, 240 MVA cable produces 6.25–7.5 MVAr per km; a 132 kV cable roughly
1.9 MVAr per km; and a 33 kV cable, 0.125 MVAr per km.

5.2.4 Loads
A load at 0.95 power factor implies a reactive power demand of 0.33 kVAr per kW of
power, which is more appreciable than the mere quoting of the power factor would
suggest. In planning a network it is desirable to assess the reactive power require-
ments to ascertain whether the generators are able to operate at the required power
factors for the extremes of load to be expected. An example of this is shown in
Figure 5.2, where the reactive losses are added for each item until the generator
power factor is obtained.

Example 5.1
In the radial transmission system shown in Figure 5.2, all p.u. values are referred to the
voltage bases shown and 100 MVA. Determine the power factor at which the generator
must operate.
Solution
Voltage drops in the circuits will be neglected and the nominal voltages assumed.
Starting with the consumer load, the VArs for each section of the circuit are added in
turn to obtain the total
Busbar A,
P ¼ 0.5 p.u. Q ¼ 0
Control of Voltage and Reactive Power 165

160 km 48 km
E D C B A
0.1 p.u. 0.04 p.u. 0.1 p.u. 0.1 p.u.
G
Load 50 MW
0.1 p.u. 0.1 p.u. 0.1 p.u. p.f. = 1

275 kV 0.04 p.u. 132 kV 11 kV

200 MW
0.8 p.f. lagging

Figure 5.2 Radial transmission system with intermediate loads. Calculation of

reactive-power requirement

I2  loss in 132 kV lines and transformers

P2 þ Q2 0:52
¼ X CA ¼ 0:1
V2 12
¼ 0:025 p:u:

Busbar C,

P ¼ 2 þ 0:5 p:u: ¼ 2:5 p:u:

Q ¼ 1:5 þ 0:025 p:u: ¼ 1:525 p:u:

I2X loss in 275 kV lines and transformers

2:52 þ 1:5252
¼ 0:07
12
¼ 0:6 p:u:

If the I2X loss in the large generator-transformer is ignored, the generator must deliver
P ¼ 2.5 and Q ¼ 2.125 p.u. and operate at a power factor of 0.76 lagging.

5.3 Relation between Voltage, Power, and Reactive Power

at a Node

The voltage V at a node is a function of P and Q at that node, that is.

V ¼ f ðP; QÞ

The voltage also depends on that of adjacent nodes and the present treatment
assumes that these are infinite busbars.
166 Electric Power Systems, Fifth Edition

The total differential of V,

@V @V
dV ¼ dP þ dQ
@P @Q

and using

@P @V @Q @V
: ¼1 and : ¼1
@V @P @V @Q
ð5:2Þ
dP dQ
dV ¼ þ
ð@P= Þ ð@Q=@V Þ
@V

It can be seen from equation (5.2) that the change in voltage at a node is defined by
the two quantities
   
@P @Q
and
@V @V

As an example, consider a line with series impedance (R þ jX) and zero shunt admit-
tance as shown in Figure 5.3. From equation (2.15),

ðV1  VÞV  PR  XQ ¼ 0 ð5:3Þ

where V1, the sending-end voltage, is constant, and V, the receiving-end voltage,
depends on P and Q.
From equation (5.3)
@P V 1  2V
¼ ð5:4Þ
@V R
Also,
@Q V 1  2V
¼ ð5:5Þ
@V X

R + jX
P + jQ

V1 V Load

Figure 5.3 Single-phase equivalent circuit of a line supplying a load of P þ jQ from an

infinite busbar of voltage V1
Control of Voltage and Reactive Power 167

Hence,
dP dQ RdP þ XdQ
dV ¼ þ ¼ ð5:6Þ
@P/@V @Q/@V V 1  2V

For constant V and DV, RdP þ XdQ ¼ 0 and dQ ¼ (R/X)dP, which is obtainable
directly from Equation (5.1).
Normally, @Q=@V is the quantity of greater interest. It can be found experimen-
tally using a load-flow calculation (see Chapter 6) by the injection of a known quan-
tity of VArs at the node in question and calculating the difference in voltage
produced. From the results obtained,

DQ Qafter  Qbefore
¼
DV Vafter  V before

DV should be small for this test, a few per cent of the normal voltage, thereby giving
the sensitivity of the node to the VAr change.
From the expression,
@Q V1  2V
¼
@V X

proved for a single line, it is evident that the smaller the reactance associated with a
node, the larger the value of @Q=@V for a given voltage drop, that is, the voltage
drop is inherently small. The greater the number of lines meeting at a node, the
smaller the resultant reactance and the larger the value of @Q=@V. Obviously,
@Q=@V depends on the network configuration, but a high value would lie in the
range 10–15 MVAr/kV. If the natural voltage drop at a point without the artificial
injection of VArs is, say, 5 kV, and the value of @Q=@V at this point is 10 MVAr/kV,
then to maintain the voltage at its no-load level would require 50 MVAr. Obviously,
the greater the value of @Q=@V, the more expensive it becomes to maintain voltage
levels by injection of reactive power.

5.3.1 @Q=@V and the Short-Circuit Current at a Node

It has been shown that for a connecting circuit of reactance X with a sending-end
voltage V1 and a received voltage V
@Q V1  2V
¼
@V X

If the three-phases of the connector are now short-circuited at the receiving end
(i.e. a three-phase symmetrical short circuit applied), the current flowing in the lines

V1
I¼ assuming RX
X
168 Electric Power Systems, Fifth Edition

With the system on no-load

   
@Q V 1 
V ¼ V1 and  ¼ 
@V   X 

Hence the magnitude of @Q=@V is equal to the short-circuit current. With normal
operation, V is within a few per cent of V1 and hence the value of @Q=@V at V ¼ V1
gives useful information regarding reactive power/voltage characteristics for small
excursions from the nominal voltage. This relationship is especially useful as the
short-circuit current will normally be known at all substations.

Example 5.2
Three supply points A, B, and C are connected to a common busbar M. Supply point A
is maintained at a nominal 275 kV and is connected to M through a 275/132 kV trans-
former (0.1 p.u. reactance) and a 132 kV line of reactance 50 V. Supply point B is nomi-
nally at 132 kV and is connected to M through a 132 kV line of 50 V reactance. Supply
point C is nominally at 275 kV and is connected to M by a 275/132 kV transformer (0.1
p.u. reactance) and a 132 kV line of 50 V reactance.
If, at a particular system load, the line voltage of M falls below its nominal value by
5 kV, calculate the magnitude of the reactive volt-ampere injection required at M to
re-establish the original voltage.
The p.u. values are expressed on a 500 MVA base and resistance may be neglected
throughout.
Solution
The line diagram and equivalent single-phase circuit are shown in Figures 5.4 and 5.5.
It is necessary to determine the value of dQ/dV at the node or busbar M; hence the
current flowing into a three-phase short-circuit at M is required.

275 kV 132 kV
j 50Ω M j 50Ω

0.1 p.u. B
A j 50Ω

0.1 p.u.

C
275 kV

Figure 5.4 Schematic diagram of the system for Example 5.2

Control of Voltage and Reactive Power 169

j 0.1 j1.43 j1.43

j1.43

j 0.1

Figure 5.5 Equivalent single-phase network with the node M short-circuited to

neutral)

The base value of reactance in the 132 kV circuit assuming a 500 MVA system base is

1322
Zbase ¼ ¼ 35 V
500

Therefore the line reactances

j50
XL ¼ ¼ j1:43 p:u:
35

The equivalent reactance from M to N ¼ j0.5 p.u.

Hence the fault MVA at M

500
¼ ¼ 1000 MVA
0:5

and the fault current

1000  106
¼ pffiffiffi ¼ 4380 A
3  132  103

pffiffiffi
It has been shown that @QM / 3@V M ¼ three-phase short-circuit current when QM and
VM are three-phase and line values

@QM pffiffiffi
¼ 4380  3 ¼ 7:6 MVAr=kV
@VM

Assuming the natural voltage drop at M ¼ 5 kV.

170 Electric Power Systems, Fifth Edition

Therefore the value of the injected VArs required to offset this drop

¼ 7:6  5 ¼ 38 MVAr

An alternative approach is to consider:

The source impedance is 0.5 p.u. on a 132 kV, 500 MVA base
A 5 kV voltage drop is 5/132 ¼ 0.038 p.u.
The current flow for this volt drop is

0:038/0:5 ¼ 0:076 p:u:

At close to 1 p.u. voltage this is also the reactive power flow.

Q ¼ 0:076 p:u:
¼ 0:076  500 ¼ 38 MVAr

5.4 Methods of Voltage Control: (a) Injection of Reactive Power

In transmission systems with X  R, busbar voltages can be controlled by the injec-

tion or absorption of reactive power. However, controlling network voltage through
reactive power flow is less effective in distribution networks where the higher cir-
cuit resistances lead to the reactive power flows having less effect on voltage and
causing an increase in real power losses.
Although reactive power does no real work, it does lead to an increase in the
magnitude of current in the networks and hence real power losses. Electricity
suppliers often penalize loads with a poor power factor by applying charges
based on kVAh (or even kVArh) in addition to kWh or even basing part of the
charge on peak kVA drawn. The provision of static capacitors to improve the
power factors of factory loads has been long established. The capacitance
required for the power-factor improvement of loads for optimum economy is
determined as follows.
Let the tariff of a consumer be based on both kVA and kWh

charge ¼ $A  kVA þ $B  kWh

A load of P kilowatts at power factor w1, lagging has a kVA of P/cosw1. If this power
factor is improved to cos w2, the new kVA is P/cos w2. The saving is therefore
 
1 1
saving ¼ $PA 
cos w1 cos w2
Control of Voltage and Reactive Power 171

The reactive power required from the correcting capacitors

Pðtan w1  tan w2 Þ kVAr

Let the cost per annum in interest and depreciation on the capacitor installation be
$C per kVAr or
$CPðtan w1  tan w2 Þ

The net saving

  
1 1
¼ P $A   $Cðtan w1  tan w2 Þ
cos w1 cos w2

This saving is a maximum when

     
d saving sin w2 1
¼ P $A  þ $C ¼0
dw2 cos2 w2 cos2 w2

that is when sin w2 ¼ $C/$A.

It is interesting to note that the optimum power factor is independent of the origi-
nal one. The improvement of load power factors in such a manner will help to allevi-
ate the whole problem of VAr flow in the distribution system.
The main effect of transmitting power at non-unity power factors is to increase
losses and reduce the ability of the circuits to transport active power. Thus both
operating and capital costs are increased by low power factor. It is evident from
equation (2.15) that, for circuits with a significant X/R ratio, the voltage drop is
largely determined by the reactive power Q. At non-unity power factors the line
currents are larger, giving increased I2R losses and hence reduced thermal capa-
bility. One of the obvious places for the artificial injection of reactive power is at
the loads themselves.
In general, four methods of injecting reactive power are available, involving the
use of:

1. static shunt capacitors;

2. static series capacitors;
3. synchronous compensators;
4. static VAr compensators and STATCOMs.

5.4.1 Shunt Capacitors and Reactors

Shunt capacitors are used to compensate lagging power factor loads, whereas
reactors are used on circuits that generate VArs such as lightly loaded cables. The
effect of these shunt devices is to supply or absorb the requisite reactive power to
maintain the magnitude of the voltage. Capacitors are connected either directly to a
busbar or to the tertiary winding of a main transformer. In the USA they are often
172 Electric Power Systems, Fifth Edition

installed along the routes of distribution circuits to minimize the losses and voltage
drops. Unfortunately, as the voltage reduces, the VArs produced by a shunt capaci-
tor or absorbed by a reactor fall as the square of the voltage; thus, when needed
most, their effectiveness drops. Also, with light network load when the voltage is
high, the capacitor output is large and the voltage tends to rise to excessive levels,
requiring some capacitors or cable circuits to be switched out by local overvoltage
relays.

5.4.2 Series Capacitors

Capacitors can be connected in series with overhead lines and are then used to
reduce the inductive reactance between the supply point and the load. One major
drawback is the high overvoltage produced across the capacitor when a short-circuit
current flows through the circuit, and special protective devices need to be incorpo-
rated (e.g. spark gaps) and non-linear resistors. The phasor diagram for a line with a
series capacitor is shown in Figure 5.6(b).
The relative merits between shunt and series capacitors may be summarized as
follows:

1. If the load VAr requirement is small, series capacitors are of little use.
2. With series capacitors the reduction in line current is small; hence if thermal con-
siderations limit the current, little advantage is obtained and shunt compensation
should be used.
3. If voltage drop is the limiting factor, series capacitors are effective; also, voltage
fluctuations due to arc furnaces, and so on, are evened out.

XL XC
I VR
VS

IXC I _Φ
IXL
(a)
VS(without capacitor)
X L= ω L
X C= 1 IXC
ωC
VS (with capacitor)
I(XL - XC)

Φ VR
I
ω
VC (b)

Figure 5.6 (a) Line with series capacitor, (b) Phasor diagram for fixed VR
Control of Voltage and Reactive Power 173

4. If the total line reactance is high, series capacitors are very effective in reducing
voltage drops and stability is improved.

Both shunt and series capacitors need to be applied with care as they can both
lead to resonance with the inductive reactance of the power system. Shunt
capacitors are benign as long as their network is connected to the main power
system and the voltage is controlled. However, if a section of network containing
both shunt capacitors and induction generators is isolated then self-excitation of
the induction generators can lead to very high resonant voltages. The use of series
capacitors, although very effective in reducing voltage drop on heavily loaded
circuits, can lead to sub-synchronous resonance with rotating machines. Capacitors
are not commonly used in distribution systems in the UK, partly because of
concerns over resonance.

5.4.3 Synchronous Compensators

A synchronous compensator is a synchronous motor running without a mechanical
load and, depending on the value of excitation, it can absorb or generate reactive
power. As the losses are considerable compared with static capacitors, the power
factor is not zero. When used with a voltage regulator the compensator can automat-
ically run overexcited at times of high load and underexcited at light load. A typical

275 kV

415 V supply
To AVR
(via VT)
CB Earthing
transformer To AVR
11 kV

CT Synchronous
compensator

66 kV

Figure 5.7 Typical installation with synchronous compensator connected to tertiary

(delta) winding of main transformer. A neutral point is provided by the earthing trans-
former shown. The Automatic Voltage Regulator (AVR) on the compensator is con-
trolled by a combination of the voltage on the 275 kV system and the current output;
this gives a droop to the voltage-VAr output curve which may be varied as required
174 Electric Power Systems, Fifth Edition

1.07

1.05

Voltage p.u.
5% Droop

1.0

0.95
-20 0 +40
Absorb Q MVAr Generate

Figure 5.8 Voltage-reactive power output of a typical 40 MVAr synchronous

compensator

connection of a synchronous compensator is shown in Figure 5.7 and the associated

Volt-VAr output characteristic in Figure 5.8. The compensator is run up as an induc-
tion motor in 2.5 min and then synchronized.
A great advantage is the flexibility of operation for all load conditions. Although
the cost of such installations is high, in some circumstances it is justified, for exam-
ple at the receiving-end busbar of a long high-voltage line where transmission at
power factors less than unity cannot be tolerated. Being a rotating machine, its
stored energy is useful for increasing the inertia of the power system and for riding
through transient disturbances, including voltage sags.

5.4.4 Static VAr Compensators (SVCs) and STATCOMs

Synchronous compensators are rotating machines and so are expensive and have
mechanical losses. Hence they are being superseded increasingly by power elec-
tronic compensators: SVCs and STATCOMs.
SVCs use shunt connected reactors and capacitors controlled by thyristors. The
reactive power is provided by the shunt elements (capacitors and inductors), as dis-
cussed in Section 5.4.1 but these are controlled by thyristors. The output of the
Control of Voltage and Reactive Power 175

TCR / MSR MSC / TSC TCR / TSC TCR / MSC

Figure 5.9 Possible combinations of controlled reactors and capacitors forming an

SVC. TCR: Thyristor Controlled Reactor, MSR: Mechanically Switched Reactor, MSC:
Mechanically Switched Capacitor, TSC: Thyristor Switched Capacitor

Thyristor Controlled Reactor (TCR) is controlled by delaying the switching on of the

thyristor within the 50/60 Hz cycle. The thyristor switches off when the current
drops to zero. The firing angle of the thyristor can be varied within each cycle and
hence the VAr absorption by the TCR controlled. As shown in Figure 5.9, TCRs may
be used with Mechanically or Thyristor Switched Capacitors to create an SVC to
export and import VArs.
When a capacitor is connected to a strong voltage source, very large currents can
flow. Hence Thyristor Switched Capacitors are only operated in integral cycles and
the operation of the thyristors is timed so that they switch when there is no instanta-
neous voltage across the capacitor.
A STATCOM (Static Compensator) is also a power electronic device to provide
reactive power but it operates on a different principle (Figure 5.10). A STATCOM
consists of a Voltage Source Converter (VSC) connected to the power system
through a coupling reactance (L). The VSC uses very large transistors that can be
turned on and off to synthesize a voltage sine wave of any magnitude and phase.
VSTATCOM is a 50/60 Hz sine wave kept in phase with Vterminal (Figure 5.10). If
the magnitude of VSTATCOM is greater than that of Vterminal then reactive power is
generated by the STATCOM while if the magnitude of VSTATCOM is less than that of
Vterminal then reactive power is absorbed by the STATCOM. A very small phase
angle is introduced between VSTATCOM and Vterminal so that a small amount of real
power flows into the STATCOM to charge the DC capacitor and provide for the
losses of the converter. However, the principle of operation is that the reactive
power is provided by the interaction of the two voltage magnitudes across the
reactor. The DC capacitor is only used to operate the power electronics and control
176 Electric Power Systems, Fifth Edition

Vterminal

ISTATCOM (Generating)Q
ISTATCOM
L
VSTATCOM_2
VSTATCOM
Vterminal
VSTATCOM_1
VSC

ISTATCOM
(Absorbing) Q

VDC

(a) (b)

Figure 5.10 Operation of a STATCOM

the ripple current. STATCOMs can be controlled very fast and have a smaller
physical equipment footprint than SVCs.

5.5 Methods of Voltage Control: (b) Tap-Changing Transformers

The basic operation of the tap-changing transformer has been discussed in Chapter
3. By changing the transformation ratio, the voltage in the secondary circuit is
varied. Hence voltage and reactive power control is obtained.
In distribution circuits, tap-changing transformers are the primary method of
voltage control. In a distribution transformer, the tap-changer compensates for the
voltage drop across the reactance of the transformer but also for the variations in
the voltage applied to the primary winding caused by changes of load within the
high voltage network. In transmission circuits reactive power is dispatched by alter-
ing the taps of transformers and this, in turn, controls the network voltages.

5.5.1 Use of Tap-Changing Transformers to Control Voltage

in a Distribution System
Consider the 40 MVA 132/11 kV transformer with a reactance of 13% on its rating
shown in Figure 5.11. It is equipped with an on-load tap-changer that is used to
maintain constant voltage at an 11 kV busbar by compensating for variations in the
voltage of the 132 kV network and for the voltage drop across the transformer. The
variation of network voltage at the 132 kV transformer busbar for heavy and light
loading conditions, and the loads of the transformer are given in Table 5.1. Active
power losses in the transformer are ignored and it is assumed that the value of the
reactance of the transformer is not influenced by the change in the turns ratio.
Control of Voltage and Reactive Power 177

t:1

X
Vr
Vs

Figure 5.11 Tap changing transformer in a distribution circuit

t is the fraction of the nominal transformation ratios, that is the tap ratio/nominal
ratio. For example, a transformer of nominal ratio 132 to 11 kV when tapped to give
144 to 11 kV has a t of 144/132 ¼ 1.09.
Choosing SBASE of 40 MVA and VBASES of 132 kV and 11 kV.

Under heavy loading conditions,

Vr ¼ 11 kV, 1 p.u.
Vs ¼ 120 kV, 0.909 p.u.
Q ¼ 13.94 MVAr, 0.3485 p.u.

Q 0:3485
DV ¼ X¼  0:13 ¼ 0:05 p:u:
VS 0:909

V r t ¼ ðV s  DV Þ
ðV s  DV Þ 0:909  0:05
t¼ ¼ ¼ 0:86
Vr 1

Under light loading conditions,

Vr ¼ 11 kV, 1 p.u.
Vs ¼ 145 kV, 1.1 p.u.
Q ¼ 2.11 MVAr, 0.053 p.u.

Table 5.1 Loading of transformer

Load Power Voltage Desired Voltage

Factor at 132 kV at 11 kV
of Load Busbar Vs Busbar Vr

Heavy loading conditions 32 MVA 0.90 120 kV 11 kV

Light loading conditions 4 MVA 0.85 145 kV 11 kV
178 Electric Power Systems, Fifth Edition

Q 0:053
DV ¼ X¼  0:13 ¼ 0:006 p:u:
VS 1:1
V r t ¼ ðV s  DVÞ
ðV s  DVÞ 1:1  0:006
t ¼ ¼ ¼ 1:094
Vr 1

A radial distribution circuit with two tap-changing transformers, is shown in the

equivalent single-phase circuit of Figure 5.12. V1 and V2 are the nominal voltages; at

R +jX I
Line
Is IR –Φ or (P + jQ)

V1 Vs Vr V2 Load

1 : ts tr : 1
(a)

X
Vs
Vr
t :1

(b)

X 2 P
t
Vs
t Vr
Q

R + jX
t 2r

t V2 (c)
V1( t s )
r

Figure 5.12 (a) Coordination of two tap-changing transformers in a radial transmis-

sion link (b) and (c) Equivalent circuits for dealing with off-nominal tap ratio, (b) Single
transformer, (c) Two transformers
Control of Voltage and Reactive Power 179

the ends of the circuit the actual voltages are tsV1 and trV2. It is required to deter-
mine the tap-changing ratios needed to compensate completely for the voltage drop
in the line. The product tstr will be made unity; this ensures that the overall voltage
level remains in the same order and that the minimum range of taps on both trans-
formers is used.
(Note that all values are in per unit; t is the off-nominal tap ratio.)
Transfer all quantities to the load circuit.
The line impedance becomes ðR þ jXÞ=t2r ; V s ¼ V 1 ts and, as the impedance has
been transferred V r ¼ V1 ts . The input voltage to the load circuit becomes V1 ts =tr
and the equivalent circuit is as shown in Figure 5.10(c). The arithmetic voltage drop

ts RP þ XQ
¼ V1  V2

tr t2r V 2
When tr ¼ 1/ts

t2s V 1 V2  V 22 ¼ ðRP þ XQÞt2s

And 

1
V 2 ¼ 1/2 t2s V1 ts t2s V21  4ðRP þ XQÞ 2 ð5:7Þ

If ts is specified then tr is defined. There are then two values of V2 for a given V1,
one low current, high voltage and one high current and low voltage. Only the high
voltage, low current solution is useful in a power system.

Example 5.3
A 132 kV line is fed through an 11/132 kV transformer from a constant 11 kV supply. At
the load end of the line the voltage is reduced by another transformer of nominal ratio
132/11 kV. The total impedance of the line and transformers at 132 kV is (25 þ j66) V.
Both transformers are equipped with tap-changing facilities which are arranged so that
the product of the two off-nominal settings is unity. If the load on the system is 100 MW
at 0.9 p.f. lagging, calculate the settings of the tap-changers required to maintain the
voltage of the load busbar at 11 kV. Use a base of 100 MVA.
Solution
The line diagram is shown in Figure 5.13. As the line voltage drop is to be completely
compensated, V1 ¼ V2 ¼ 132 kV ¼ 1 p.u. Also, ts  tr ¼ 1. The load is 100 MW,
48.3 MVAr., that is, 1 þ j0.483 p.u.
Using equation (5.7)


1
1 ¼ 1/2 t2S 1 tS t2S 1  4ð0:14  1 þ 0:38  0:48Þ 2

1
;2 ¼ t2S tS t2S  1:28 2

2

; 2  t2S ¼ t2S t2S  1:28
180 Electric Power Systems, Fifth Edition

V1 = 132kV V2 = 132kV
Vr
VS Z
1 : tS tr : 1 P, Q

Figure 5.13 Schematic diagram of system for Example 5.3

Hence,

ts ¼ 1:21 and tr ¼ 1/1:21 ¼ 0:83

These settings are large for the normal range of tap-changing transformers (usually not
more than 20% tap range). It would be necessary, in this system, to inject VArs at the
load end of the line to maintain the voltage at the required value.

A transformer at the receiving end of a line does not improve the VAr flow in the
circuit and the current in the supplying line is increased if the ratio is reduced. In
countries with long and inadequate distribution circuits, it is often the practice to
boost the received voltage by a variable ratio transformer so as to maintain rated
voltage as the power required increases. Unfortunately, this has the effect of increas-
ing the primary supply circuit current by the transformer ratio, thereby decreasing
the primary voltage still further until voltage collapse occurs.

5.5.2 Use of Tap-Changing Transformers to Despatch VArs

in a Transmission System
In transmission networks VArs may be dispatched by the adjustment of tap settings
on transformers connecting busbars. Consider the situation in Figure 5.14(a), in
which Vs and Vr are constant voltages representing the two connected systems. The
circuit may be rearranged as shown in Figure 5.14(b), where t is the off-nominal (per
unit) tap setting; resistance is zero. The voltage drop between busbars
 
Vs X QT
DV ¼  Vr ¼ 
t t2 V r

Hence,


V s V r t  V 2r t2 1/X ¼ QT
Control of Voltage and Reactive Power 181

Vs t:1 Vr
X

Infinite busbars
(a)

X / t2

Vs QT Vr
Vs / t
(b)

Figure 5.14 (a) Two power systems connected via a tap-change transformer,
(b) Equivalent circuit with impedance transferred to receiver side

and
tð1  tÞV2 /X ¼ QT when Vs ¼ Vr ¼ V ð5:8Þ

When

t < 1, QT is positive, that is a flow of VArs into Vr

t > 1, QT is negative, a flow of VArs out of Vr

Also, QT ¼ t(1  t)S, where S ¼ short-circuit level, that is V2/X.

Thus, by suitable adjustment of the tap setting, an appropriate injection of reactive
power is obtained.
The idea can be extended to two transformers in parallel between networks. If one
transformer is set to an off-nominal ratio of, say, 1.1 : 1 and the other to 0.8 : 1 (i.e. in
opposite directions), then a circulation of reactive power occurs round the loop,
resulting in a net absorption of VArs. This is known as ‘tap stagger’ and is a compar-
atively inexpensive method of VAr absorption.

Example 5.4
A synchronous generator (75 MVA, 0.8 p.f., 11.8 kV and XS ¼ 1.1 p.u.) is connected
through an 11/275 kV tap changing transformer (75 MVA, XT ¼ 0.15 p.u., tap range ¼
þ/20%) to a very large 275 kV power system, as shown in Figure 5.15.
a. What is the value of the internal emf and power angle of the generator when it
exports 60 MW of active and zero MVAr of reactive power to the system? With the
182 Electric Power Systems, Fifth Edition

V t:1
E

XS XT
275kV

Figure 5.15 Tap changer in a transmission circuit – Example 5.4

transformer tap in the neutral position, what is the value of reactive power output at
the generator terminals?
b. What is the value of the transformer tap at which 20 MVAr is imported from the
275 kV system, if the terminal voltage of the generator V is maintained at 1 p.u.
when the generator does not export any active power.
Solution
Choosing a common base
SBASE ¼ 75 MVA
VBASES ¼ 11 kV; 275 kV

Converting to the common base (Equation (2.7))

 2
11:8
XS ¼ 1:1  ¼ 1:26 p:u:
11
XT ¼ 0:15 p:u:
X ¼ XS þ XT ¼ 1:41 p:u:

At 60 MW and 0 MVAr exported into the 275 kV system,

I ¼ 0:8 þ j0 p:u:

and the generator internal voltage and power angle are

E ¼ 1 þ IX ¼ 1 þ 0:8  j 1:41 ¼ 1 þ j1:13

E ¼ 1:5ff48

With the transformer tap in the neutral position the reactive power output at the
generator terminals is

V ¼ 1 þ IX ¼ 1 þ 0:8  j0:15 ¼ 1 þ j0:12

S ¼ VI ¼ ð1 þ j0:12Þ0:8 ¼ 0:8 þ j0:096 p:u:
Q ¼ 0:096  75 ¼ 7:2 MVAr
Control of Voltage and Reactive Power 183

If 20 MVAr is absorbed from the 275 kV system and V and the voltage of the 275 kV
system are at 1 p.u. then using Equation (5.8)

tð1  tÞV2 /XT ¼ QT

20
tð1  tÞ1/0:15 ¼
75
tð1  tÞ ¼ 0:04

t2  t  0:04 ¼ 0
t ¼ 1:04

The solution for t near 1 is chosen

Thus the transformer taps are set to þ4%

5.6 Combined Use of Tap-Changing Transformers and

Reactive-Power Injection

A common practical arrangement is shown in Figure 5.16, where the tertiary wind-
ing of a three-winding transformer is connected to a VAr compensator. For given
load conditions it is proposed to determine the necessary transformation ratios with
certain outputs of the compensator.
The transformer is represented by the equivalent star connection and any line
impedance from V1 or V2 to the transformer can be lumped together with the trans-
former branch impedances. Here, VN is the phase voltage at the star point of the
equivalent circuit. The secondary impedance (XS) is usually approaching zero and
hence is neglected. Resistance and losses are ignored.

P2 Q2
V1 V2 V1 T12 V2
P S XP VL XS

T
V3
XT

Compensator C
P3= 0 T23
(a)
Q3
V3
C
(b)

Figure 5.16 (a) Schematic diagram with combined tap-changing and synchronous
compensation, (b) Equivalent network
184 Electric Power Systems, Fifth Edition

The allowable ranges of voltage for V1 and V2 are specified and the values of the
three-phase real and reactive power; P2, Q2, P3, and Q3 are given. P3 is usually taken
as zero.
The volt drop V1 to VL is given by

Q2 /3
DV p
Xp
VN

or

Q2
DV p
Xp pffiffiffi
VL 3
pffiffiffi
where V L is the line voltage ¼ 3V N and Q2 is the total VArs.
Also,

P2
pffiffiffi
DV q
Xp
VL 3

2
2
; VN þ DVp þ DV q ¼ V21

(see phasor diagram of Figure 2.24; phase values used)

and
   2
VL Q2 2 P2
pffiffiffi þ Xp pffiffiffi þ Xp pffiffiffi
2
¼ V 21
3 VL 3 3V L

2
; V2L þ Xp Q2 þ X2p P22 ¼ V21L V 2L
pffiffiffi
where V 1L is the line voltage ¼ 3V 1
rhffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiiffi
V2  2Xp Q2 1

;V2L ¼ 1L V21L V21L  4Xp Q2  4X2p P22
2 2

Once VL is obtained, the transformation ratio is easily found. The procedure is

best illustrated by an example.

Example 5.5
A three-winding grid transformer has windings rated as follows: 132 kV (line), 75 MVA,
star connected; 33 kV (line), 60 MVA, star connected; 11 kV (line), 45 MVA, delta con-
nected. A VAr compensator is available for connection to the 11 kV winding.
The equivalent circuit of the transformer may be expressed in the form of three wind-
ings, star connected, with an equivalent 132 kV primary reactance of 0.12 p.u., negligi-
ble secondary reactance, and an 11 kV tertiary reactance of 0.08 p.u. (both values
expressed on a 75 MVA base).
Control of Voltage and Reactive Power 185

60 MW T12 0 T12
V1 N V2 V1 N V2

120 kV 34 kV 143 kV 30 kV
30 MVAr
0.8 p.u.

T23 T23

C C
(a) (b)

Figure 5.17 Systems for Example 5.4. (a) System with loading condition 1.
(b) System with loading condition 2

In operation, the transformer must deal with the following extremes of loading:

1. Load of 60 MW, 30 MVAr with primary and secondary voltages governed by the lim-
its 120 kV and 34 kV; compensator disconnected.
2. No load, primary and secondary voltage limits 143 kV and 30 kV; compensator in
operation and absorbing 20 MVAr.
Calculate the range of tap-changing required. Ignore all losses.
Solution
The value of XP, the primary reactance (in ohms)

1322
¼ 0:12  ¼ 27:8 V
75

Similarly, the effective reactance of the tertiary winding is 18.5 V. The equivalent star
circuit is shown in Figure 5.17.
The first operating condition is as follows:

P1 ¼ 60 MW Q1 ¼ 30 MVAr V1L ¼ 120 kV

Hence,

1

V 2L ¼ 120 0002  2  27:8  30  106
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi


1
120 0002 120 0002  4  27:8  30  106  4  27:82  602  1012
2
 
122
¼ 63:6 108  124:4  108
2
;VL ¼ 111 kV
186 Electric Power Systems, Fifth Edition

The second set of conditions are:

V 1L ¼ 143 kV P2 ¼ 0 Q2 ¼ 20 MVAr

Again, using the formula for VL,

VL ¼ 138:5 kV

The transformation ratio under the first condition

¼ 111/34 ¼ 3:27

and, for the second condition

¼ 138:5/30 ¼ 4:64

The actual ratio will be taken as the mean of these extremes, that is, 3.94, varying by
0.67 or 3.94 17%. Hence the range of tap-changing required is 17%.

Example 5.6
In the system shown by the line diagram in Figure 5.18, each of transformers TA
and TB have tap ranges of 10% in 10 steps of 1.0%. It is necessary to find the
voltage boost needed on transformer TA to share the power flow equally between
the two lines.
The system data is as follows (on a common base):
All transformers: XT ¼ 0.1 p.u.
Transmission lines: R ¼ R0 ¼ 0
X ¼ 0.20 p.u.
X0 ¼ 0.15 p.u
VA ¼ 1:1ff5
VB ¼ 1:0ff0
Solution
We must first calculate the current sharing in the two parallel lines:

1:1ff5  1:0ff0
I1 ¼ ¼ 0:2397  j0:2397
j0:4
1:1ff5  1:0ff0
I2 ¼ ¼ 0:2740  j0:2740
j0:35

Any boost by transformer TA will cause a current to circulate between the two busbars
because the voltages VA and VB are assumed to be held constant by the voltage regula-
tors on the generators.
Control of Voltage and Reactive Power 187

VA 33 : 132kV 132 : 33kV VB

TA I1 TB
R + jX
EA EB
I2
R + jX
TC TD
(a)

i 0.1 j0.2 i 0.1

Vboost
Icirc
i 0.1 j 0.15 i 0.1
(b)

Figure 5.18 (a) Line diagram of system for Example 5.6. (b) Equivalent network
with voltage boost Vboost acting

To equalize the currents, a circulating current is required, as in Figure 5.18(b),

giving

I2  I 1
I circ ¼
2

0:0343  j0:0343
I circ ¼ ¼ 0:0241ff  45
2

;VBoost ¼ 0:0241ff  45  j0:75 ¼ 0:0180ff45 V

To achieve this boost, ideally TA should be equipped with a phase changer of 45 and
taps to give 1.8% boost. In practice, a tap of 2% would be used in either an in-phase
boost (such as obtainable from a normal tapped transformer) or a quadrature boost
(obtainable from a phase-shift transformer. In transmission networks it should be noted
that because of the generally high X/R ratio, an in-phase boost gives rise to a quadra-
tive current whereas a quadrature boost produces an in-phase circulating current,
thereby adding to or subtracting from the real power flow.
Circulating reactive current by adjusting the taps in transformers in parallel circuits
has been used to de-ice lines in winter by producing extra I2R losses for heating. Two
transformers in parallel can be tap-staggered to produce I2X absorption under light-
load, high-voltage conditions.
188 Electric Power Systems, Fifth Edition

R
Y (a)
1
VYB VYB
B

VB

VR
VYB
θ 1
VYB (b)
1
VR

VY

Figure 5.19 (a) Connections for one phase of a phase shift transformer. Similar con-
nections to other two phases. (b) Corresponding phasor diagram

5.7 Phase-Shift Transformer

A quadrature phase shift can be achieved by the connections shown in Figure 5.19(a).
The booster arrangement shows the injection of voltage into one phase only; it is
repeated for the other two phases. In Figure 5.19(b), the corresponding phasor
diagram is shown and the nature of the angular shift of the voltage boost VYB
indicated. By the use of tappings on the energizing transformer, several values of
phase shift may be obtained.

Example 5.7
In the system shown in Figure 5.20, it is required to keep the 11 kV busbar at constant
voltage. The range of taps is not sufficient and it is proposed to use shunt capacitors
connected to the tertiary winding.
All impedances are referred to 33 kV. The impedance of the overhead line, ZL
referred to 33 kV ¼ (2.2 þ j5.22) V. For the three-winding transformer the measured
impedances between the windings and the resulting equivalent star impedances Z1, Z2
and Z3 are given in Table 5.2.
Solution
The equivalent circuit referred to 33 kV is shown in Figure 5.20(b).
The voltage at point C (referred to 33 kV) is

33 000
pffiffiffi  DVp
3
 Control of Voltage and Reactive Power
Table 5.2 Data for three-winding transformer

Winding MVA Voltage p.u. Z p.u. Z Z(V) referred Equivalent

MVA kV referred to referred to 33 kV side Z(V) referred
nameplate VA to 15 MVA (ZBASE ¼ 72.6 V) to 33 kV side.
(Equation (3.14))

P-S 15 33/11 0.008 þ j0.1 0.008 þ j0.1 0.58 þ j7.26 Z1 ¼ 0.212 þ j8.21
1
ðZPS þ ZPT  ZST Þ
2
P-T 5 33/1.5 0.0035 þ j0.0595 0.0105 þ j.1785 0.76 þ j12.96 Z2 ¼ 0.368  j0.945
1
ðZPS þ ZST  ZPT Þ
2
S-T 5 11/1.5 0.0042 þ j0.0175 0.0126 þ j0.0525 0.915 þ j3.81 Z3 ¼ 0.547 þ j4.76
1
ðZPT þ ZST  ZPS Þ
2

189
190 Electric Power Systems, Fifth Edition

Tap range
33 kV (constant) +
- 10%
11 kV
Load 10 MVA 0.8 p.f. lagging

A B C
(a)
0.368 - j0.945Ω
C
2.2 + j5.22Ω 0.212 + j8.21Ω Z2
A N
ZL Z1 0.547 + j4.76Ω
B
33 kV Z3
10 MVA
Tertiary L L 0.8 p.f. lagging
load load

(b)

Figure 5.20 (a) Line diagram for Example 5.7. (b) Equivalent network – referred
to 33 kV

where
RP þ XQ
DVp

V
2:78  8=3  106 þ 12:485  6=3  106
;DVp
pffiffiffi
33 000= 3
7:4 þ 24:95
DV p ¼ ¼ 1:703 kV and VC ¼ 17:3 kV
19

Vc referred to 11 kV ¼ 17.3/3 ¼ 5.77 kV (phase) or 10 kV (line). In order to maintain

11 kV at C, the voltage is raised by tapping down on the transformer. Using the full
range of 10%, that is t ¼ 0.9, the voltage at C is

pffiffiffi
17:3  3
¼ 11 kV
ðð33  0:9Þ=11Þ

The true voltage will be less than this as the primary current will have increased by (1/
0.9) because of the change in transformer ratio. The tap-changing transformer is not
able to maintain 11 kV at C and the use of a static capacitor connected to the tertiary
will be investigated.
Consider a shunt capacitor of capacity 5 MVAr (the capacity of the tertiary).
Assume the transformer to be at its nominal ratio 33/11 kV. The voltage drop to
point N
Control of Voltage and Reactive Power 191

2:412  8=3  106 þ 13:42  1=3  106

¼
VN
19 kV
¼ 0:574 kV

V N ¼ 19  0:574 ¼ 18:43 kV

Therefore the volt drop N to C

0:368  8=3  106  0:945  6=3  106

DVC ¼ ¼ 0:049 kV
18:43  103
;VC ¼ 18:43  0:049 ¼ 18:381 kV

Referred to 11 kV, VC ¼ 10.65 kV (line). Hence, to have 11 kV the transformer will tap
such that t ¼ (1  0.35/11) ¼ 0.97, that is, a 3% tap change, which is well within the
range and leaves room for load increases.
On no-load

2:959  0 þ 18:19  ð5=3Þ

DV p ¼ kV
19
30:3
¼ ¼ 1:594 kV
19
VC ¼ 19 þ 1:6 ¼ 20:6 kV ðphaseÞ

On the 11 kV side
VC ¼ 11:9 kV ðlineÞ

therefore the tap change will have to be at least 8.1%, which is well within the range.

5.8 Voltage Collapse

Voltage collapse is an important aspect of system stability.

Consider the circuit shown in Figure 5.21(a). If VS is fixed (i.e. an infinite busbar),
the graph of VR against P for given power factors is as shown in Figure 5.21(b). In
Figure 5.21(b), Z represents the series impedance of a 160 km long, double-circuit,
400 kV, 260 mm2 conductor overhead line. The fact that two values of voltage exist
for each value of power is easily understood by considering the analytical solution of
this circuit. At the lower voltage a very high current is taken to produce the power.
The seasonal thermal ratings of the line are also shown, and it is apparent that for
loads of power factor less than unity (lagging) the possibility exists that, before the
thermal rating is reached, the operating power may be on that part of the character-
istic where small changes in load cause large voltage changes and voltage instability
will have occurred. In this condition the action of tap-changing transformers is
192 Electric Power Systems, Fifth Edition

Z P, Q
VS VR

(a)

Hot weather Normal Cold weather

rating rating rating
1.2
p.f. = 0.90 lead
1.1 p.f. = 0.95
lead
p.f. =
1.0 0
p.f. = .97 lead
p.f. 0 .99 le
p
0.9 p.f p.f .f. = 0 =1
.0
ad
.= .= .
0.9 0.9 99 log
0.8 7l
p.

0l
VR (p.u.)

og
f.

og
=
0.

0.7
95
lo
g

0.6
0.5
0.4
0.3
0.2
0.1
500 1000 1500 2000
Power (MW)
(b)

Figure 5.21 (a) Equivalent circuit of a line supplying a load P þ jQ. (b) Relation
between load voltages and received power at constant power factor for a 400 kV,
2  260 mm2 conductor line, 160 km in length. Thermal ratings of the line are
indicated

interesting. If the receiving-end transformers ‘tap up’ to maintain the load voltage,
the line current increases, thereby causing further increase in the voltage drop. It
would, in fact, be more profitable to ‘tap down’, thereby reducing the current and
voltage drop. It is feasible therefore for a ‘tapping-down’ operation to result in
increased secondary voltage, and vice versa.
The possibility of an actual voltage collapse depends upon the nature of the load. If
this is stiff (constant power), for example induction motors, the collapse is aggravated.
If the load is soft, for example heating, the power falls off rapidly with voltage and the
situation is alleviated. Referring to Figure 5.21 it is evident that a critical quantity is the
power factor; at full load a change in lagging power factor from 0.99 to 0.90 will pre-
cipitate voltage collapse. On long lines, therefore, for reasonable power transfers it is
necessary to keep the power factor of transmission approaching unity, certainly above
Control of Voltage and Reactive Power 193

Infinite busbar

C C C
2 2 2

X X X

C C C
2 2 2

Load Zg GL

Figure 5.22 Line diagram of three long lines in parallel – effect of the loss of one line.
GL ¼ local generators

0.97 lagging, and it is economically justifiable to employ VAr injection by static capaci-
tors, synchronous compensators or Static VAr Compensators (SVCs) close by the load.
A problem arises with the operation of two or more lines in parallel, for example
the system shown in Figure 5.22, in which the shunt capacitance has been repre-
sented as in a p section. If one of the three lines is removed from the circuit because
of a fault, the system series reactance will increase from XL/3 to XL/2, and the
capacitance, which normally improves the power factor, decreases to 2C from 3C.
Thus the overall voltage drop is greatly increased and, owing to the increased I2XL
loss of the lines and the decreased generation of VArs by the shunt capacitances, the
power factor decreases; hence the possibility of voltage instability. The same argu-
ment will, of course, apply to two lines in parallel.

Example 5.8
Figure 5.23 shows three parallel 400 kV transmission circuits each 250 km long. The
parameters of the circuits are

Resistance : 0:02 ohm=km

Inductance : 1:06 mH=km
Capacitance : 0:011 mF=km
194 Electric Power Systems, Fifth Edition

G D
L1
Demand
L2
L3

Figure 5.23 Diagram of Example 5.8

The load demand varies between

Demand PD (MW) QD (MVAr)

Peak 1100 532.8

Off-peak 220 106.6

A power flow calculation for the two loading conditions shows the voltage at the
demand busbar to be 0.894 during peak demand and 1.015 for off-peak demand
conditions.

VG (p.u.) PG (MW) QG (MVAr) VD (p.u.) PD (MW) QD (MVAr)

1 1117.5 451.7 0.894 1100 532.8

1 220.6 304.4 1.015 220 106.6

To maintain the demand busbar voltage at 1 p.u. reactive compensation is needed

499:5 MVar ðcapacitiveÞ during peak demand condition

83:3 MVAr ðinductiveÞ under off-peak condition

VG (p.u.) PG (MW) QG (MVAr) VD (p.u.) PD (MW) QD (MVAr) Compensation

(MVAr)

1 1112.9 166.2 1 1100 532.8 499.5

1 220.5 216.3 1 220 106.6 83.3

If one line is lost during peak demand (N-1 security requires that no load is shed if
one circuit trips) the voltage at the receiving end would reduce to 0.745 p.u. and reactive
compensation of 625.5 MVAr (capacitive) would be required to bring the voltage to 1 p.u.

VG (p.u.) PG (MW) QG (MVAr) VD (p.u.) PD (MW) QD (MVAr) Compensation

(MVAr)

1 1139.9 983 0.745 1100 532.8 0

1 1119.7 40.6 1 1100 532.8 625.5
Control of Voltage and Reactive Power 195

Usually, there will be local generation or compensation feeding the receiving-end

busbars at the end of long lines. If this generation is electrically close to the load
busbars, that is low connecting impedance Zg, a fall in voltage will automatically
increase the local VAr generation, and this may be sufficient to keep the reactive
power transmitted low enough to avoid large voltage drops in the long lines. Often,
however, the local generators supply lower voltage networks and are electrically
remote from the high-voltage busbar of Figure 5.20, and Zg is high. The fall in volt-
age now causes little change in the local generator VAr output and the use of static-
controlled capacitors at the load may be required. As Zg is inversely proportional to
the three-phase short-circuit level at the load busbar because of the local generation,
the reactive-power contribution of the local machines is proportional to this fault
level. When a static or synchronous compensator reaches its rated limit, voltage can
no longer be controlled and rapid collapse of voltage can follow because any VArs
demanded by the load must now be supplied from sources further away electrically
over the high-voltage system.
In the UK and some other countries, many generators are some distance from the
load-centres. Consequently, the transmission system operator is required to install
local flexible VAr controllers or compensators to maintain a satisfactory voltage at
the delivery substations supplying the local distribution systems. Such flexible con-
trollers, based on semiconductor devices which can vary the VAr absorption in a
reactor or generation in a capacitor, are called FACTS (Flexible a.c. Transmission
System).
Typical values of compensation required for a 400 kV or 500 kV network are:

Peak load ¼ 0:3 kVAr=kW generating VArs

Light load ¼ 0:25 kVAr=kW absorbing VArs

5.9 Voltage Control in Distribution Networks

Single-phase supplies to houses and other small consumers are tapped off from
three-phase feeders connected between one phase and the neutral. Although
efforts are made to allocate equal loads to each phase the loads are not applied
at the same time and some unbalance occurs. In the distribution network (British
practice) shown in Figure 5.24 an 11 kV distributor supplies a number of lateral
feeders in which the voltage is approximately 400 V and then each phase, loaded
separately.
The object of design is to keep the consumers’ nominal 230 V supply within
6/ þ 10% of the declared voltage. The main 33/11.5 kV transformer is controlled
with an on-load tap changer to maintain the 11 kV busbar at a voltage approxi-
mately 5% above 11 kV. The distribution transformers have a secondary phase
voltage of 433/250 V which is some 8.5% higher than the nominal value of 230 V.
These transformers have taps of þ/5% that are only adjustable when the trans-
former is off-circuit (isolated).
 Power System Security Sll
I
Most of the security related functions deal with static "snapshots" of the
power system.They have to be executedat intervalscompatiblewith the rate
of changeof systemstate.This quasi-staticapproachis, to a large extent,the
only practical approachat present,since dynamic analysisand optimization are
conslder4bly mole {!fficu!! 4nd cqmpurallo44lly 1aqtelime corrsulurg,
System security can be said to comprise of three major functions that are
carried out in an energycontrol centre: (i) systemmonitoring, (ii) contingency
analysis,and (iii) comectiveaction analysis.
System monitoring suppliesthe power systemoperatorsor dispatcherswith
pertinentup-to-dateinformation on the conditionsof the power system on real
time basisas load and generationchange.Telemetrysystemsrneasure,monitor
and transmit the data, voltages,currents,current flows and the statusof circuit
breakersand switchesin every substationin a transrnission network. Further,
other critical and important information such as frequency, generator outputs
and transformertap positions can also be telemetered.Digital computers in a
control centre then processthe telemetereddata and place them in a data base
form and inform the operatorsin case of an overload or out of limit voltage.
13.1 INTRODUCTION
Important data are also displayed on large size monitors. Alarms or warnings
may be given if required.
In Chapter7, we have beenprimarily concerned with
the economical operation Stateestimation (Chapter 14) is normally used in such systemsto combine
of a power system' An equally important factor in
the operation of a power telemetereddata to give the best estimate (in statisticalsense)of the curreltt
system is the desire to maintain system security.
System security involves systemcondition or "state". Such systemsotten work with supervi$orycontrol
practicessuitably designedto keep the system operating
when componentsfail. systemsto help operatorscontrol circuit breakersand operateswitches and taps
Besideseconomizingon fuel cost and minimizrngemission
of gases(co, cor, remotely. These systemstogetherare called SCADA (supervisorycontrol and
Nox, sor), the power systernshould be operationally,.secure,,.
An operation_ data acquisition)systelns.
ally "secure" power system is one with low probability
of, systern black out The second ma-ior security function is contingency analysis. Modern
(collapse)or equipmentdamage.If the pro."r,
uf cascadingfailurescontinses. operationcomputershavecontingencyanalysisprogramsstoredin them.These
the systernas a whole or its tnajor parts may completely loreseepossiblcsystetntroubles(outages)beforethey occur.They study outage
collapse. This is
normally referred to as system blackout. All these events and alert the operators to any potential overloads or serious voltage
aspects require security
c ons t r a i n epdo w e r s y s te mo p ti mi z a ti on(S C O). vi ol ati tl ns.For exalnple,t he sir nplestf ir r m of cont ingency
analysiscan be put
Since security and economy normally have conflicting together with a standard LF program as studied in Chapter 6, along with
requirements, it is
inappfopriateto treat them separately.The fina.laim proceduresto set up the load flow dafa for each outageto be studied by the
of economy is the security
lunction of the utility company.The energy management LF plogram. This allows the system operatorsto locate def'ensiveoperating
system (EMS) is to
operate the system at minimum cost, with the guaranteed stateswhere no single contingencyevent will generateoverloadsand/or voltage
alleviation of
emergencyconditions.The emergencycondition will violation:;.This analysis thus evolves operatingconstraintswhich may be
dependon the severity of
t iolat io n so f o p e ra ti n gl i rn i ts(b ra n c hf' l ow sand bus cntpi oycdin t hc liD ( ccot r olnicdispat ch)and UC ( unitcor nnr it r nclrpr
t ) ogr ar r r .
vol tagel i mi ts).The most
severeviolationsresult fiom contingencies. Thus contingencyanalysiscarricsout ornergcncyidentil'icationancl"what if''
An irnportantpart of securitystudy,
therefbre,moves around the power system'sability simulations.
to withstanrjthe effects of
contingencies.A particular systemstateis said to be The third major security function, corrective action analysis, permits the
secureonly with reference
to one or more specific contingency cases, and operatorto changethe operationof the power systemif a contingencyanalysis
a given set of quantities
monitoredfor violation. Most power systemsare operated program predicts a serious problem in the event of the occurrenceof a certain
in such a way that outage.Thus this provides preventive and post-contingencycontrol. A simple
any singlecontingencywill not leaveother.o-pon"nts
heavily overloaded,so example of corrective action is the shifting of generationfrom one station to
that cascadingfailures are avoided.
another.This may result in change in power flows and causing a change in
loading on overloadedlines.
 j5! 2 ,1 M o d e rnP o @i s I srr
Thresethreeftrnctionstogetherconsist oi a very compiex set of toois that heip
in the secureoperationol'a power system.
o o
o -o = o
T3.2 SYSTEM STATE CLASSIFICATION o c o - L

o-
o o o
o
o 3 pb.
6 q

Dyliacco [13] and further clarified by Fink and Carlsen l23l in order to define
o, EE
relevant EMS (Energy ManagementSystem) functions. Stott et. al [15] have c
o (!6
8E
also presenteda more practical static security level diagram (see Fig. 13.1) by L
q)
o. BE
incorporating correctively secure(Level 2) andcorrectableemergency(Level4) o r: .9
security levels.
o
z qF - o =

In the Fig. 13.1, arrowed linep represent involuntary transitions between

E o
o
o o
o_ gH€
=E;.
' = O -
o. o_o x
Levels 1 to 5 due to contingencies..The removal of violations from Level 4 o- o--c.Y o
o
^. ;EEt b- E
normally requiresEMS directed "corrective rescheduling" or "remedial action"
!ie
E
o o
g,
o o c o
bringing the system to Level 3, from where it can return to either Level I or o
2 by further EMS, directed "preventive rescheduling" depending upon the
o
o) abg =98
c
o
(J
o

desired operational security objectives. o

L

o P c 9 c ()
f

Levels I and 2 representnormal power system operation.Level t has the E O = O o

U)
ideal security but is too conservativeand costly. The power system survives any d H d 8 C)
(E
of the credible contingencieswithout relying on any post-contingency corrective x g > o
action. Level2 is more economical,but dependson post-contingencycorrective
rescheduling to alleviate violations without loss of load, within a specified
'oE9 .b
.=
g)'E
g, =-E
.FOE

fi
o
c
c)
o)
E
o
o
c c c
period of time. Post-contingencyoperating limits might be different from their F O
(g(J E8E o
o o
pre-contingencyvalues. b o d o
8bb o

13.3 SECURITY ANALYSIS : 5d ' 58 o_

d( l ) - . o 9 c
. = 9 =
at
System security can be broken down into two major functions that are carried los i 6 H
5 c o
(t)
.d,
g#EE
.= z E C g
out in an operationscontrol centre: (i) security assessment,and (ii) security HE ' = lr
e o y 9 c
control. The tormer gives the security level of the system operating state. The
latter determines the appropriate security constrained scheduling required to =ee d 8 F.O
('5 o 3

o
optimally attain the target security level.
6- 0)
The security functions in an EMS can be executedin 'real time' and 'study' L o)
o o
modes. Real time application functions have a particular need for computing c)
a
E
o) -o
speed and reliability.
'fhe o
o
-o
([
8o o s 9
':=
static securit.vlevel of a power systemis characterisedby the presence - E r'/^ Y
() L

*, ^'kua to
- A

*, ^9i :
:( l:) -0 q)
or rrtherwise of emergency operating conditions (limit violations) in its actual a d
e ai
L
ocfr o)c,

(pre-contingency) or potential (post-contingency)operating states. System gE > o

_9!) C)
> o x
Az
J3(D
C.
> o
Ail
J3

security assessmentis the process by which any such violations are detected.
System e.:^
s€sstlt€rrt involves two func tions:
(i) system monitoring and (ii) contingencyanalysis.Systemmonitoring provides
the operator of the power system with pertinent up-to-date information on the
current condition:;clf the power system. In its simplest form, this just detects
violations in the actual systemoperating state.Contingency analysisis much
 ,.sfe ;l Modern power Svstem Analvsis Power System Security I_ srai
Only a small proportion of work on optimal power flow (OPF) has taken into SLACKBUS
account the security constraints.The most successfulapplications have beento
t +s*y ts f+o*ys
1.02421-5.0" 3
the security constrained MW dispatch OPF sub-problem. The contingency- 41.0236t-5.3"
1 . O 6 t O "1 --.>40.7+ j1.2 _ 39.5_i3.0:_ + 1 8 . 9- y 5 . 2 _ 1 6 . 9+ 1 3 . 2 - <
_
constrainedvoltageivar reschedulingproblem, as of the writing of this text, still
,f88.e-y8.6
,l
-Ja.a--/e.8 i6.3-j2.3
remains to be solved to a satisf desree. +
The total number of contingency constraintsimposed on SCO is enormous. --zt.s-i5.9
t
The SCO or contingency constrained OPF problem is solved with or without
first optimizing with respectto the base case(precontingency)constraints.The
I z+.2*js.a 'L?:'-q''---'----/
general procedure adopted is as follows: t- az.s* '-"''
l+-u
*S4.9 +17.3 -.i;n;l--+s
53.7-i7.2
' -* 5
(i) Contingency analysis is carried out and cases with violations or near
1.0474t-2 i1.o17gt-6.2"
violations are identified. Y
(ii) The SCO problem is solved. ) , zo+1t o *oo*ito
!G-)
(iii) The rescheduling in Step 1 might have created new violations, and t40 +i30
therefore step 1 should be repeatedtill no violations exist. Base Case AC Lineflow for sample5 bus system
Fig. 13.2
Hence, SCO representsa potentially massiveadditional computing effort.
An excellent comprehensiveoverview of various available methods is SLACKBUS
{+s*yts f+o+7s
presentedby Stott et. al [15].
1 . 0 1 0 7 1 -5 . 9 " 3
There is still great potential for further improvement in power system 1.0610" 1 4 1.00682- 6.6"
->48.6 + j5.2 -46.7 - j5.3<- + 38.5-10.6 -38.4 -i1.1
security control. Better problem formulations, theory, computer solution :-
-- { - r . o- 7 a. s
methods and implementation techniquesare required. +81.8-i 5.5 + 3a?-/'er

T3.4 CONTINGENCYANALYSIS

In the past many widespreadblackouts have occurred in interconnectedpower t- eo.o* i r . s- ; t . r

systems. Therefore, it is necessaryto ensure that power systems should be :63.1 + j1O.2 -<-- - 61.6 -i8.9 5
1.0468t-2. 1.O114t-6.4"
operatec!mosf economic:r!lysuch that povrer is cle!i.rerecl reliably. Reliable
operation implies that there is adequatepower generation and the same can be rlr { 2s+ilo loo *lt o
lJ./
\

transmittedreliably to the loads.Most power systemsare designedwith enough t40+i30

redundancyso that they can withstand all rnajor tailure events.Here we shall
Fig. 13.3 PostoutageAC LoadFlow(Linebetween2 and4 is open)
sttrdy thc possiblc consccprcnccs :rnclrcrncdialactions rcquircd by two nrain
f ailur e ev e n ts ;l i n e .o rrta g easn c lg e n e ra t i ngrrni tfai l ures. SLACKBUS
(c) t a sr r t s I a o, 7 s
To explainthe problelnbriefly,we considerthe five-bussystemof Ref'erence
0.se35/- 6.s" o1
LI0J. The basecaseload llow resultslbr the exampleare given in t ig. 13.2and, |
1 ' 0 6 t 0 "1 l ' 5 4 . 4+
s lt r r wir f ' lo w< tf2 4 .7 MW a n d 3 .6 MV AR on the l i ne f} om bus 2 to hus 3. L.ct _1__r i 1 i . 3- s 2 . 1 - i 9 . -7 | - s 1 . 4+ i 7 . 3 - s 1 . 1
| tl;f
Itl_qq.z_j1z.s
t ls lls s t llll cth a t a t p rc s c n l w, e i l rc o n l y i n tcrestecl
i n the MW l oadi ngof the l i ne. +80.e+13.21
Let us examinewhat will happenif the line from bus 2 to bus 4 were to open*. - 52.s-i 13.0+
' l' lr c | +s,s* 1tz.t
r c s ulti n gl i n c l ' l o w sa rtc vl o l ta g c su l c show ni n l ri g. 13.3.tt nray be notcd ':t't'
that the flow on the line 2-3 has increasedto 31.5 MW and that most of the ':: )ll
other line flows are also changed. It may also be noted fhat bus voltage
- 5
magnitudesalso get aff'ected,particularly at bus 4, the change is almost2To less
0.8994t-16.?
from 1.0236to 1.0068pu. Supposethe line from bus 2 to bus 5 were to open.
Figure 13.4shows the resultingflows and voltages.Now the inaximum change 1 2 0+ i 1 0
'foo*yro
t:fgl.g.j 5 which is almost 107o less.
",jus
xSimulationof line outageis more complex than a generatoroutage, 2 and5 is open)
Fig. 13.4 PostoutageAC Lcad Flow(Linebetween
since line
outageresultsin a changein systemconfigurations.
 -;'518
t- | rr,rodern
Po@is
5LAUK HUS

e t 4 s+ l 1 5 l+o*1s
| 1 . o o 6 1_t s . 7 " 3 1
1 . 0 6 t 0 "1'l_
l -!.6 + ^ _ I 41.O043t-6.1"
1t.s 4s.9_j7.6 <_i __22.4_ j2.6 _22 .3 +j0.7 <_
{nze+iz1.7l L - --i-1-I
I --- V_21.5_j4.8 { t.s - 1t.t

-25.2-j4.4 +

i '''"
I ---t
1 - r z o .-g1 t s . t (l - ' - _ - GiveAlarm signal
l l r f-zs-7s.s
2..]--u+ _+ _53.6+16.8 <- -52.5 -16.5 5
1.o24st-3.7.
i o 9956/_-7
.1"
Y
l , z o+ 1 l o
{oo*iro
GiveAlarm signal
Fig. 13.5 PostoutageAC Load Flow(Generator
2 outage,lostgeneration
is
pickedup by generator1)

Figure 13.5is an exampleof generatoroutageanclis selectedto explainthe

tact that generatoroutagescan also result in changes in line flows and
bus
voltages.In the exampleshowr in Fig. 13.5 all the generationlost from
bus 2
is picked up on the generatorat bus 1. Had there been more thanZgenerarors
in the sample system say at bus 3 also, it was possible the loss of
leneration
on bus 2 is made up by an increase in generation at buses 1 and 3.
The
differencesin Iine flows anclbus voltagesvrould show how the lost gener.ation
is shared by the remaining units is quite significant.
It is important to know which line or unit outageswill render line flows or
voltagesto crossthe lirnjts. To find the eff'ectsof outages,contingencyanalysis
techniques are empioyeci. Contingency analysis models single failure events --'t A-..

1;no*rifii.iuior".""-i;{offi 'l
( i' e' one -l i n co u ta g c soiro l l c u n i t o u tu g cs)or nrul ti pl c
ccpri pnrcnt
fi ri l urccvc^ts ' ''l''''
(failure of multiple unit or lines or their combination) one after another l'r'.,o
until all
"credible outages"are considered.For each outage,all lines ancl
voltagesin the
netrvork are checkedagainsttheir respectivelimits. Figure 13.6 depictsa flow _ _ l
,rr\=_ ves jciventarn
chart illustrating a simple method for carrying out a contingency analysis. violation?
->*-t_-qJ
One of the important problems is the selection of "all credible outages,,. -T-
Execution time to analyseseveralthousandoutagesis typically I min based
computerand analyticaltechnologyas of 2000.An erpproxirnate
DC load flow may be used to achievespeedy solution if voltage is
rnodelsuclras
on
,*" --\
I
aiso , N Ail
required, then full AC load flow analysishas to be carried out. '--o-i
I

FIg. 13.6 A simpletechniquefor contingency

analysis
 620j;:f uodern PowerSystemAnatysis
I
^ a ? frErrarE?rtrElt
IJ.C DIIIYDIIIVII Y .FAUI-L'I(s
limits and those violating their limit can be informed to the operator for
A .securityanalysisprogramis run in a load dispatchcentre very quickly to help necessarycontrol action.
the operators.This can be attempted by carrying out an approximate analysis The generation shift sensitivity factors are linear estimatesof the change in
and using a computer system having multiple processors or vector processors line flow with a change in power at a bus. Thus, the effects of simultaneous
y anarysts.I ne s may uate an equrvalent
should be used for neighboursconnectedthrough tie-linos. We can eliminate all principle of superposition.
non-violation casesand run complete exact program for "critical" casesonly. Let us assumethat the loss of the ith generatoris to be made up by governor
This can be achievedby using techniquessuch as "contingency selection"or action on all generatorsof the interconnectedsystemand pick up in proportion
"contingency screening", or "contingency ranking". Thus it will be easy to to their maximum MW ratings. Thus, the proportion of generationpick up from
warn the operation staff in advance to enable them to take corrective action if unit k (k * i) would be
one or more outageswiil result in seriousoverloadsor any violations. One of
(13.7)
the simplest ways to present a quick calculation of possible overloads is to
employ network (linear) sensitivity factors. Thesefactors give the approximate
change in line flows for changes in generationin the system and can be
calculated from the DC load flow. They are mainly of two types: where
1. Generation shift factors = maximum MW rating for rnth generator
Pn,,,,,u,^
2. Line outage distribution factors
g*i= proportionality factor for pick up on kth unit when ith unit fails.
Briefly we shall now describe the use of those factors without deriving them.
Now, for checking the /th line flow, we may write
Reference [7] gives their deri iation.
The generation shift factorsl cr.,;are defined as: j, = ff * 0r; APo, - \ri LPotl (13.8)
E,lau,
In Eq. (13.8) it is assumedthat no unit will violate its maximdm limit. For
(13.4)
unit limit violation, algorithm can easily be modified.
where, Similarly the line outage distribution factors can be used for checking if the
line overloadswhen solllc of the lines are lost.
4t = Change in MW power flow on hne I when a changein genera-
The line outage distribution factor is defined as:
tion, AP", takes place at the ith bus
Here, it is assumedthat LPotis fully compensatedby an equal and opposite d,,,= * (13.e)
change in generation at the slack (reference) bus, with all other generators Ji
remaining fixed at their original power generations.The factor al,then gives the where
sensitivityof the /th line flow to a changein generationat ith bus. Let us now
dt,i = line outagedistribution factor when monitoring /th line atter an
study the outage of a large generating unit and assume that all the lost outage of ith line.
generation (Pod would be supplied by the slack bus generation. Then
Aft = change in MW flow on /th iine'
APo, - -P1i (13.s) - precontingency line flow on ith line
fi
and the new power flow on each line could be calculatedusing a precalculated
set of " d' factors as given below. lf precontingency line flows on lines / and i, tlre power flow on line / with line
i out can be found out employing "d" factors.
ft = f i * dti APc, for all lines V / (13.6) (13.10)
?,= ff *d,,,f,o
where, ft - power flow on /th line after the failure of ith generator Here,
^d foi =precontingency or preoutageflows on lines / and i respectively
f i = power flow on /th line before the failure or precontingency fi
power flow
fr = power flow on /th line with ith line out.
 iii l Po
Modern J"srr
a
'd' factors ali the lines for
Thus one can check quickiy by precaiculating
overloading for the outage of a particular line. This can be repeated fbr the
j=l j=2 j=3 j=4 j=5 j=6 j = 7
outage of each line one by one and overloads can be found out for corrective
action. (line 1-2)(linel-3)(Iine 2-3)(line2-4)(line2-5)(line3-4)(line4-5)
It may be noted that a line flow can be positive or negative. Hence we must (l i n e l- 2) 0. 0 1. 0001- 0. 3331- 0. 2685 - 0. 2094 0. 3735 0. 2091
check / agarnst - Jt ^u* as well as h **. Llne tlows can be louncl out usmg
telemetry systemsor with stateestimation techniques.If the network undergoes =3
-J (line
\rr\rw 2-3)
L -J) --0.4542
v 0.4545 0 0.4476 0.3488 -0.6226 -0.3488
= 4 (hne 2-4) -0.3634 0.3636 0.4443 0.0 0.4418 0.6642 -O.44r8
any significant structural change, the sensitivity factors must be updated.
= 5 (line 2-5) -0.1819 0.1818 0.2222 0.2835 0.0 0'3321 1.0
= 6 (line 3-4) 0.5451 -0.5451 -0.6662 0.7161 0.5580 0.0 -0.5580
= 7 (line 4-5) 0.18i6 -0.1818 -0.2222-0.2835 1.0002 -0.3321 0'0

It has been found that if we calculate the line flows by the sensitivity
Find the generationshift factors and the line outagedistribution factors for the methods,they come out to be reasonablyclose to the valuescalculatedby the
five-bus sample network discussedearlier. full AC load flows. However, the calculations carried out by sensitivity
Solution Table 13.1 gives the [x] matrix for the five bus sample system, methods are faster than those made by full AC load flow methodsand therefore
together with the generation shift distribution factors and the line outage are used for real time monitoring and control of power systems.However,
distribution factors are given in Tables I3.2 and 13.3 respectively. where reactive power flows are mainly required, a full AC load flow method
(NR/FDLF) is preferred for contingency analysis.
Table 13.1 X Matrixfor Five-busSampleSystem(Bus 1 as a reference) The simplest AC security analysisproceduremerely needsto run an AC load
flow analysis for each possible unit, line and transformer outage.One normally
0 does ranking or shortlisting of most likely bad caseswhich are likely to result
in an overload or voltage limit violation and other casesneed not be analysed.
0 0.05057 0.03772 0.04029 0.4714 Any good P1(performanceindex can be selected)is usedfor rankirig. One such
P/ is
0 0.03772 0.08914 0.07886 0.05143
0 0.04029 0.07886 0.09514 0.05857 (13.11)

0. 0.04714 0.05143 0.05857 0.13095 For large n, PI will be a small numberif all line flows are within limit, and
will be large if one or more lines are overloaded.
Shift Distribution
Table 13.2 Generation Factorfor Five-busSystem For rr = I exact calculationsciur be done fbr P1. P1 tablecan be orderedfrom
largest value to least. Suitable number of candidatesthen can be chosen for
Bus I Bus 2 further analysis[7].
- 0.8428 If voltagesare to be included, then the following PI can be employed.
l=1(line1-2)
/=2(line1-3) - o.t572
(13.12)
/=3(line2-3) o.0714
I=4(line2-4) 0.0571
/=5(line2-5) 0.0286 Here, Alvil is the difference between the voltage magnitude as obtained at the
/=6(line3-4) - 0.0857 end of the lPlQ FDLF algorithm Alvl-u* it the value fixed by the utility.
-ue
I=l (line4-5) - 0.0285 Largest vaiue oi Pi is piaceciat the top. The security arraiysistrray rrow
startedfor the desired numbel of casesdown the ranking list.

$ummary and Further Reading:

Reference [25] has discussedthe concept for screening contingencies. Such
contingency selection/screeningtechniquesfonn the foundationfor many real-
time computer security analysis algorithms.
 l#GrffiE
IE!F}ILfr

Reference [15] gives a broad overview of security

assessmentand contain an stability, the dynamics mainly involves the loads and the means for voltage
bibliography covering the literature on"security
assessmenrup to control. Ref [11] provides a comprehensivelist of books, reports, workshops
;;;;t]r",
Reference [11] gives an excellent bibliography and technical papers related to voltage stability and security.
on voltage stability. This
topic is discussedbriefly in the next section. Definitions: [2]

13.6 POWER SYSTEM VOITAGE A power system at a given operating state is small-disturbance voltage stable
STABILITY if, following any small disturbance, voltages near loads are identical or close
Power transmissiol tuglgility has traditionally to the pre-disturbancevalues. The concept of small-disturbance voltage stability
been Iimited by either rotor is related to steady-statestability (Chapter 12) and can be analysedusing small-
angle (synchronous) stability or by thermal loaoing
capabilities. The blackout signal (linearised) model of the system.
problem has been linked with transient stability.
L-uckily this ;roblem is now A power systemat a given operating stateand subject to a given disturbance
not that serious because of fast short circuii
clearing; po*Lrru excitation ts voltage stable if voltages near loads approach post-disturbanceequilibrium
systems, and other special stability controls.
ElectriJ *rnpuni"s are now valuei. The concept of voltage stability is related to the ffansient stability of a
required to squeezethe maximum possible power
through networks power system. The analysis of voltage stability normally requires simulation of
owing to various constraintsin the construction of generation "ii.G
and transmission the system modelled by non-linear diffdrential-algebraic equations.
facilities.
voltage (load) stability, however, is now a A power systemat a given operating stateand subject to a given disturbance
main issue in planning and undergoes voltage collapse if post-disturbanceequilibrium voltages are below
operating electric power systemsand is a factor reading
to limit po-w; transfers. acceptable limits. Voltage collapse may be total (blackout) or partial. The
voltage stability is concerned with the ability
of a power system to maintain voltage instability and collapse may occur in a time frame of a second.In this
acceptable voltages at all busesin the system under
normal conditions and after case the term transient voltage stability is used. Sometimes it may take up to
being subjected to a disturbance.A power system
is said to have entereda state tens of minutes in which case the term long-term voltage stability is used.
of voltage instability when a disturbante
resurts in a pro!."rriu" and The term voltage security means the ability of a system, not only to operate
uncontrollable decline in voltage
Inadequate reactive.-powersupport from generators stably, but also to remain stable following any reasonably crediblebontingency
and transmission lines or adverse system change such as load increases[2].
leads to voltage.instability ot uoitug" collapse,
which have resulted in several Voltage stability involves dynamics, but load flow based static analysis
major system failures in the world. Th"y -",
(i) south Florida, usA, system disturbanc methods are generally used for quick and approximate analysis.
e of 17 May r9g5, (transient, 4 Figure 13.7 depicts how voltage stability can be classified into transient and
sec)
long-term time frame l2l.
(ii) French systemdisturbancesof Decembe
r 19, r97g and January 12, rgg7,
(longer term). Transientvoltage stability Longer-term voltage stability
(iii) swedish systemdisturbanceof December
27, rgg3 (longer term, 55 sec) Inductionmotor dynamics Increase in load/powertransfer
(iv) Japanese(Tokyo) system disturbance
of July 23, 1gg7irorg", term, z0
min) Generator/excitationdvnamics LTCtransf& Distvolt.Reg.
(v) NREB grid disturbancein India in 19g4 Primemovercontrol Load diversity /thermostat
and r9g7.
(vi) Belgium, Aug 4, 1992. (longer term, Mech. switched capacitors/reactors Excitationlimiting Gasturbinestart-up
4.5 min)
(vii) Baltimore, washington DC, usA,
5th July 1990 (longer rerm, insecure Under voltageload shedding
for hours)
Hence, a full understandingof voltage stability phenomena
and designing
mitigation schemesto prevent voltase instabilitv is nf o,raqr'or,,o+^,.+:r:a:^^
', ;,ilir; ;;,:;; ffiri#il;"; Protective relaying includingoverload protection
;agestability.
lf phenomena.Becadseof this, voltage
'ent engineers.
Voltage instabilitv and
rterchangeably.by many ."r"ur"h..r. 100
voltage instability or co[apse is a faster dynamic-process. Time-seconds-
As opposedto angle
Fiq. 13.7 Voltage stabilitVphenomenaand time responses
 SPfti*f uooern po@is wffi;Ewe
I
Voltage stability problems normaliy occur
in heavily stressed systems.
voltage stability and rotor angle (or synchronous) desirable.
stability are more or less
interlinked' Rotor angle stability,
as voltage stability is affected by
.a3-.wex
reactive power control. voltage stability
is concern"o *itt, load areas and load Unity powerfactor

rvr^rvlv vv'Yvr Pr'lurs tu a large system

;;;''',h.;A#";
over long transmission lines. Voltage
':::i:y,"::,*:,','31^ and ,oto."ungl"stabilityis basicauy Nosepoint(knee)
Seneratorstability.In a ^,?::^s,112itiu
large inter-connectedsystem,voltagecoilapse t
area is possible without lois of synchroni* of a load \
or any generators. Loaw of V6 and Pr"r;
The slower forms of voltage instability same for const Z load
are often analysed as steady-state for other types of load,
problems' 'snapshots' in time following V degradationis faster.
an outage or during load buildup are
simulated' In addition to post-disturbancl
load floivs, two other load flow based
methods are widely used: P-V curves and
e-v curves. pV curves are used for
especially for radial systems.
/ curves (Fig. 13.g),
eV curves (Fig.
'ig. 13.8) and methods PplP6zy
to quantify nose
puted. Power flow analysis cletermines 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

asvanous
svstem
parameters
and |HI'I,:IJHI# Fig. 13.8 PV curveswith differentload powerfactors
""lT:ir'#n#'f
voltage exist for each value of load. The
uppe. on" indicates stable ;;i;;;
whereas lower one is the unacceptable value Only the operating points above the critical points represent satisfactory
(multiple load flow). At limiting
of voltage stability i.e. at nose point single operating conditions. At the 'knee' of the V-P curve, the voltage drops rapidly
l!ug" ioad flow solution exists.
Nearer the nose point, lesser is the staLility with an increasein load demand. Power-flow solution fails to convefgebeyond
maigin,
this limit indicating instability. Operation at or near thel stability limit is
Effective counter Measures to prevent impractical and a satisfactory operating condition is ensured by permitting
or contain voltage
Instability sufficient "power margin".

(i) Generator terminal voltage shourd

be raised. 1400 range
(ii) Generator transformer tap value j+Allowable
may be increased.
(iii) Q-injection should be carried out 1200
at an appropriate location.
(iv) Load-end oLTc (on-load tap 1000
changer) should be suitably used. P> P, > P,, iNMW
(v) For under voltage conditions, strategic
to'
load shedding should be resorted ffuo. Boo
System reinforcement may be carried out
by instpiling new transmission
lines between generation and load centres. Systemcharacteristics
series and shunt compensationmay
be carried out and svcs (static var compensation)
may be installed. Generation
rescheduling and s_tarting-upof gas ,rrui-"*
,.y i" .,*.i.,t out.
Practical aspectsof Q-flow problemsleading
to voltage collapsein EHv lines:
(i)
For lons line.s urifh 'r-^^nrrnu^r L--^^- 0 . 9 0 . 9 51 . 0 1 . 0 5
"-!' s'vv'Lr.,'e' uus's, recelvlng_end or road voltages
increase for light load conditions and Capacitor Vin pu
decrease for heavy load conditions.
(ii) For radial transmission lines, characteristics
if any loss of a line takes place, reactance
goes up, I2x loss increases resulting
in increase in voltage drop. This
Fig. 13.9 Systemand shunt capacitorsteady-stateQ-V characteristics,
should be suitably compensated capacitorMVAr shown at rated voltage
by local e injection.of course this
involvescost.If thereis a shortageof rocal q
,ourrr, thenimport of e
 M
Voltage Collapse
;
Voltage collapse is the processby which the seguenceof events accompanying (v) Post-disturbanceMwA4vAR margins should be ffanslated to pre-
voltage instability le_adsto unacceptable voltag6 profile in a significant part disturbanceoperatinglimits that operatorscan monitor.
of
the power system. It may be manifested in ieverat different"ways. (vi) Training in voltage stability basis (a training simulator) for control centre
Voltage
collapse may be characterised as follows:
and power plant operators should be i

SUIvIMARY
Power system security (including voltage stability) is likey to challenge
planneqs,analysts; researchersand operators for the foreseeablefuture. As load
grows, and as new transmission lines and new generations would be
(iii) The voltage collapse generallymanifests itself as a slow increasingly difficult to build or add, more and more utilities will face the
decay of voltage.
It is the result of an accumulative process involving the actions and security challenge.
interactions of many devices, controlJ, and protective iyrt"*r. The time Deregulation and socio-economic ffends compounded by technological
frame of collapse in such caseswould be of the order of several minutes. developmentshave increasedthe likelihood of voltage instability.
Voltage collapse is strongly influenced by system conditions Luckily many creative persons are working tirelessly to find new methods
and
characteristics. and innovative solutions to meet this challenge.
(iv) Reactivecompensationcan be made most effective by the judicious
choice
of a mixture of shunt capacitors, static var system and possibly
synetuonouscondensers.
NCES
REFERE
Methods of Improvlng Voltage Stabllity
Books
voltage stability can be improved by adopting the following rneans:
(i) Enhancing the localisedreactivepower support (SVC) L l.J. Narguthand D.P. Kothari, PttnterSystemEngincering,'fataMc0raw.Hill, New
is more efl,ective
and C-banks are lnore economical. ra-tS devices or synchronous Dclhi, 1994.
condenser may also be used, 2, C.W, Taylor, Power SystemVoltag,eStabiliry,McGraw-Hill, New york, 1994.
(ii) Compensating the line length reduces net reactance and power 3. P. Kundur, Power SystemStability and Contol, Sections2.12, ll.2 and Chapter
flow
increases. 14, McGraw-Hill, New York, 1994.
(iii) Additional transmissionlinc muy be crectccl.It also 4, T,J.E,Miller, Editor, ReactivePower Control in ElectricSyslens,John Wiley and
improves reliability.
(iv) Enhancing excitation of generator, system voltage improves Sons, New York, 1982.
and e is
supplied to the system. 5. A. chakrabarti, D.P. Kothari and A.K. Mukhopadhyay, Perforurutnce,operation
(v) HVDC tie may be used between regional grids. and Control of EHV Power Transmission Systems,Wheeler Publishing, New
(vi) By resorting to strategic load shedding, voltage goes Delhi. 1995.
up as the reactive 6. T.V. Cutsemand C. Vournas, VoltageStability of Electric Power Syslerns,Kluwer
. burden is reduced.
Academic Publishers,London, 1998.
Future Trends and Challenges 7. A.J. Wood and W.F. Wollenberg, Power Generation,Operation, and Control, Znd
Edn, John Wiley, New York, 1996.
(i\ Ontirnql oifi-- n'F EA/-'n-r ,{^*,:^^^
urrrrr6 vr I nv
\-./ i, \rlvv.ltvgD.
E. John J. Gratngerand W.D. Stevenson,Power SystemAnalysis,McGraw-Hill, New
(ii) Betterand probabilisticloadmodelling. York, 1994.
(iii) Developtechniquesandmodelsfor studyof non-lineardynamics 9. G.L. Kusic, Computer-AidedPower SystemsAnalysis,Prentice-Hall,New Jersey,
of largr
siz.esystems.For example,new methodsto obtainnetworkequivalentr 1986.
suitablefor voltagestabilityanalysis. 10. G.W. Stagg and A.H. El-Abiad, Computer Methods in Power System Analysis,
McGraw-Hill, New York, 1968.
Papers
11. v. Ajjarapu and B. Lee, "Bibliography on voltage stability", IEEE Trans. on
Power Systems,Vol. 13, No. 1, February 1998, pp lI5-125,
12. L.D. Arya, "Security ConstrainedPower System Optimization", PhD thesis, IIT
Delhi, 1990.
t4
13. T.E. Dyliacco, "The Adaptive Reliability Control System", IEEE Trans. on pAS,
Vol. PAS-86, May 1967, pp 517-531
(This is a key paper on system security and energy control system)
14. A.A. Fouad, "Dynamic Security AssessmentPractices in North America", IEEE
Trans. on Power Systems,Vol. 3, No. 3, 1988, pp 1310-1321.
15. B. Stott, O. Alsac and A.J. Monticelli, "security Analysis and Optimization", proc
IEEE, VoL 75, No. 12, Dec. 1987, pp 1623-1644.
16. Special issue of Proc. IEEE, February 2000.
17. P.R. Bijwe, D.P. Kothari and L.D. Arya, "Alleviation of Line Overloads and
voltage violations by corrective Rescheduling", IEE proc. c, vol. 140, No. 4,
T4.I INTRODUCTION
July 1993, pp 249-255.
18. P.R. Bijwe, D.P. Kothari and L.D. Arya, "Overload Ranking of Line Outageswith
State estimation plays a very important role in the monitoring and control of
postourage generation rescheduling", Int. J. of Electric Machines and Power
modern power systems. As in case of load flow analysis, the aim of state
Systems,Yol. 22, No. 5, 1994, pp 557-568.
estimation is to obtain the best possible values of the bus voltage magnitudes
19. L.D. Arya, D.P. Kothari et al, "Post Contingency Line Switching for Overload
and angles by processing the available network data. Two modifications are,
Alleviation or Rotation", Int J. of EMPg Vol 23. No. 3, 1995, pp 345-352.
however, introduced now in order to achieve a higher degreeof accuracy of the
20. P.R. Bijwe, S.M. Kelapure, D.P. Kothari and K.K. Saxena,"Oscillatory Stability
solution at the cost of some additional computations. First, it is recognised that
Lirnit Enhancement by Adaptive Control Rescheduliig, Int. J. of Electric Power
the numerical values of the data to be processed for the state estimation are
and Energy Systems,Vol. 21, No. 7, 1999, pp 507-514.
generally noisy due to the errors present. Second, it is noted that there are a
21. L.D. Arya, S.C. Chaube and D.P. Kothari, "Line switching for Alleviating
larger number of variables in the system (e.9. P, Q line flows) which can be
Overloadsunder Line OutageCondition Taking Bus Voltage Limits into Account",
measured but arc not utilised in the load flow analysis. Thus, the process
Int. J. of EPES, Yol. 22, No. 3, 2000, pp 213-ZZl.
involves imperfect measurements that are redundant and the process of
22. P.R. Bijwe, D.P. Kothari and S. Kelapure, "An Effective Approach to Voltage
estimating the system statesis based on a statistical criterion that estimates the
Security and Enhancement",'Int. J. of EPES, Yol. 22, No 7, 2000, pp 4g3-4g6.
true value of the state variables to minimize or maximize the selected criterion.
23. L. Fink and K. carlsen, "operating under sttress and Strain", IEEE spectrum,
A well known and commonly used criterion is that of minimizing the sum of the
March 1978, pp. 48-50.
squares of the differences between the estimated and "u1le" (i.e. measured)
24. S.M. Kelapure, "Voltage Security Analysis and Enhancement", Ph.D. thesis, IIT
values of a function.
Delhi,2000.
Most state estimation programs in practical use are formulated as
25. G.C. Ejebe, et. al, "Fast Contingency Screening and Evaluation for Voltage
overdetermined systems of non-linear equations and solved as weighted least-
security Analysis", IEEE Trans. on Power systems,vol. 3, No. 4, Nov. l9gg, pp
squares(WLS) problems.
1582-1590.
State estimatorsmay be both static and dynamic. Both have been developed
26. T. Van Cutsen, Voltage Instability: "Phenomena,Counter measures,and Analysis
for power systems.This chapter will introduce the basic principles of a static-
Methods", Proc. IEEE, Vol. 88, No. 2, Feb. 2000, pp 208-227.
state estimator.
ln a power system, the state variables are the voltage magnrtudes and phase
angles at the buses. The inputs to an estimator are imperfect (noisy) power
system measurements.The estimator is designedto give the "best estimate" of
the system voltage and phase angles keeping in mind that there are errors in the
measuredquantities and that there may be redundant measurements.The output
data are then used at the energy control centres for carrying out several
 An lntroductionto State Estimationof Power Systems
or,01-li,r:,systeT^studics
suchas cconornicclisplrch(Chaprcr7),
::if.-jl:t
strLurl-y analysts (Lnapter J =7'V /\( r|1+/. J4,(, .\ 5 )
lJ).

From Eqs. (14.31) and (14.4),one gets the following expressionfor the index:
r4-2 LEAST souARES ESTIMATION:THE BAsrc ( r4.6)
soluTroN J =yt!- y'H*.- i<'H'y+ *.'HtH*.
I7l- IeI
For minimizing J = f$), we must satisfy the tollowing condition.
A lrr
ns wilf be seen iater in Section r4.3, the probrem
of power system state gradoJ = 0 \r+.t)
estirnation is a special case of the more general problem
of estimation of a
random vector x from the numerical values of anotirer It is easyto check (see,e.g. [1]) that Eq. (1a.7)leadsto the following result.
related random vector y
with relatively little statistical information being available
for both x and y. In HtH*,-H'y-O (14.8)
such cases,the method of least-squared-errorestimation
may be utilised with 'notmal equation' and may be solved explicitly
good results and has accordingly been widely This equation is called the
employed.
Assume that x is a vector of n random variabres for the LSE of the vector i as
x1tx2, ..., x' that y is
another vector of m (> n) random variables (14.e)
!1, J2, ..., J^ and both are related *. = (H,Il)-t Ht y
as
!=Hx+r (r4.1)
where H is a known matrix of dimensionmx n
anclr is a zero mean rancklm
E;il;il]
r . i
variable of the same din-rensionas y. The vector
x representsthe variablesto
be estimated,while the vector y representsthe ln orclerto illustrate the methodof LSE, let us considerthe simple problem of
variables whose numerical
v aluesa re a v a i l a b l eE. q u a ti o n( l 4 .l ) suggests thatthe measurement estimating two random variables x, and .rz by using the data for a thret:
vectory i s
linearly related to the unknown vector x and in dimensionalvector y.
addition is corrupted by the
vector r (error vector).
The problem is basicallyto obtain thc best possible l-' 0.1
value of the vector x
from the given values of the vector y. Since the variable Assume H=10 ll
r is assumedto be zero
m ean,o n e ma y ta k e th e e x p e c ta ti o n o f E q. (14.1)anclget the rel ati on Lrrl
f = Hx (r4.2) The matrix Ht F1 is then given bY
whcre I , , = cxpcctcclvalue ol' x iurd y, respec:tively. I
u'u =12 I unaits inverse
is
This shows that the load flow methods of chapter 6 Ll 2)
the mean valuesof the bus voltages.Howeu"r, tn.
could be used to estimate
=f ''':' -:':1
woulcl like to estimatethe
actual values of bus voltagesrather than their averages. (H'm-' u3 2131
One possibleway of obtainingthe bestpossibleestimate
l-
of the vectorx from It is easy to form the vector Hty and combining this with the itrverse of
y lies in the use of the method of leastsquareestimation
(LSE). To developthis (H'H), the following estimateof x is obtained.
rneth<td,assllmethat i rcpresetttsthe desireclestirnate
of ,r so that y given by f rrt lat -. /1 t1\/ r
-l

the equation ^ \Ltr)/r-\rrt)\/2-Yz)

* =| |

j' = Hf,
/
L-(t 3 ) y r * ( 2 / 3 ) y r 1 ( r / 3 )
l v-,
(r4.3)
representsthe estimateof the vector y. The effor Weighted LSE
! of the estimation of y is then
given bv mI ..
lnq
grsllll|:.iltr
Eilv€il !rl D9. \!+,2) ls ulrerr lr-rsrrsu r\, (ri) rrru \rr\.rrrr(uJ r1,61ir

(r4.4) squaresestimate and is obtainedby minimising the index function that puts
The estimate i is defineclto irc thc I-sE if it is equal weightage to the effors of estimation of all componentsof the vector y.
cornputcclby rninirnizingthe
estimation index J given by It is often desirableto put.dift'erentweightageson the ditt'erentcomponentsof
y since some of the measurementsmay be more reliable and accuratethan the
othersand these should be given more importance.T'o achievethis we define
the estimation index as
 rvtouern Power

I = l'wf / 1 4 1 n \
An lntroduction of PowerSystems-T--l" ,3iiSl:
to StateEstimation
\ r.i. r vi,
where W is a real symmetric weighting matrix of dimension m x ru. This
is [o.r I
often chosen as a diagonal matrix for simplicity.
It is relatively straightforward to extend the method of LSE to the weighted
w =t l I I
l
form of "I and to derive the foilowing form of the normal equation. L o.tj
The matrir HLWH ts
H|WH?_ H'Wy _0
G4.Ila)
This leads to the desired weighted least squares estimate (WLSE) H,wH-l''' o'tl
*. = (H,W H)-t H,W y 10.1 1.lJ
(14.11b)
This pertains to minimization as the hessian 2H|WH is a non-negative and the matrix HtW is obtained as

H ' | w = f o 'ot o t - l
definite.
Some Properties:
L0 1 0.1j
Rewriting Eq. (14.ilb) as
Theweightedleastsquares
estimateof the vectorx is thenobtainedas(from
i=ky (I4.I2a). E q .( l a . 1 1 b ) )
where k = (H'WH)-' Ht W. (r4.rzb) ^ (lr/21) y, - (r}/Zt) y, (t0/2t) yr
Here the matrix k dependson the value of H and the choice of w. *=[ I
L- ellr) y, (2ot2r)
y, (r/2\ y, )
using Eqs. (14.1) and (r4.lzb) it is easy to get the relation as follows.
If this result is compared with the result in Example I4.1, the effect of
r introducing the weighting on the estimate is apparent.Note that the choice of
=f::;,7, ,,,wH)x + kr W in this case suggeststhe data for y2 is consideredmore valuable and this
resultsin the components of x being more heavily dependenton
)2.
Or X =r + kr (14.r3) The matrix ft is in this case found to be (Eq. (1a.12b))
and E{i} =E{xl (14.r4)
-ro/2r ro/zrf
rc=lrr/21
ln IJq. (14.14) it is assumedthat the error r is statistically independent
of L- rlzt 20/21 r/21J
columns of H and the vector r has a zero mean. An estimate that If the covariance of the measurementerror is assumedto be R = L the
satisfied
Eq. (l 4)4) is calledan unbiasedestimate.This implies that the estimation covarianceof the estimation error is obtained as (Ref. Eq. (14.15c))
error
ls zero on an averase.
P,= (1tr47) *:_
x =kr (14.15a) |L - 6 7 .::1
r34 )
The covariance of the error of estimation is therefore given by
The choice of W above yields unacceptablylarge estimation error variances.
P, = KRK (14.1sb) Let us now choose the weighting matrix W = I. The matrix Kis then obtained
where R is the covarianeeo1 the error vector r. Note that the covariancep" AS
is
a lneasureof the accuracy of the estirnaiion and a smaiier trace of this -1/3t/3f
matrix
indicatesa better estimate.Eq. (14.15b) suggeststhat the best possible
choice
*=l''3
of the weighting matrix is to set w - R-1.The optimum value of L- U3 2/3 1/3J
the error
covariance'matrix is then given by The error covariance matrix is then given by
P, - (HtR-rH)-l 6 -3]
(14 15c)
rD- x = \/ rL/ tn>\ ) [
t---*--
** L_ E 6 J
1Et<ample14;2 The error variances are now seen to be much smaller as is to be expected.

Non-linear Measurements
Assumethat in the Example I4.1, we want to obtain the WLSE of the variable
x by choosing the following weighting matnx The case of special interest to the power system state estimation problem
correspondsto the non-linear measurementmodel.
 I
536 | Modern Power SystemAnalysis An lrylfgduction
to_StateEstimation
of PowerSystems
I
I
I S3T
I
!=h(x)+ r (14.16) useful result in the sensethat it shows a mechanismfor improving on the initial
where h(x) representsan rz dimensional vector of nonlinear functions of the estimafeby making use of the available nleasurements. Having obtainednew
variablex. It is assumedthat the componentsof the vector h(x) arecontinuous estimatei , the processof linearisationis repeatedas many times as desiredand
in their arguments and therefore may be differentiated r.vith respect to the this leads to the {qllq*tng 4erylle {qryq qf the qqb]ro4 of thq 4on-linear
componentsof x. The problem is to extendthe methodof least squaresin order estimationproblem.
to estimate the vector x from the data for the vector y with these two variables
i ( t + 1 ) = i ( t )+ K ( t ) { y - h t t ( / ) l } (r4.23)
being relatedthrough Eq. (14.16).
To mimic our treatment of the linear measurementcase, assumethat i where the matrix K(D is deflned as
representsthe desired estimate so that the estimateol the measLlrementy could K(t) - LH/ wHil-' al w (r4.24)
be obtained using the relation The index / representsthe iteration number and H lrepresentsthe value of the
I = h(i) (1.4.17a) Jacobian evaluated at x = i (l). Usualiy the iterative process is terminated
wheneverthe norm of the differenceof two successivevalues of the estimate
This yields the error of estimationof the vectol y
i 1t + l) - .i (/) reachesa pre-selecredrhresholdlevel.
j =y_ h(3) (r4.r7b) A flowchart for implementing the iterative algorithm is shown in Fig. 14.7.
In ordcr to obtain the WLSE of ,,r,we nrust cho<lsethe index of estimation A major sourceof computationin the algorithmlies in the need to updatethe
J as follows. Jacobianat every stageof iteration.As discussedearlier in Chapter6 (seeEqs.
(6.86) and (6.87)) it is often possibleto reducethe computationsby holding rhe
J=U- h ( f i ) l ' W l J- h ( i ) l (14.18)
value of H a constant,possibly after / exceeds2 or 3. T,ris is in general,
The necessarycondition for the index ,I to have a minimum at .x,is given by permissible in view of the fact that the change in estimate tends to be rather
Eq. (14.1e). small after a couple of iterations.

ty-h(x)lH(i)=o (r4.re)
where H (fr) is the Jacobianof h (x) evaluatedat i. In general this non-linear
equation can not be solved for the desired estimate, i. n way out of this
difficulty is to make use of the linearisa-tiontechnique. Let us assumethat an
a priori estimate xu of the vector x is available (say from the load flow
solution).
Using Taylor seriesapproximation,we get
! = h ( x d + H u & - x 1 ) +r (14.20)
where 1/o standsfor the Jacobian evaluated at x = x9 and the noise term r is
now assurnedto include the effects of the higher order terms in the'Taylor
series.Equation (14.20)can be rewrittenas:
A ) ' = y - h ( x o ) =H s A x + r (14.2I) i ,00"," ) eq.(+..227
/
where Ay is the p^-rturbedmeasurementand Ax is the perturbed value of the I
vector r. An \\LSE of .r is then ea^silvobtaineda-sdiscussedearlier and this -,=n) \,* ! -_I\,,ll
c"l::'1:
leads to the desired expressionfor the lineaized solurion of the non-linear -''!'
estjmat.ionproblem. ./ ls
:- \-
.-2a.,
t No
.i. = .rrr- [H,,' I+'HQ]- H()' ]l' {-r - h t.t,,,)} r11.22)
r Yes
I r is no r l i k e l v th a t th e e s ti ma te.i o b tai nedfrom E q. (11.22r i s goi ngto be
t - r f I r t u c i r u s e s i l t . ' c . i r t t e n e r . r i . t l t r ' . r i , t ' r t t t ' t c ' s t i r l u t e . l ' , . rI l ) J \ - I l r r t h ' c l o s e t i r t h f
\
Stcp
(Dt-irnalydue rf rhe \ect()r x. Hrt,*'ever,F,q.(14.22) prot'ides us wirh a very
rig. r+.1
 ',:83--E-,
I
I
todern powerSystemAnalvsis An tntroductionto State Estimationof Power Systems _r-
| 539

It is thus apparentthat the problem of estimation of the power system state

'is a non-linear problennand may be solved using either the batchpr:ocessingor
sequentialprocessingformula [see Section 3.3 of Reference 1]. Also, if the
Considerthe simple caseo1'ascalar variablex and assunlethat the relationship svstem is assumedto have reacheda steady-state condition. the voltageangles
_xls glven Dy
Y = x 3+ r problem is then a static problem and the methodsof Sec. 14.2 may be used. if
so desired. To develop explicit solutions, it is necessaryto start by noting the
The JacobianH, is easily obtained in this caseand the iterative algorithm takes
exact forms of the model equations for the components of the vector y (k).
the explicit form
Let P, arfi Qi denote the active and reactive power injections of ith bus.
i (t + 1)= i (t) + t3i (Dl-z {y - ti (Dl3} These are related to the components of the state vector through the following
where we have used W = 1. equations.
Let the correct value of x/ be eclualto 2 and assumethat due to the effect N
of r the measuredvalue of y is found to be 8.5. Also, assumethat the initial P , =D I Yj | | Vjl lYij I cos (- 6t + 6, * 4ti) (r4.2s)
estimatex (0) is taken to be equal to 1. The table below gives the results of the j --r
first few iterations. N

e,= -D | % l I v j l l Y i j l s i n( - 6 , + 5 t + 0 ; ) (r4.26)
, (t) j:1
0 1.0
I 3.5 where I IU I repre5entsthe uragnitudeand l/U- representsthe angle of the
2 2.56 admittance of the line connecting the lth and7th buses.The active and reactive
_)
a
2.t6 componentsof the power flow from the ith to the7th bus, on the other hand are
given by the tbllowing relations.
It is apparentthat the algorithmwould yield the correctsolutionafter several P,j= | vil I Vjl I Yij I cos (d, - 6i * 0,) - l v i P l Y , , l c o s9 , ,
iterations. (r4.27)
- -
14.3 STATIC STATE ESTIMATION OF POWER SYSTEMS Q , j = l % l I V j l l Y i j l s i n ( d ' 6 i * 0 t ) I viP I Y,,I sin Iij
(14.28)
lrol-lr2l
Let us assume that fhe vector y has the general form
As noted earlier, for a system with N buses,the state vector x may be defined J = lPt . . . PN Q t . . . Q r u Pr z . . . Pu - l, N Qn . . . Q 7 , J- 1 , 7 s ,
as t hc 2N - | v c c l tl rtl l ' l l tc N -- | v o l l agcangl cs62,..., 6" and thc N vol tage
)2 b i nl vl 1 l , . . . . ,I y N I l /
magnitudes/1, v2, ..., v". The load flow data,dependingon type of bus, are
generally comrpted by noise and the problem is that of processing an adequate (r4.2e)
set of available data in order to estimate the state vector. The readily available
The Jacobian H will theh have the form
data may not provide enough redundancy (the large geographical area over
Ht Hz
which the system is spreadoften prohibits the telemeteringof all the available
tnedsurementsto the central computing station).The redundancyfactor, defined H3 H4
as the ratio m./n should have a value in the range 1.5 to 2.8 in order that the Hs H6
computedvalue of the statemay have the desiredaccuracy.It may be necessary (14.30)
H_ H7 Hs
to irleh-rdethe data for the power flows in both the directions of some of the tie
lines in order to increase the redundancy factor. In fact, some 'psuedo
measurements'which representthe computedvalues of such quantitiesas the /ru-r o
active andreactiveinjectionsat some remotebusesmay alsobe includerlin the o IN
vector y (k). (N - l; submatrix
where /" is the iclentity matrix of dimension N, H, is the N x
of the partial derivatives of the active power injections wrt 6's, H2 is the
11 Power System Security

11.1 INTRODUCTION

Up until now we have been mainly concerned with minimizing the cost of
operating a power system. An overriding factor in the operation of a power
system is the desire to maintain system security. System security involves
practices designed to keep the system operating when components fail. For
example, a generating unit may have to be taken off-line because of auxiliary
equipment failure. By maintaining proper amounts of spinning reserve, the
remaining units on the system can make up the deficit without too low a
frequency drop or need to shed any load. Similarly, a transmission line may be
damaged by a storm and taken out by automatic relaying. If, in committing
and dispatching generation, proper regard for transmission flows is maintained,
the remaining transmission lines can take the increased loading and still remain
within limit.
Because the specific times at which initiating events that cause components
to fail are unpredictable, the system must be operated at all times in such a
way that the system will not be left in a dangerous condition should any credible
initiating event occur. Since power system equipment is designed to be operated
within certain limits, most pieces of equipment are protected by automatic
devices that can cause equipment to be switched out of the system if these limits
are violated. If any event occurs on a system that leaves it operating with limits
violated, the event may be followed by a series of further actions that switch
other equipment out of service. If this process of cascading failures continues,
the entire system or large parts of it may completely collapse. This is usually
referred to as a system blackout.
An example of the type of event sequence that can cause a blackout might
start with a single line being opened due to an insulation failure; the remaining
transmission circuits in the system will take up the flow that was flowing on
the now-opened line. If one of the remaining lines is now too heavily loaded,
it may open due to relay action, thereby causing even more load on the
remaining lines. This type of process is often termed a cascading outage.
Most power systems are operated such that any single initial failure event
will not leave other components heavily overloaded, specifically to avoid
cascading failures.
Most large power systems install equipment to allow operations personnel
to monitor and operate the system in a reliable manner. This chapter will deal
410
 INTRODUCTION 41 1

with the techniques and equipment used in these systems. We will lump these
under the commonly used title system security.
Systems security can be broken down into three major functions that are
carried out in an operations control center:

1. System monitoring.
2. Contingency analysis.
3. Security-constrained optimal power flow.

System monitoring provides the operators of the power system with pertinent
up-to-date information on the conditions on the power system. Generally
speaking, it is the most important function of the three. From the time that
utilities went beyond systems of one unit supplying a group of loads, effective
operation of the system required that critical quantities be measured and the
values of the measurements be transmitted to a central location. Such systems
of measurement and data transmission, called telemetry systems, have evolved
to schemes that can monitor voltages, currents, power flows, and the status of
circuit breakers, and switches in every substation in a power system transmission
network. In addition, other critical information such as frequency, generator
unit outputs and transformer tap positions can also be telemetered. With so
much information telemetered simultaneously, no human operator could hope
to check all of it in a reasonable time frame. For this reason, digital computers
are usually installed in operations control centers to gather the telemetered
data, process them, and place them in a data base from which operators can
display information on large display monitors. More importantly, the computer
can check incoming information against prestored limits and alarm the
operators in the event of an overload or out-of-limit voltage.
State estimation is often used in such systems to combine telemetered system
data with system models to produce the best estimate (in a statistical sense) of
the current power system conditions or “state.” We will discuss some of the
highlights of these techniques in Chapter 12.
Such systems are usually combined with supervisory control systems that
allow operators to control circuit breakers and disconnect switches and
transformer taps remotely. Together, these systems are often referred to as
SCADA systems, standing for supervisory control - and data
- acquisition system.
a
The SCADA system allows few operators to monitor the generation and
high-voltage transmission systems and to take action to correct overlords or
out-of-limit voltages.
The second major security function is contingency analysis. The results of
this type of analysis allow systems to be operated defensively. Many of the
problems that occur on a power system can cause serious trouble within such
a quick time period that the operator could not take action fast enough. This
is often the case with cascading failures. Because of this aspect of systems
operation, modern operations computers are equipped with contingency analysis
programs that model possible systems troubles before they arise. These
412 P O W E R SYSTEM S E C U R I T Y

programs are based on a model of the power system and are used to study
outage events and alarm the operators to any potential overlords or out-of-limit
voltages. For example, the simplest form of contingency analysis can be put
together with a standard power-flow program such as described in Chapter 4,
together with procedures to set up the power-flow data for each outage to be
studied by the power-flow program. Several variations of this type of contingency
analysis scheme involve fast solution methods, automatic contingency event
selection, and automatic initializing of the contingency power flows using actual
system data and state estimation procedures.
The third major security function is security-constrained optimal power flow.
In this function, a contingency analysis is combined with an optimal power
flow which seeks to make changes to the optimal dispatch of generation, as
well as other adjustments, so that when a security analysis is run, no
contingencies result in violations. To show how this can be done, we shall divide
the power system into four operating states.

0 Optimal dispatch this is the state that the power system is in prior to
any contingency. It is optimal with respect to economic operation, but it
may not be secure.
0 Post contingency: is the state of the power system after a contingency has
occurred. We shall assume here that this condition has a security violation
(line or transformer beyond its flow limit, or a bus voltage outside the
limit).
0 Secure dispatch: is the state of the system with no contingency outages,
but with corrections to the operating parameters to account for security
violations.
0 Secure post-contingency: is the state of the system when the contingency
is applied to the base-operating condition-with corrections.

We shall illustrate the above with an example. Suppose the trivial power system
consisting of two generators, a load, and a double circuit line, is to be operated
with both generators supplying the load as shown below (ignore losses):

OPTIMAL DISPATCH

We assume that the system as shown is in economic dispatch, that is the

500 MW from unit 1 and the 700 MW from unit 2 is the optimum dispatch.
Further, we assert that each circuit of the double circuit line can carry a
 INTRODUCTION 413

maximum of 400 MW, so that there is no loading problem in the base-operating

condition.
Now, we shall postulate that one of the two circuits making up the
transmission line has been opened because of a failure. This results in

Unit 1 nit 2

___t
500 M w (OVERLOAD) 1200 Mw
POST CONTINGENCY STATE

Now there is an overload on the remaining circuit. We shall assume for this
example that we do not want this condition to arise and that we will correct
the condition by lowering the generation on unit 1 to 400 MW. The secure
dispatch is

Unit 1 Unit 2

mm
1200 M W
SECURE DISPATCH

Now, if the same contingency analysis is done, the post-contingency condition is

Unit 1 Unit 2

1200 M W
SECURE POST CONTINGENCY STATE
By adjusting the generation on unit 1 and unit 2, we have prevented the
post-contingency operating state from having an overload. This is the essence
of what is called “security corrections.” Programs which can make control
adjustments to the base or pre-contingency operation to prevent violations in
the post-contingency conditions are called “security-constrained optimal power
flows” or SCOPF. These programs can take account of many contingencies
and calculate adjustments to generator MW, generator voltages, transformer
taps, interchange, etc. We shall show how the SCOPF is formed in Chapter 13.
414 POWER SYSTEM SECURITY

Together, the functions of system monitoring, contingency analysis, and

corrective action analysis comprise a very complex set of tools that can aid in
the secure operation of a power system. This chapter concentrates on contingency
analysis.

11.2 FACTORS AFFECTING POWER SYSTEM SECURITY

As a consequence of many widespread blackouts in interconnected power

systems, the priorities for operation of modern power systems have evolved to
the following.

0 Operate the system in such a way that power is delivered reliably.

0 Within the constraints placed on the system operation by reliability
considerations, the system will be operated most economically.

The greater part of this book is devoted to developing methods to operate

a power system to gain maximum economy. But what factors affect its operation
from a reliability standpoint? We will assume that the engineering groups who
have designed the power system’s transmission and generation systems have
done so with reliability in mind. This means that adequate generation has been
installed to meet the load and that adequate transmission has been installed
to deliver the generated power to the load. If the operation of the system went
on without sudden failures or without experiencing unanticipated operating
states, we would probably have no reliability problems. However, any piece of
equipment in the system can fail, either due to internal causes or due to external
causes such as lightning strikes, objects hitting transmission towers, or human
errors in setting relays. It is highly uneconomical, if not impossible, to build a
power system with so much redundancy (i.e., extra transmission lines, reserve
generation, etc.) that failures never cause load to be dropped on a system.
Rather, systems are designed so that the probability of dropping load is
acceptably small. Thus, most power systems are designed to have sufficient
redundancy to withstand all major failure events, but this does not guarantee
that the system will be 100% reliable.
Within the design and economic limitations, it is the job of the operators to
try to maximize the reliability of the system they have at any given time. Usually,
a power system is never operated with all equipment ‘‘in” (i-e., connected) since
failures occur or maintenance may require taking equipment out of service.
Thus, the operators play a considerable role in seeing that the system is
reliable.
In this chapter, we will not be concerned with all the events that can cause
trouble on a power system. Instead, we will concentrate on the possible
consequences and remedial actions required by two major types of failure
events-transmission-line outages and generation-unit failures.
Transmission-line failures cause changes in the flows and voltages on the
 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 415

transmission equipment remaining connected to the system. Therefore, the

analysis of transmission failures requires methods to predict these flows and
voltages so as to be sure they are within their respective limits. Generation
failures can also cause flows and voltages to change in the transmission system,
with the addition of dynamic problems involving system frequency and
generator output.

11.3 CONTINGENCY ANALYSIS DETECTION OF NETWORK

PROBLEMS

We will briefly illustrate the kind of problems we have been describing by use
of the six-bus network used in Chapter 4. The base-case power flow results for
Example 4A are shown in Figure 11.1 and indicate a flow of 43.8 MW and
60.7 MVAR on the line from bus 3 to bus 6. The limit on this line can be
expressed in MW or in MVA. For the purpose of this discussion, assume that
we are only interested in the MW loading on the line. Now let us ask what
will happen if the transmission line from bus 3 to bus 5 were to open. The
resulting flows and voltages are shown in Figure 11.2. Note that the flow on
the line from bus 3 to bus 6 has increased to 54.9 MW and that most of the
other transmission lines also experienced changes in flow. Note also that the
bus voltage magnitudes changed, particularly at bus 5 , which is now almost
5% below nominal. Figures 11.3 and 11.4 are examples of generator outages
and serve to illustrate the fact that generation outages can also result in changes
in flows and voltages on a transmission network. In the example shown in
Figure 11.3, all the generation lost from bus 3 is picked up on the generator
at bus 1. Figure 11.4 shows the case when the loss of generation on bus 3 is
made up by an increase in generation at buses 1 and 2. Clearly, the differences
in flows and voltages show that how the lost generation is picked up by the
remaining units is imporant.
If the system being modeled is part of a large interconnected network, the
lost generation will be picked up by a large number of generating units outside
the system’s immediate control area. When this happens, the pickup in
generation is seen as an increase in flow over the tie lines to the neighboring
systems. To model this, we can build a network model of our own system plus
an equivalent network of our neighbor’s system and place the swing bus or
reference bus in the equivalent system. A generator outage is then modeled so
that all lost generation is picked up on the swing bus, which then appears as
an increase on the tie flows, thus approximately modeling the generation loss
when interconnected. If, however, the system of interest is not interconnected,
then the loss of generation must be shown as a pickup in output on the other
generation units within the system. An approximate method of doing this is
shown in Section 11.3.2.
Operations personnel must know which line or generation outages will cause
flows or voltages to fall outside limits. To predict the effects of outages,
416 POWER SYSTEM SECURITY

246.1 kV a’
24 1

-
0-
*
33.1- I-
50.0
74.4
IA
-
+23.2
19.1
4 3 . 8 -+- 42.8
46.1 6

-
+12.4

c-27.8
26.2

15.5 +15.4
+16.0

--k 12.8 70

Bus 1
I
241.5kV La” --.f 231.0/-5.9’ 70
t
+15.4 15.0- Bus5

--t 35.6
--k 11.3
-
18.0 ---k

34.5
--j

-++- 9.7
1.6

-
-107.9
+26.1
+ 16.0 --f 43.6

Bus 4 I
-2.8
4.0

ait
.
I 70. 70.

-
1

- -
-42.5 226.7
+19.9 MW
-4.1
31.6 4.9 -+t MVAR
+45.1

I
227.6kV mo70.
FIG. 11.1 Six-bus network base case AC power flow (see Example 4A).

contingency analysis techniques are used. Contingency analysis procedures

model single failure events (i.e., one-line outage or one-generator outage) or
multiple equipment failure events (i.e., two transmission lines, one transmission
line plus one generator, etc.), one after another in sequence until “all credible
outages” have been studied. For each outage tested, the contingency analysis
procedure checks all lines and voltages in the network against their respective
limits. The simplest form of such a contingency analysis technique is shown in
Figure 11.5.
 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 417
Bus 3 246.1 kV mo
-60.0

241.5kV k3.4' -5.1 5.1 C-

-
-50
+91.2
37.3
-49.1
- 54.9 ----F 53.6
6

.--t5.9
-+ 22.9 -22.3
+ 16.2 --tr19.9

Bus 1 I I
I1.5kV fQo
26.8
----t
C+ 14.6
- 19.9
Bus 5

+-k 15.3
229.8kV k5.2'

0- +21.7 -37.0
*22.0
108.5
U
I+- 29.8

Bus 4
-7.2
7.4
~21 t70 70

-
219.3kV U"

42.1

-
22.3
I
+7.6 -I+- 0.0
35.5
-47.6

226.4kV /-4.1" 70 70

FIG. 11.2 Six-bus network line outage case; line from bus 3 to bus 5 opened

The most difficult methodological problem to cope with in contingency

analysis is the speed of solution of the model used. The most difficult logical
problem is the selection of "all credible outages." If each outage case studied
were to solve in 1 sec and several thousand outages were of concern, it would
take close to 1 h before all cases could be reported. This would be useful if the
system conditions did not change over that period of time. However, power
418 POWER SYSTEM SECURITY
Bus 3
Bus 2
24 1.5kV L 2 J I - O -21.2
+28.6
-20
+31
J*
221.9kV &QJo

+-- 0.6
c>--
+ 187.9
-6.1

-21.1 +20.8
Bus 6

+25.2 +25.8
1 -20.0
-57.6
-42.9
++ 44.2
-40.3

Bus 1
- 52.4
28.1
s

241.5kV La" I
--f 55.7
-+I- 25.9

55.2
17.1
29'7 -
--t

-52.5
Bus 5

-8.9
-++- 2.6
214.7kV ElQJ."

Q01-75. -18.9 *
-0.1 10.5
-I+ 16.6
-64.1
-10.0
Bus 4
216.8kV uo

56.2

226.5kV L = U 0
FIG. 11.3 Six-bus network generator outage case. Outage of generator on bus 3; lost
generation picked up on generator 1.

systems are constantly undergoing changes and the operators usually need to
know if the present operation of the system is safe, without waiting too long
for the answer. Contingency analysis execution times of less than 1 min for
several thousand outage cases are typical of computer and analytical technology
as of 1995.
One way to gain speed of solution in a contingency analysis procedure is to
24 1
‘5kV
Bus 2
-24.6
-27.7
-23.9
*30.3
J*
222.0kV

-3.1
+4.6
mo

-
-110.5
-159.1 Bus 6
--F 20.8 -20.6
39.6 *25.7J -5.5
+-t49.8
--F 46.7
-I-+ 43.0
-44.0
+40.1
- -C-r- 3.6

241.5kV
--
25.8
4 14.2 +23.4
-27.3
Bus 5
-5.5
214.7kV mo

+1.6
- -41.4 39.6
’-15-:0.
+34.3
-
+24.9
43.3
-24‘0 d 3 . 0

+37.7
++-
48.1
--+9.8
*1.2
-42.1
ft 23.1

ali 70 70

FIG. 11.4 Six-bus network generator outage case. Outage of generator on bus 3; lost
generation picked up on generator 1 and generator 2.

use an approximate model of the power system. For many systems, the use of
DC load flow models provides adequate capability. In such systems, the voltage
magnitudes may not be of great concern and the DC load flow provides
sufficient accuracy with respect to the megawatt flows. For other systems,
voltage is a concern and full AC load flow analysis is required.
420 POWER SYSTEM SECURITY

'r'
START

I SET SYSTEM MODEL

TO INITIAL CONDITIONS I
b i =1

OUTAGE OF GENERATOR i

- DISPLAY
. EXCEED LIMIT
~

ALARM MESSAGE
I

!
SIMULATE AN OUTAGE
OF LINE 2' USING THE
SYSTEM MODEL

J ~~

FIG. 11.5 Contingency analysis procedure.

 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 421

11.3.1 An Overview of Security Analysis

A security analysis study which is run in an operations center must be executed
very quickly in order to be of any use to operators. There are three basic ways
to accomplish this.

0 Study the power system with approximate but very fast algorithms.
0 Select only the important cases for detailed analysis.
0 Use a computer system made up of multiple processors or vector
processors to gain speed.

The first method has been in use for many years and goes under various names
such as “ D factor methods,” “linear sensitivity methods,” “DC power flow
methods,” etc. This approach is useful if one only desires an approximate
analysis of the effect of each outage. This text presents these methods under
the name linear sensitivity factors and uses the same derivation as was presented
in Chapter 4 under the D C power flow methods. It has all the limitations
attributed to the DC power flow; that is, only branch MW flows are calculated
and these are only within about 5% accuracy. There is no knowledge of MVAR
flows or bus voltage magnitudes. Linear sensitivity factors are presented in
Section 1 1.3.2.
If it is necessary to know a power system’s MVA flows and bus voltage
magnitudes after a contingency outage, then some form of complete A C power
flow must be used. This presents a great deal of difficulty when thousands of
cases must be checked. It is simply impossible, even on the fastest processors
in existence today (1995) to execute thousands of complete AC power flows
quickly enough. Fortunately, this need not be done as most of the cases result
in power flow results which do not have flow or voltage limit violations. What
is needed are ways to eliminate all or most of the nonviolation cases and only
run complete power flows on the “critical” cases. These techniques go under
the names of “contingency selection” or “contingency screening” and are
introduced in Section 1 1.3.4.
Last of all, it must be mentioned that there are ways of running thousands
of contingency power flows if special computing facilities are used. These
facilities involve the use of many processors running separate cases in parallel,
or vector processors which achieve parallel operation by “unwinding” the
looping instruction sets in the computer code used. As of the writing of this
edition (1 995), such techniques are still in the research stage.

11.3.2 Linear Sensitivity Factors

The problem of studying thousands of possible outages becomes very difficult
to solve if it is desired to present the results quickly. One of the easiest ways
to provide a quick calculation of possible overloads is to use linear sensitivity
factors. These factors show the approximate change in line flows for changes
422 POWER SYSTEM SECURITY

in generation on the network configuration and are derived from the DC load
flow presented in Chapter 4.These factors can be derived in a variety of ways
and basically come down to two types:

1. Generation shift factors.

2. Line outage distribution factors.

Here, we shall describe how these factors are used. The derivation of sensitivity
factors is given in Appendix 11A.
The generation shift factors are designated a,i and have the following
definition:
(11.1)

where
G = line index
i = bus index
Af/ = change in megawatt power flow on line e when a change in
generation, A e , occurs at bus i
AP, = change in generation at bus i

It is assumed in this definition that the change in generation, APi, is exactly

compensated by an opposite change in generation at the reference bus, and that
all other generators remain fixed. The a,, factor then represents the sensitivity
of the flow on line t to a change in generation at bus i. Suppose one wanted
to study the outage of a large generating unit and it was assumed that all the
generation lost would be made up by the reference generation (we will deal
with the case where the generation is picked up by many machines shortly). If
the generator in question was generating Po MW and it was lost, we would
represent AP, as

Api= -Po (11.2)

and the new power flow on each line in the network could be calculated using
a precalculated set of “ a ” factors as follows:

f,,= f,“ + a , , A e for t = 1 . . . L (11.3)

where
f, = flow on line e after the generator on bus i fails
f /“ = flow before the failure
The “outage flow,” f;, on each line can be compared to its limit and those
exceeding their limit flagged for alarming. This would tell the operations
 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 423

personnel that the loss of the generator on bus i would result in an overload
on line L.
The generation shift sensitivity factors are linear estimates of the change in
flow with a change in power at a bus. Therefore, the effects of simultaneous
changes on several generating buses can be calculated using superposition.
Suppose, for example, that the loss of the generator on bus i were compensated
by governor action on machines throughout the interconnected system. One
frequently used method assumes that the remaining generators pick up in
proportion to their maximum MW rating. Thus, the proportion of generation
pickup from unit j ( j # i) would be
pmax
y..= 'j (11.4)
c PYX
kfi
where

Pya' = maximum MW rating for generator k

yji = proportionality factor for pickup on generating unit j when unit i fails

Then, to test for the flow on line L, under the assumption that all the generators
in the interconnection participate in making up the loss, use the following:

f; = f /" + a,,Ae - j1
#i
[afjyjiApl:] (11.5)

Note that this assumes that no unit will actually hit its maximum. If this
is apt to be the case, a more detailed generation pickup algorithm that took
account of generation limits would be required.
The line outage distribution factors are used in a similar manner, only they
apply to the testing for overloads when transmission circuits are lost. By
definition, the line outage distribution factor has the following meaning:

(11.6)
where

d ( , k = line outage distribution factor when monitoring line L after an

outage on line k
Af/ = change in MW flow on line L
f , " = original flow on line k before it was outaged (opened)

If one knows the power on line G and line k, the flow on line L with line k out
can be determined using " d " factors.

(11.7)
 424 POWER SYSTEM SECURITY

where
f /",f,"= preoutage flows on lines L and k , respectively
f f= flow on line L with line k out

By precalculating the line outage distribution factors, a very fast procedure can
be set up to test all lines in the network for overload for the outage of a
particular line. Furthermore, this procedure can be repeated for the outage of
each line in turn, with overloads reported to the operations personnel in the
form of alarm messages.
Using the generator and line outage procedures described earlier, one can
program a digital computer to execute a contingency analysis study of the power
system as shown in Figure 11.6. Note that a line flow can be positive or negative
so that, as shown in Figure 11.6, we must check f against -fYx as well as
f y x This
. figure makes several assumptions; first, it assumes that the generator
output for each of the generators in the system is available and that the line
' flow for each transmission line in the network is also available. Second, it
assumes that the sensitivity factors have been calculated and stored, and that
they are correct. The assumption that the generation and line flow MWs are
available can be satisfied with telemetry systems or with state estimation
techniques. The assumption that the sensitivity factors are correct is valid as
long as the transmission network has not undergone any significant switching
operations that would change its structure. For this reason, control systems
that use sensitivity factors must have provision for updating the factors when
the network is switched. A third assumption is that all generation pickup will
be made on the reference bus. If this is not the case, substitute Eq. 11.5 in the
generator outage loop.

EXAMPLE 11A

The [ X I matrix for our six-bus sample network is shown in Figure 11.7,
together with the generation shift distribution factors and the line outage
distribution factors.
The generation shift distribution factors that give the fraction of generation
shift that is picked up on a transmission line are designated a t i . The a factor
is obtained by finding line P along the rows and then finding the generator to
be shifted along the columns. For instance, the shift factor for a change in the
flow on line 1-4 when making a shift in generation on bus 3 is found in the
second row, third column.
The line outage distribution factors are stored such that each row and
column corresponds to one line in the network. The distribution factor dt,k is
obtained by finding line t along the rows and then finding line k along that
row in the appropriate column. For instance, the line outage distribution factor
that gives the fraction of flow picked up on line 3-5 for an outage on line 3-6
 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 425

P: AT EACH
GEN.BUS
:f ON A L L
I
\

4 READ EXISTING
SYSTEM CONDITIONS
1

CHECK A L L LINES
FOR OVERLOAD

CHECK A L L LINES
FOR OVER LOAD
AFTER LINE
OUTAGES

FIG. 11.6 Contingency analysis using sensitivity factors.

 X Matrix for Six-BusSample System (Reference at Bus 1) Generation Shift Factors For Six-Bus Sample System
0 0 0 0 0 Bus 1 Bus 2 Bus 3
0.09412 0.08051 0.06298 0.06435 0.08129 C = 1 (line 1-2) 0 - 0.47 - 0.40
0.08051 0.16590 0.05897 0.09077 0.12895 f = 2 (line 1-4) 0 -0.31 - 0.29
0.06298 0.05897 0.10088 0.05422 0.05920 f = 3 (line 1-5) 0 -0.21 -0.30
0 0.06435 0.09077 0.05422 0.12215 0.08927 f = 4 (line 2-3) 0 0.05 -0.34
0 0.08129 0.12895 0.05920 0.08927 0.16328 / = 5 (line 2-4) 0 0.31 0.22
C = 6 (line 2-5) 0 0.10 - 0.03
C = 7 (line 2-6) 0 0.06 -0.24
C = 8 (line 3-5) 0 0.06 0.29
C = 9 (line 3-6) 0 -0.01 0.37
C = I0 (line 4-5) 0 0 -0.08
f = 11 (line 5-6) 0 -0.06 -0.13

Line Outage Distribution Factors lor Six-Bus Sample System

k=8 k=lO k=ll

i
k=l k=2 k=3 k=4 k=S
,Line 1-2) (Line 1-4) (Line 1-5) (Line 2-3) (Line 2-4) (Line 3-5)
C = I (line 1-2) 0.64 0.54 -0.11 -0.50 -0.2 I -0.12 -0.14
t = 2 (line 1-4) 0.59 0.46 - 0.03 0.61 -0.06 - 0.04 - 0.33
C=3 (line 1-5) 0.41 0.36 0.15 -0.11 0.18 - 0.02 -0.17
C =4 (line 2-3) -0.10 -0.03 0.18 0.12 -0.40 -0.53 0.17
C=5 (line2-4) - 0.59 0.76 -0.17 0.16 0.30 0.19 -0.02 -0.67 -0.19
C=6 (line 2-5) -0.19 -0.06 0.33 0.22 0.23 0.27 -0.03 -0.26
C = 7 (line 2-6) -0.12 -0.04 0.21 0.51 0.15 0.27 - 0.20 0.58 0.20
C = 8 (line 3-5) -0.12 - 0.04 0.20 -0.38 0.14 0.27 -0.17 0.47 0.19 -0.42
C = 9 (line 3-6) 0.01 0 - 0.03 - 0.62 -0.02 - 0.03 0.64 0.60 -0.02
C = 10 (line 4-5) 0.01 - 0.24 0.29 0.13 -0.39 0.24 0.14 0.15 - 0.02 -0.15
C = I I (line 5-6) 0.11 0.03 -0.18 0.12 -0.13 -0.23 0.36 - 0.40 0.42 -0.18

FIG. 11.7 Outage factors for a six-bus system.

 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 427

is found in the eighth row and ninth column. Figure 11.3 shows an outage of
the generator on bus 3 with all pickup of lost generation coming on the
generator at bus 1. To calculate the flow on line 1-4 after the outage of the
generator on bus 3, we need (see Figure 11.1):

Base-case flow on line 1-4 = 43.6 MW

Base-case generation on bus 3 = 60 MW
Generation shift distribution factor = a1-4,3= -0.29

Then the flow on line 1-4 after generator outage is = base-case flow1-, +
al-4,3APgen3= 43.6 + (-0.29)(-60 MW) = 61 MW.
To show how the line outage and generation shift factors are used, calculate
some flows for the outages shown in Figures 11.2 and 11.3. Figure 11.2 shows
an outage of line 3-5. If we wish to calculate the power flowing on line 3-6 with
line 3-5 opened, we would need the following.

Base-case flow on line 3-5 = 19.1 MW

Base-case flow on line 3-6 = 43.8 MW
Line outage distribution factor: d,-,.,-, = 0.60

Then the flow on 3-6 after the outage is = base flow,-, + d,-,,,-, x base
flow,*, = 43.8 + (0.60) x (19.1) = 55.26 MW.
In both outage cases, the flows calculated by the sensitivity methods are
reasonably close to the values calculated by the full AC load flows as shown
in Figures 1 1.2 and 1 1.3.

11.3.3 AC Power Flow Methods

The calculations made by network sensitivity methods are faster than those
made by AC power flow methods and therefore find wide use in operations
control systems. However, there are many power systems where voltage
magnitudes are the critical factor in assessing contingencies. In addition, there
are some systems where VAR flows predominate on some circuits, such as
underground cables, and an analysis of only the MW flows will not be adequate
to indicate overloads. When such situations present themselves, the network
sensitivity methods may not be adequate and the operations control system
will have to incorporate a full AC power flow for contingency analysis.
When an AC power flow is to be used to study each contingency case, the
speed of solution and the number of cases to be studied are critical. To repeat
what was said before, if the contingency alarms come too late for operators to
act, they are worthless. Most operations control centers that use an AC power
flow program for contingency analysis use either a Newton-Raphson or the
decoupled power flow. These solution algorithms are used because of their
428 POWER SYSTEM SECURITY

speed of solution and the fact that they are reasonably reliable in convergence
when solving difficult cases. The decoupled load flow has the further advantage
that a matrix alteration formula can be incorporated into it to simulate the
outage of transmission lines without reinverting the system Jacobian matrix at
each iteration.
The simplest A C security analysis procedure consists of running an A C
power flow analysis for each possible generator, transmission line, and trans-
former outage as shown in Figure 11.8. This procedure will determine the
overloads and voltage limit violations accurately (at least within the accuracy
of the power flow program, the accuracy of the model data, and the accuracy
with which we have obtained the initial conditions for the power flow). It does
suffer a major drawback, however, and that concerns the time such a program
takes to execute. If the list of outages has several thousand entries, then the
total time to test for all of the outages can be too long.
We are thus confronted with a dilemma. Fast, but inaccurate, methods
involving the a and d factors can be used to give rapid analysis of the system,
but they cannot give information about M V A R flows and voltages. Slower, full
A C power flow methods give full accuracy but take too long.

+
Pick outage i from the list and remove
that component from the power flow
model

I)
Run an AC Power Flow on the
current model updated to reflect
the outage

+
Test for overloads and voltage
Alarm List
limit violations. Report all
limit violations in an alarm
list.

+
Last outage done?
Yes
No

I i=i+l
End

FIG. 11.8 AC power flow security analysis.

 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 429

Fortunately, there is a way out of this dilemma. Because of the way the
power system is designed and operated, very few of the outages will actually
cause trouble. That is, most of the time spent running A C power flows will go
for solutions of the power flow model that discover that there are no problems.
Only a few of the power flow solutions will, in fact, conclude that an overload
or voltage violation exists.
The solution to this dilemma is to find a way to select contingencies in such
a way that only those that are likely to result in an overload or voltage limit
violation will actually be studied in detail and the other cases will go
unanalyzed. A flowchart for a process like this appears in Figure 11.9. Selecting

Select the bad cases from the full

case list and store in a short list

Short List of most-

likely bad cases

-
List of Possible
Outages
i-1 is1

Pick outage i from the short list and

1
remove that component from the power
flow model
I

+
Run an AC Power Flow on the
I

current model updated to reflect

the outage

+
Test for overloads and voltage
Alarm List
limit violations. Report all
limit violations in an alarm
list.

Last outage done?

i=i+l
I +
End

FIG. 11.9 AC power flow security analysis with contingency case selection.
430 POWER SYSTEM SECURITY

the bad or likely trouble cases from the full outage case list is not an exact
procedure and has been the subject of intense research for the past 15 years.
Two sources of error can arise.
1. Placing too many cases on the short list: this is essentially the “conservative”
approach and simply leads to longer run times for the security analysis
procedure to execute.
2. Skipping cases: here, a case that would have shown a problem is not placed
on the short list and results in possibly having that outage take place and
cause trouble without the operators being warned.

11.3.4 Contingency Selection

We would like to get some measure as to how much a particular outage might
affect the power system. The idea of a performance index seems to fulfill this
need. The definition for the overload performance index (PI) is as follows:
2n
(11.8)
a11 branches
I

If n is a large number, the PI will be a small number if all flows are within
limit, and it will be large if one or more lines are overloaded. The problem then
is how to use this performance index.
Various techniques have been tried to obtain the value of PI when a branch
is taken out. These calculations can be made exactly if n = 1; that is, a table
of PI values, one for each line in the network, can be calculated quite quickly.
The selection procedure then involves ordering the PI table from largest value to
least. The lines corresponding to the top of the list are then the candidates for
the short list. One procedure simply ordered the PI table and then picked the
top N, entries from this list and placed them on the short list (see reference 8).
However when n = 1, the PI does not snap from near zero to near infinity
as the branch exceeds its limit. Instead, it rises as a quadratic function. A line
that is just below its limit contributes to PI almost equal to one that is just
over its limit. The result is a PI that may be large when many lines are loaded
just below their limit. Thus the PI’S ability to distinguish or detect bad cases
is limited when ti = 1. Ordering the PI values when n = 1 usually results in a
list that is not at all representative of one with the truly bad cases at the top.
Trying to develop an algorithm that can quickly calculate PI when n = 2 or
larger has proven extremely difficult.
One way to perform an outage case selection is to perform what has been
references 9 and 10). Here, a decoupled power flow
called the I P l Q tiic~t/~od(see
is used. As shown in Figure 11.10, the solution procedure is interrupted after
one iteration (one P - c) calculation and one Q - Vcalculation; thus, the name
1 P1 Q). With this procedure, the PI can use as large an n value as desired, say
n = 5. There appears to be sufficient information in the solution at the end of
 C O N T I N G E N C Y ANALYSIS: D E T E C T I O N OF N E T W O R K PROBLEMS 431

.
Begin power flow solution

Build B' and B" matrices

Full outage
case list
J===== L
Model outage case
I

Solve the P-theta equation for

the AB's

Solve the Q-V equation for the

AIEI'S

Calculate flows and voltages for this case .-

then calculate the PI

- Pick next outage case

FIG. 11.10 The 1 P1Q contingency selection procedure.

PI List
(one entry for
each outage
case)

the first iteration of the decoupled power flow to give a reasonable PI. Another
advantage to this procedure is the fact that the voltages can also be included
in the PI. Thus, a different PI can be used, such as:

n PIz = ) c ' ( + c ( AIEiI )zm

(11.9)
all branches all buses AIE("""
i i
432 POWER SYSTEM SECURITY

where AIEil is the difference between the voltage magnitude as solved at the
end of the l P l Q procedure and the base-case voltage magnitude. A/(E("""is a
value set by utility engineers indicating how much they wish to limit a bus
voltage from changing on one outage case.
To complete the security analysis, the PI list is sorted so that the largest PI
appears at the top. The security analysis can then start by executing full power
flows with the case which is at the top of the list, then solve the case which is
second, and so on down the list. This continues until either a fixed number of
cases is solved, or until a predetermined number of cases are solved which do
not have any alarms.

11.3.5 Concentric Relaxation

Another idea to enter the field of security analysis in power systems is that an
outage only has a limited geographical effect. The loss of a transmission line
does not cause much effect a thousand miles away; in fact, we might hope that
it doesn't cause much trouble beyond 20 miles from the outage, although if the
line were a heavily loaded, high-voltage line, its loss will most likely be felt
more than 20 miles away.
To realize any benefit from the limited geographical effect of an outage, the
power system must be divided into two parts: the affected part and the part
that is unaffected. To make this division, the buses at the end of the outaged
line are marked as layer zero. The buses that are one transmission line or
transformer from layer zero are then labeled layer one. This same process can
be carried out, layer by layer, until all the buses in the entire network are
included. Some arbitrary number of layers is chosen and all buses included in
that layer and lower-numbered layers are solved as a power flow with the outage
in place. The buses in the higher-numbered layers are kept as constant voltage
and phase angle (i.e., as reference buses).
This procedure can be used in two ways: either the solution of the layers
included becomes the final solution of that case and all overloads and voltage
violations are determined from this power flow, or the solution simply is used
to form a performance index for that outage. Figure 11.11 illustrates this
layering procedure.

FIG. 11.11 Layering of outage effects.

 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 433

The concentric relaxation procedure was originally proposed by Zaborsky

(see reference 13). The trouble with the concentric relaxation technique is that
it requires more layers for circuits whose influence is felt further from the
outage.

11.3.6 Bounding
A paper by Brandwajn (reference 11) solves at least one of the problems in
using the concentric relaxation method. Namely, it uses an adjustable region
around the outage to solve for the outage case overloads. In reference 11, this
is applied only to the linear (DC) power flow; it has subsequently been extended
for AC network analysis.
To perform the analysis in the bounding technique we define three subsystems
of the power system as follows:

N1 = the subsystem immediately surrounding the outaged line

N2 = the external subsystem that we shall not solve in detail
N3 = the set of boundary buses that separate N1 and N2

The subsystems appear as shown in Figure 11.12. The bounding method is

based on the fact that we can make certain assumptions about the phase angle
spread across the lines in N2, given the injections in N1 and the maximum
phase angle appearing across any two buses in N3. In Appendix 11A of this
chapter we show how to calculate the APk and the AP,,, injections that will
make the phase angles on buses k and rn simulate the outage of line
k-m.
If we are given a transmission line in N2 with flow f:q, then there is a
maximum amount that the flow on p q can shift. That is, it can increase from

FIG. 11.12 Layers used in bounding analysis.

434 POWER SYSTEM SECURITY

f;, to its upper limit or it can decrease to its lower limit. Then,
(11.10)

Further, we can translate this into a maximum change in phase angle difference
as follows:
(1 1.11)
or
1
Afpq = - (AO, - AO,) (11.12)
XP4
and finally:
(AO, - = Af PmqaxxPq (11.13)

Thus, we can define the maximum change in the phase angle difference across
p q . Reference 11 develops the theorem that:

IA8, - A8,I < lA8i - A0jI (1 1.14)

where i and j are any pair of buses in N3, Adi is the largest A 8 in N3, and Adj
is the smallest A 8 in N3 (see Appendix 11B).
Equation 11.14 is interpreted as follows: the right-hand side, IA8, - ABj(,
provides an upper limit to the maximum change in angular spread across any
circuit in N2. Thus, it provides us with a limit as to how far any of the N2
circuits can change their flow. By combining Eqs. 11.13 and 11.14 we obtain:

Af ;;'xPq < lA8i - A8jI (11.15)

Figure 11.13 shows a graphical interpretation of the bounding process. There

are two cases represented in Figure 11.13: a circuit on the top of the figure that

FIG. 11.13 Interpretation of bounding.

 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 435

cannot go over limit, while that on the bottom could. In each case, the
horizontal line represents the change in flow on circuit pq times its reactance,
Afp4xpq;the dotted line, labeled Af2;'xPq, represents the point where circuit p q
will go into overload and is determined as explained previously. Any value of'
Afpqxgq to the right of the dotted line represents an overload.
The solid line labeled lAOi - AO,j represents the upper limit on Afp,xp,.
Thus, if the solid line is below (to the left) of the dotted line, then the
circuit theory upper limit predicts that the circuit cannot go into overload;
if on the other hand, the solid line is above (to the right of) the dotted
line, the circuit may be shifted in flow due to the outage so as to violate
a limit.
A completely safe N2 region would be one in which the maximum I AOi - AB,I
upper limit is small enough to be less than all of the AfF;'xpq limits. In fact,
as the N1 region is enlarged, the value of \bei- AOjj will become smaller and
smaller. Therefore, the test to determine whether the N1 region encompasses
all possible overloaded circuits should be as follows:

All circuits in N2 are safe from overload if the value of lAOi - AO,/ is less
than the smallest value of Af ;;'x,, over all pairs p q , where pq corresponds
to the buses.at the ends of circuits in N2

If this condition fails, then we have to expand N1, calculate a new lAOi - AOjl
in N3, and rerun the test over the newly defined N2 region circuits. When an
N2 is found which passes the test, we are done and only region N1 need be
studied in detail.
References 10 and 12 extend this concept to screening for AC contingency
effects. Such contingency selection/screening techniques form the foundation
for many real-time computer security analysis algorithms.

EXAMPLE 11B

In this example, we shall take the six-bus sample system used previously and
show how the bounding technique works so that not all of the circuits in the
system need be analyzed. Note that this is a small system so that the net savings
in computer time may not be that great. Nonetheless, it demonstrates the
principles used in the bounding technique quite well.
We shall study the outage of transmission line 3-6. The DC power flow will
be used throughout and the initial conditions will be those shown in Figure
4.12. The MW limits on the transmission lines are shown in the table
at the top of the next page.
 436 POWER SYSTEM SECURITY

Line M W Limit
1-2 30
1-4 50
1-5 40
2-3 20
2-4 40
2-5 20
2-6 30
3-5 20
3-6 60
4-5 20
5-6 20

In this example, we shall proceed in steps. Step A will analyze the system as if
the N1 and N3 regions consist of only line 3-6 itself, as shown in Figure 11.14.
If the bounding criteria is met, no other analysis need be done as it will establish
that no overloads exist anywhere in the system. If the bounding criteria fails,
we still proceed to step B. Step B expands the bounded region from line 3-6 to
include all buses which are once removed from buses 3 and 6; that is, it includes
buses 2, 3, 5, and 6 as shown in Figure 11.15, and in this case the boundary of
the region, N3, consists of buses 2 and 5.
To start, we need to calculate Af Pmqax and then Af :''xpq as given in Eqs.
11.10 through 11.13. These values are given below where the flows and flow
limits are all converted to per unit on a 100 MVA base. (The line reactances
are found in the appendix to Chapter 4.)

MW Limit fL
Line (per unit) (per unit) Af ",:.; XPq Af;;xxpq
1-2 0.30 0.253 0.047 0.20 0.0094
1-4 0.50 0.416 0.084 0.20 0.0168
1-5 0.40 0.331 0.069 0.30 0.0207
2-3 0.20 0.018 0.182 0.25 0.0455
2-4 0.40 0.325 0.075 0.10 0.0075
2-5 0.20 0.162 0.038 0.30 0.01 14
2-6 0.30 0.248 0.052 0.20 0.0104
3-5 0.20 0.169 0.03 1 0.26 0.00806
3-6 0.60 0.449 - - -
4-5 0.20 0.04 1 0.159 0.40 0.0636
5-6 0.20 0.003 0.197 0.30 0.059 1

For step A, we use Eq. 11A.13 from Appendix 11A to calculate 6 3 , 3 6 and 6 6 , 3 , as

BLOG FIEE http://fiee.zoomblog.com

 CONTINGENCY ANALYSIS: DETECTION OF NETWORK PROBLEMS 437
,.-.-.-.-.-. -.-. -.-.-.-.-.-.-.-.-.-. -.--.-.-. -.-.- .-.-
*\

I
Bus I
i

6
32.5-
50.0

I
i\. ,
+44.9 44.9 -# I
I I
i
i
I

+24.8 \.-. -.-. -.-.-. *. 24.8-

-0.3 I
\ I I
1

-
d16.2
25.3
i !
f
1 I
i
I

- - 25.3
-16.2
Bus 5

-0.3
I
I
i
100 M W 433.1 -33.1 I

c k -41.6

70.0 i
I
I
i
I

Bus 4 I

-- 41
4.
where - MW
I
i
I

-
I
I
generator
32.5
!
I
I
'.
1
1 70.0 %.-.-.-.-.-. load
I
i
.-.-.-.-.-.-.-.-. -.-.-.-.-.-.-.-.-.-.-
FIG. 11.14 Step A of Example 11B.

BLOG FIEE http://fiee.zoomblog.com

 438 POWER SYSTEM SECURITY

!
!
I
I
I
i
I
I
I
i
I
I
I
I
I
I
- -
I
!
I
I
-
1 0 0 MW +33.1
25.3
0.3

I -41.6
I
I
i
I
Bus 4
I
i N2
!
I
I
where - MW

I
I 0-i generator

I
!
'.- -.-.-.-.-.-.-.-.-.-.-.-.70.0 -.-.-. - .-.-.- h-
.-.-.-. load
.- .--.-.-.'
FIG. 11.15 Step B of Example 11B.
 CALCULATlON OF NETWORK SENSITIVITY FACTORS 439

shown below.

Then using Eq. 1 1A. 1 1

lAO3 - A061 = 0.111437

According to the criterion in Eq. 11.14, the value lAOi - AOjl must be less that
the smallest value of lAOp - AOq/ which equals Af pmxxpq and is found in the
table above to be at line 2-4. Since lA03 - A061 = 0.111437 and the minimum
lAOi - AO,I is lA02 - A041 which has a value of 0.0075, the criteria fails. We
must proceed to step B.
Step B requires that we calculate IAOi - AOjI for buses 2 and 5. This value
is 0.003564 and the bounding criteria is satisfied.
If we had used the d factors for the six-bus system as shown in Example
1 lA, we could simply find all the line flows for the 3-6 outage as shown in the
table below.

M W Limit f p”,
Line (per unit) (per unit) f ;i6 Out

1-2 0.30 0.253 0.251

1-4 0.50 0.4 16 0.416
1-5 0.40 0.33 1 0.322
2-3 0.20 0.018 -0.220 overload
2-4 0.40 0.325 0.316
2-5 0.20 0.162 0.148
2-6 0.30 0.248 0.508 overload
3-5 0.20 0.169 0.380 overload
3-6 0.60 0.449 -
4-5 0.20 0.04 1 0.320
5-6 0.20 0.003 0.191

Note that three overloads exist on lines 2-3, 2-6, and 3-5, which are all within
the bounded region N1 + N 3 in Figure 11.15.

APPENDIX 11A
Calculation of Network Sensitivity Factors

First, we show how to derive the generation-shift sensitivity factors. To start,

repeat Eq. 4.36.
e = [XIP (llA.l)
440 POWER SYSTEM SECURITY

This is the standard matrix calculation for the DC load flow. Since the DC
power-flow model is a linear model, we may calculate perturbations about a
given set of system conditions by use of the same model. Thus, if we are
interested in the changes in bus phase angles, 88, for a given set of changes in
the bus power injections, AP, we can use the following calculation.

A0 = [X]AP ( 1 1A.2)

In Eq. l l A . l , it is assumed that the power on the swing bus is equal to the
sum of the injections of all the other buses. Similarly, the net perturbation
of the swing bus in Eq. l l A . 2 is the sum of the perturbations on all the other
buses.
Suppose that we are interested in calculating the generation shift sensitivity
factors for the generator on bus i. To d o this, we will set the perturbation
on bus i to + 1 and the perturbation on all the other buses to zero. We can
then solve for the change in bus phase angles using the matrix calculation in
Eq. 11A.3.

(1 1A.3)

The vector of bus power injection perturbations in Eq. l l A . 3 represents the

situation when a 1 pu power increase is made at bus i and is compensated by
a 1 pu decrease in power at the reference bus. The A8 values in Eq. l l A . 3 are
thus equal to the derivative of the bus angles with respect to a change in power
injection at bus i. Then, the required sensitivity factors are

(1 1 A.4)

where
d8
Xni = "- = nth element from the A0 vector in Eq. 11A.3
dc

Xmi = dem
~ = mth element from the A0 vector in Eq. 11A.3
dP;:
x/ = line reactance for line 8

A line outage may be modeled by adding two power injections to a system,

one at each end of the line to be dropped. The line is actually left in the system
and the effects of its being dropped are modeled by injections. Suppose line k
 CALCULATION OF NETWORK SENSITIVITY FACTORS 441

BUS n BUS rn

LINES TO
REMAINDER
OF NETWORK
' LINES TO
REMAINDER
OF NETWORK

LINE k
+ LINE k BEFORE
P"rn OUTAGE

I
\I

FIG. 11.16 Line outage modeling using injections.

from bus n to bus m were opened by circuit breakers as shown in Figure 11.16.
Note that when the circuit breakers are opened, no current flows through them
and the line is completely isolated from the remainder of the network. In the
bottom part of Figure 11.16, the breakers are still closed but injections APn and
AP, have been added to bus n and bus rn, respectively. If APn = F,,,,,, where Fnm
is equal to the power flowing over the line, and AP, = -prim, we will still
have no current flowing through the circuit breakers even though they are
closed. As far as the remainder of the network is concerned, the line is
disconnected.
Using Eq. l l A . 2 relating to A0 and AP, we have

A0 = [XlAP
442 POWER SYSTEM SECURITY

where

so that
A0, = X,,AP, + XnmAPm (llA.5)
A0, = Xm,AP, + XmmAP,

define

On, Om, Pnm to exist before the outage, where P,, is the flow on line k
from bus n to bus rn
A0,, A0,, AP,, to be the incremental changes resulting from the outage
e n9 8, > R m to exist after the outage

The outage modeling criteria requires that the incremental injections AP,
and APmequal the power flowing over the outaged line after the injections are
imposed. Then, if we let the line reactance be x k
-
P,, = AP, = - AP, (1 1 A.6)
where
- 1 -
prim = - (en- 0),
xk

then

(11A.7)

and
8, = 0, + A0,
- (llA.8)
0, = 0, + Atlm
giving
- 1 -
P,, = - (6, - 0,)
- 1
= - (0, - 0,) + -1 (Adn - A0,)
xk xk xk
 CALCULATION OF NETWORK SENSITIVITY FACTORS 443

Then (using the fact that Fnmis set to AP,)

1
1
APn = (llA.lO)

prim

Define a sensitivity factor 6 as the ratio of the change in phase angle 8, anywhere
in the system, to the original power P,,, flowing over a line n m before it was
dropped. That is,

(1 1A. 11)

If neither n or m is the system reference bus, two injections, AP,, and APm, are
imposed at buses n and m, respectively. This gives a change in phase angle at
bus i equal to
Aei = XinAPn+ Xi,APm ( 1 1A. 12)

Then using the relationship between APn and APm, the resulting 6 factor is

(1 1A. 13)

If either n or m is the reference bus, only one injection is made. The resulting
S factors are

If bus i itself is the reference bus, then 6 i , n m= 0 since the reference bus angle is
constant.
The expression for d [ , k is

(llA.15)
444 POWER SYSTEM SECURITY

if neither i nor j is a reference bus

- X i m ) X k - cxjn - x j m > x k
- ( x n n + xmm - 2 x n m )

3 (Xi,- xj,- xi, + X j m )

-
- x/ (1 1A.16)
xk - (xnn + xmm - 2 x n m )
The fact that the a and d factors are linear models of the power system allows
us to use superposition to extend them. One very useful extension is to use the
a and d factors to model the power system in its post-outage state; that is, to
generate factors that model the system’s sensitivity after a branch has been lost.
Suppose one desired to have the sensitivity factor between line / and
generator bus i when line k was opened. This is calculated by first assuming
that the change in generation on bus i, A e , has a direct effect on line / and an
indirect effect through its influence on the power flowing on line k , which, in
turn, influences line / when line k is out. Then

‘f/ = a/iAfl + d/,kAfk (llA.17)

However, we know that
Afk = akiAfl (1 1A.18)
therefore,
Af/ = a/,AP, + d/?,akiAP,= ( a / , + d,,,aki)AP, (1 1A.19)

+
We can refer to a,, d / , , a k ias the “compensated generation shift sensitivity.”
The compensated sensitivity factors are useful in finding corrections to the
generation dispatch that will make the post-contingency state of the system
secure from overloads. This will be dealt with in Chapter 13 under the topic
of “security-constrained optimal power flow.”

APPENDIX 11B
Derivation of Eauation 11.14

Equation 11.14, repeated here as Eq. 1 lB.l

lAep - Aeql < lAe, - A8,I (llB.l)

is proved as shown in reference 11 (the proof is attributed to Moslehi).

Suppose that buses i and j have the highest and lowest values of A0 in the
N3 region. Then the following both hold:

Aei > A e ,
 PROBLEMS 445

and
AOj < AOf

for all buses f in N3. Taking any external bus in N2, call it bus e, we shall
state that
AOe < AOi (11B.2)
and
AOe > AOj (1 1B.3)

Proof: Suppose Eq. 11B.2 is not true and there exists a bus e’ such that

Age! > AOi

and, further, suppose that
Age* > AOe (11B.4)

for all the buses in N3. This implies that Eq. 11B.4 holds for the union of buses
in N2 and N3. If we now look at the network as a DC power flow network,
with no impedances to ground, and only the two injections at buses k and rn,
then all incremental power flows leaving node e’ must be positive, since the
incremental flows leaving node e’ are found from

1
Af,,, = -(AOe, - AOe) (llB.5)
Xe’e

However, since the network in N2 and N3 is strictly passive, and there are no
impedances to ground, this violates Kirchoffs current law, which requires all
branch flows incident to a bus to sum to zero. The only way for this to be true
would be if all flows were zero; that is, all incremental angle spreads were equal.
We can continue this reasoning to the neighbor buses of e‘ until we reach node
i and conclude that
AOe, = AOi (llB.6)

which contradicts Eq. 11B.4; thus, Eq. 11B.2 is proved. Equation 11B.3 is proved
in a similar fashion. Then, as a result, Eq. 11B. 1 is also proved.

PROBLEMS

11.1 Figure 11.17 shows a four-bus power system. Also given below are the
impedance data for the transmission lines of the system as well as the
generation and load values.
446 POWER SYSTEM SECURITY

- Bus2
(REFERENCE)

-Bus 3

FIG. 11.17 Four-bus network for Problem 11.1.

Line Line rectance (pu)

1-2 0.2
1-4 0.25
2-3 0.15
2-4 0.30
3-4 0.40

Bus Load (MW) Generation (MW)

1 150
2 350
3 220
4 280

a. Calculate the generation shift sensitivity coefficients for a shift in

generation from bus 1 to bus 2.
b. Calculate the line outage sensitivity factors for outages on lines 1-2,
1-4, and 2-3.

11.2 In the system shown in Figure 11.18, three generators are serving a load
of 1300 MW. The MW flow distribution, bus loads, and generator
outputs are as shown. The generators have the following characteristics.

1 100 600
2 90 400
3 100 500
 PROBLEMS 447

ASSUME NO LOSSES
IN THIS N E T W O R K
300 MW
FIG. 11.18 Three-generator system for Problem 11.2.

The circuits have the following limits:

CKT A 600MWmax
CKT B 600MWmax
CKT C 450MWmax
CKT D 350MWmax

Throughout this problem we will only be concerned with flows on the

circuit labeled A, B, C , and D. The generation shift sensitivity coefficients,
aTi,for circuits, A, B, C, and D are as follows.

CKT Shift on Gen. 1 Shift on Gen. 2

~~

A 0.7 0.08
B 0.2 0.02
C 0.06 0.54
D 0.04 0.36

Example: Apflow, = a/,i x APi

if
5=C and i=2
Apflow, = (0.54)AP,

Assume a shift on gen. 1 or gen. 2 will be compensated by an equal

448 POWER SYSTEM SECURITY

(opposite) shift on gen. 3. The line outage sensitivity factors df,k are

A B C D
A X 0.8 0.21 0.14
0.9 X 0.06 0.04
0.06 0.12 X 0.82
D 0.04 0.08 0.73 X

As an example, suppose the loss of circuit k will increase the loading

on circuit P as follows.

= pflo,, (before outage)

Pflow, + di,k x pflowk(before outage)
if
[=A and k=B

The new flow on G would be

a. Find the contingency (outage) flow distribution on circuits A, B, C ,

and D for an outage on circuit A. Repeat for an outage on B, then
on C , then on D. (Only one circuit is lost at one time.) Are there any
overloads?
b. Can you shift generation from gen. 1 to gen. 3, or from gen. 2 to gen.
3, so that no overloads occur? If so, how much shift?

11.3 Given the three-bus network shown in Figure 11.19 (see Example 4B),
where
XI* = 0.2 pu
x I 3= 0.4 PU
x~~ = 0.25 pu

the [ X I matrix is

0.2118 0.1177 0
0.1177 0.1765 0
 PROBLEMS 449

Bus 2

Bus 3
(REFERENCE e, = 0)

FIG. 11.19 Three-bus system for Problem 11.3.

Use a 100-MVA base. The base loads and generations are as follows.

1 100 150 50 250

2 300 180 60 250
3 100 170 60 300

a. Find base power flows on the transmission lines.

b. Calculate the generation shift factors for line 1-2. Calculate the shift
in generation on bus 1 and 2 so as to force the flow on line 1-2 to
zero MW. Assume for economic reasons that any shifts from base
conditions are more expensive for shifts at the generator on bus 1 than
for shifts on bus 2, and that the generator on bus 3 can be shifted
without any penalty.

11.4 Using the system shown in Example 11B, find N1, N2 and N3 for the
outage of the line from bus 2 to bus 4. Do you need to expand region
N l ? Where are the overloads, if any? (Use the same branch flow limits
as shown in Example 11B.)

11.5 Using the data found in Figure 11.7, find the base-case bus phase angles
and all line flows using the following bus loads and generators: all loads
are 100 MW and all generators are also at 100 MW. Assume line flow
limits as shown in the following table.
450 POWER SYSTEM SECURITY

Line MW Limit
1-2 70
1-4 90
1-5 70
2-3 20
2-4 50
2-5 40
2-6 60
3-5 30
3-6 70
4-5 30
5-6 20

For a line outage on line 1-4, find the change in phase angle across each
of the remaining lines and see if the phase angle change across buses 1
and 4 meets the bounding criteria developed in the text.

11.6 Using the data from Problem 11.2, calculate the performance index, PI,
for each outage case. Use a value of n = 1 and n = 5; that is for

PI= c (flowflowij
all lines maxij
) 2n

Which PI does a better job of predicting the case with the overload?
Explain why.

FURTHER READING

The subject of power system security has received a great deal of attention in the
engineering literature since the middle 1960s. The list of references presented here is
therefore large but also quite limited nonetheless.
Reference 1 is a key paper on the topic of system security and energy control system
philosophy. Reference 2 provides the basic theory for contingency assessment of power
systems. Reference 3 covers contingency analysis using DC power flow methods.
Reference 4 is a broad overview of security assessment and contains an excellent
bibliography covering the literature on security assessment up to 1975.
The use of AC power flows in contingency analysis is possible with any AC load flow
algorithm. However, the fast-decoupled power flow algorithm is generally recognized as
the best for this purpose since its Jacobian matrix is constant and single-line outages
can be modeled using the matrix inversion lemma. Reference 5 covers the fast-decoupled
power flow algorithm and its application.
 FURTHER READING 451

Correcting the generation dispatch by sensitivity methods is covered by reference

6. The use of linear programming to solve power systems problems is covered in
reference 7.
References 8-12 cover some of the literature on contingency selection, and reference
13 gives a technique for solving the power flow using an approximation called concentric
relaxation. References 14 and 15 give an indication of recent research on dynamic
security assessment; that is, detecting fault cases that may cause dynamic or transient
stability problems. Finally, reference 16 is concerned with the emerging area of voltage
stability, which seeks to find contingencies which will cause such severe voltage problems
as to bring on what is known as a “voltage collapse.”

1. DyLiacco, T. E., “The Adaptive Reliability Control System,” I E E E Transactions on

Power Apparatus and Systems, Vol. PAS-86, May 1967, pp. 517-531.
2. El-Abiad, A. H., Stagg, G. W., “Automatic Evaluation of Power System Per-
formance-Effects of Line and Transformer Outages,” A I E E Transactions on Power
Apparatus and Systems, Vol. PAS-81, February 1963, pp. 712-716.
3. Baughman, M., Schweppe, F. C., “Contingency Evaluation: Real Power Flows from
a Linear Model,” IEEE Summer Power Meeting, 1970, Paper C P 689-PWR.
4. Debs, A. S., Benson, A. R., “Security Assessment of Power Systems,” Systems
Engineering For Power: Status and Prospects, U.S. Government Document, CONF-
750867, 1967, pp. 1-29.
5. Stott, B., Alsac, O., “Fast Decoupled Load Flow,” I E E E Transactions on Power
Apparatus and Systems, Vol. PAS-93, May/June 1974, pp. 859-869.
6. Thanikachalam, A., Tudor, J. R., “Optimal Rescheduling of Power for System
Reliability,” I E E E Transactions on Power Apparatus and Systems, Vol. PAS-90,
July/AuguSt 1971, pp. 1776-1781.
7. Chan, S . M., Yip, E., “A Solution of the Transmission Limited Dispatch Problem
by Sparse Linear Programming,” I E E E Transactions on Power Apparatus and
Systems, Vol. PAS-98, May/June 1979, pp. 1044-1053.
8. Ejebe, G. C., Wollenberg, B. F., “Automatic Contingency Selection,” I E E E Trans-
actions on Power Apparatus and Systems, Vol. PAS-98, JanuaryIFebruary 1979,
pp. 92-104.
9. Albuyeh, F., Bose, A., Heath, B., “Automatic Contingency Selection; Ranking
Outages on the Basis of Real and Reactive Power Equations,” I E E E Trans-
actions on Power Apparatus and Systems, Vol. PAS-101, No. 1, January 1982,
pp. 107-112.
10. Ejebe, G. C., VanMeeteren, H. P., Wollenberg, B. F. “Fast Contingency Screening
and Evaluation for Voltage Security Analysis,” I E E E Transactions on Power Systems,
Vol. 3, No. 4, November 1988, pp. 1582-1590.
11. Brandwajn, V., “Efficient Bounding Method for Linear Contingency Analysis,”
I E E E Transactions on Power Systems, Vol. 3, No. 1, February 1988, pp. 38-43.
12. Brandwajn, V., Lauby, M. G., “Complete Bounding Method for AC Contingency
Screening,” I E E E Transactions on Power Systems, Vol. PWRS-4, May 1989,
pp. 724-729.
13. Zaborsky, J., Whang, K. W., Prasad, K., “Fast Contingency Evaluation using
Concentric Relaxation,” I E E E Transactions on Power Apparatus and Systems, Vol.
PAS-99, January/February 1980, pp. 28-36.
452 POWER SYSTEM SECURITY

14. Fouad, A. A., “Dynamic Security Assessment Practices in North America,” I E E E

Transactions on Power Systems, Vol. 3, No. 3, August 1988, pp. 1310-1321.
15. El-Kady, M. A., Tang, C. K., Carvalho, V. F., Fouad, A. A., Vittal, V., “Dynamic
Security Assessment Utilizing the Transient Energy Function Method,” I E E E
Transactions on Power Systems, Vol. PWRS-1, No. 3, August 1986, pp. 284-291.
16. Jasmon, G. B., Lee, L. H. C. C., “New Contingency Ranking Technique Incorporating
a Voltage Stability Criterion,” I E E Proceedings, Part C : Generation, Transmission
and Distribution, Vol. 140, No. 2, March 1993, pp. 87-90.
12 An Introduction to State
Estimation in Power Systems

12.1 INTRODUCTION

State estimation is the process of assigning a value to an unknown system state

variable based on measurements from that system according to some criteria.
Usually, the process involves imperfect measurements that are redundant and
the process of estimating the system states is based on a statistical criterion
that estimates the true value of the state variables to minimize or maximize the
selected criterion. A commonly used and familiar criterion is that of minimizing
the sum of the squares of the differences between the estimated and “true” (i.e.,
measured) values of a function.
The ideas of least-squares estimation have been known and used since the
early part of the nineteenth century. The major developments in this area have
taken place in the twentieth century in applications in the aerospace field. In
these developments, the basic problems have involved the location of an
aerospace vehicle (i.e., missile, airplane, or space vehicle) and the estimation of
its trajectory given redundant and imperfect measurements of its position and
velocity vector. In many applications, these measurements are based on optical
observations and/or radar signals that may be contaminated with noise
and may contain system measurement errors. State estimators may be both
static and dynamic. Both types of estimators have been developed for power
systems. This chapter will introduce the basic development of a static-state
estimator.
In a power system, the state variables are the voltage magnitudes and relative
phase angles at the system nodes. Measurements are required in order to
estimate the system performance in real time for both system security control
and constraints on economic dispatch. The inputs to an estimator are imperfect
power system measurements of voltage magnitudes and power, VAR, or
ampere-flow quantities. The estimator is designed to produce the “best
estimate” of the system voltage and phase angles, recognizing that there are
errors in the measured quantities and that there may be redundant measure-
ments. The output data are then used in system control centers in the
implementation of the security-constrained dispatch and control of the system
as discussed in Chapters 1 1 and 13.
453
454 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

12.2 POWER SYSTEM STATE ESTIMATION

As introduced in Chapter 11, the problem of monitoring the power flows and
voltages on a transmission system is very important in maintaining system
security, By simply checking each measured value against its limit, the power
system operators can tell where problems exist in the transmission system-and,
it is hoped, they can take corrective actions to relieve overloaded lines or
out-of-limit voltages.
Many problems are encountered in monitoring a transmission system. These
problems come primarily from the nature of the measurement transducers and
from communications problems in transmitting the measured values back to
the operations control center.
Transducers from power system measurements, like any measurement device,
will be subject to errors. If the errors are small, they may go undetected and
can cause misinterpretation by those reading the measured values. In addition,
transducers may have gross measurement errors that render their output
useless. An example of such a gross error might involve having the transducer
connected up backward; thus, giving the negative of the value being measured.
Finally, the telemetry equipment often experiences periods when communi-
cations channels are completely out; thus, depriving the system operator of any
information about some part of the power system network.
It is for these reasons that power system state estimation techniques have been
developed. A state estimator, as we will see shortly, can “smooth out” small
random errors in meter readings, detect and identify gross measurement errors,
and “fill in” meter readings that have failed due to communications failures,
To begin, we will use a simple DC load flow example to illustrate the
principles of state estimation. Suppose the three-bus DC load flow of Example
4B were operating with the load and generation shown in Figure 12.1. The only
information we have about this system is provided by three MW power flow
meters located as shown in Figure 12.2.
Only two of these meter readings are required to calculate the bus phase
angles and all load and generation values fully. Suppose we use M13 and M3,
and further suppose that Mi3 and M32 give us perfect readings of the flows on
their respective transmission lines.

M13 = 5 MW = 0.05 PU
M32 = 40 MW = 0.40 PU
Then, the flows on lines 1-3 and 3-2 can be set equal to these meter readings.

1
f13=-(61 -63)=M13=0.05~~
x13

1
f32 = -(63 - 62) = M3, = 0.40 pu
x2 3
 POWER SYSTEM STATE ESTIMATION 455

PE

Bus 3
FIG. 12.1 Three-bus system from Example 4B.

Bus 1. Bus 2

Bus 3

FIG. 12.2 Meter placement.

Since we know that 8, = 0 rad, we can solve the f,, equation for el, and the
f32 equation for 8,, resulting in

8, = 0.02 rad
8, = -0.10rad
456 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

We will now investigate the case where all three meter readings have slight
errors. Suppose the readings obtained are

M , , = 62 MW = 0.62 PU
M13= 6 MW = 0.06 PU
M32 = 37 MW = 0.37 PU

If we use only the M , , and M32 readings, as before, we will calculate the phase
angles as follows:
= 0.024 rad

O2 = -0.0925 rad
0, = 0 rad (still assumed to equal zero)

This results in the system flows as shown in Figure 12.3. Note that the predicted
flows match at M I , and M,,, but the flow on line 1-2 does not match the
reading of 62 MW from M12.If we were to ignore the reading on MI, and use
M I , and M32, we could obtain the flows shown in Figure 12.4.
All we have accomplished is to match M12,but at the expense of no longer
matching M , ,. What we need is a procedure that uses the information available
from all three meters to produce the best estimate of the actual angles, line
flows, and bus load and generations.
Before proceeding, let’s discuss what we have been doing. Since the only
thing we know about the power system comes to us from the measurements,
Bus 1

-
0-t
-+-
M12
58.25

-7 --I
102 MW

t12 = -0.0925

Bus 3

FIG. 12.3 Flows resulting from use of meters M,,and M3*.

 POWER SYSTEM STATE ESTIMATION 457
Bus 1 Bus 2
-62 MW

M12

*
99 MW
--f
69.875 MW M13

BUS 3

FIG. 12.4 Flows resulting from use of meters M , , and M3,.

we must use the measurements to estimate system conditions. Recall that in

each instance the measurements were used to calculate the bus phase angles at
bus 1 and 2. Once these phase angles were known, all unmeasured power flows,
loads, and generations could be determined. We call 8, and O2 the state oariables
for the three-bus system since knowing them allows all other quantities to be
calculated. In general, the state variables for a power system consist of the bus
voltage magnitude at all buses and the phase angles at all but one bus. The
swing or reference bus phase angle is usually assumed to be zero radians. Note
that we could use real and imaginary components of bus voltage if desired. If
we can use measurements to estimate the “states” (i.e., voltage magnitudes and
phase angles) of the power system, then we can go on to calculate any power
flows, generation, loads, and so forth that we desire. This presumes that the
network configuration (i.e., breaker and disconnect switch statuses) is known
and that the impedances in the network are also known. Automatic load tap
changing autotransformers or phase angle regulators are often included in a
network, and their tap positions may be telemetered to the control as a
measured quantity. Strictly speaking, the transformer taps and phase angle
regulator positions should also be considered as states since they must
be known in order to calculate the flows through the transformers and
regulators.
To return to the three-bus DC power flow model, we have three meters
providing us with a set of redundant readings with which to estimate the two
states 8, and 02.We say that the readings are redundant since, as we saw earlier,
only two readings are necessary to calculate 8, and 8,, the other reading is
always “extra.” However, the “extra” reading does carry useful information
and ought not to be discarded summarily.
458 AN INTRODUCTION T O STATE ESTIMATION IN POWER SYSTEMS

This simple example serves to introduce the subject of static-state estimation,

which is the art of estimating the exact system state given a set of imperfect
measurements made on the power system. We will digress at this point to
develop the theoretical background for static-state estimation. We will return
to our three-bus system in Section 12.3.4.

12.3 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES

ESTIMATION

12.3.1 Introduction
Statistical estimation refers to a procedure where one uses samples to calculate
the value of one or more unknown parameters in a system. Since the samples
(or measurements) are inexact, the estimate obtained for the unknown parameter
is also inexact. This leads to the problem of how to formulate a “best”estimate
of the unknown parameters given the available measurements.
The development of the notions of state estimation may proceed along
several lines, depending on the statistical criterion selected. Of the many criteria
that have been examined and used in various applications, the following three
are perhaps the most commonly encountered.

1. The maximum likelihood criterion, where the objective is to maximize the

probability that the estimate of the state variable, 2, is the true value of
the state variable vector, x (i.e., maximize P ( 9 ) = x).
2. The weighted least-squares criterion, where the objective is to minimize
the sum of the squares of the weighted deviations of the estimated
measurements, P, from the actual measurements, z.
3. The minimum variance criterion, where the object is to minimize the
expected value of the sum of the squares of the deviations of the estimated
components of the state variable vector from the corresponding com-
ponents of the true state variable vector.

When normally distributed, unbiased meter error distributions are assumed,

each of these approaches results in identical estimators. This chapter will utilize
the maximum likelihood approach because the method introduces the measure-
ment error weighting matrix [ R ] in a straightforward manner.
The maximum likelihood procedure asks the following question: “What is
the probability (or likelihood) that I will get the measurements I have
obtained?” This probability depends on the random error in the measuring
device (transducer) as well as the unknown parameters to be estimated.
Therefore, a reasonable procedure would be one that simply chose the estimate
as the value that maximizes this probability. As we will see shortly, the
maximum likelihood estimator assumes that we know the probability density
function (PDF) of the random errors in the measurement. Other estimation
 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 459

schemes could also be used. The “least-squares” estimator does not require that
we know the probability density function for the sample or measurement errors.
However, if we assume that the probability density function of sample or
measurement error is a normal (Gaussian) distribution, we will end up with
the same estimation formula. We will proceed to develop our estimation
formula using the maximum likelihood criterion assuming normal distributions
for measurement errors. The result will be a “least-squares” or more precisely
a “weighted least-squares” estimation formula, even though we will develop
the formulation using the maximum likelihood criteria. We will illustrate this
method with a simple electrical circuit and show how the maximum likelihood
estimate can be made.
First, we introduce the concept of random measurement error. Note that we
have dropped the term “sample” since the concept of a measurement is much
more appropriate to our discussion. The measurements are assumed to be in
error: that is, the value obtained from the measurement device is close to the
true value of the parameter being measured but differs by an unknown error.
Mathematically, this can be modeled as follows.
~ ~~~~

Let zmeasbe the value of a measurement as received from a measurement

device. Let z‘IUebe the true value of the quantity being measured. Finally,
let q be the random measurement error. We can then represent our
measured value as
Zmeas = Z t r u e + v (12.1)

The random number, q, serves to model the uncertainty in the measurements.

If the measurement error is unbiased, the probability density function of q is
usually chosen as a normal distribution with zero mean. Note that other
measurement probability density functions will also work in the maximum
likelihood method as well. The probability density function of r j is

1
PDF(q) = ~ exp( - q2/2e2) (12.2)
a&

where c is called the standard deviation and c 2 is called the variance of the
random number. PDF(q) describes the behavior of q. A plot of PDF(rj) is shown
in Figure 12.5. Note that c, the standard deviation, provides a way to model
the seriousness of the random measurement error. If 0 is large, the measurement
is relatively inaccurate (i.e., a poor-quality measurement device), whereas a
small value of c denotes a small error spread (i.e., a higher-quality measurement
device). The normal distribution is commonly used for modeling measurement
errors since it is the distribution that will result when many factors contribute
to the overall error.
460 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

PDF (11)

FIG. 12.5 The normal distribution.

12.3.2 Maximum Likelihood Concepts

The principle of maximum likelihood estimation is illustrated by using a simple
DC circuit example as shown in Figure 12.6. In this example, we wish to estimate
the value of the voltage source, xtrUe,
using an ammeter with an error having
a known standard deviation. The ammeter gives a reading of zTas, which is
equal to the sum of zYUe(the true current flowing in our circuit) and v1 (the
error present in the ammeter). Then we can write
Zrneas
1
-
- Ztrue
1 +?l (12.3)

Since the mean value of q1 is zero, we then know that the mean value of zYeas
is equal to zYUe.This allows us to write a probability density function for zYeas as

( 1 2.4)

where o1 is the standard deviation for the random error g,. If we assume that
the value of the resistance, r l , in our circuit is known, then we can write

1
PDF(zyeaS)= ___ (1 2.5)
a,& exp

FIG. 12.6 Simple DC circuit with current measurement.

 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 461

Coming back to our definition of a maximum likelihood estimator, we now

wish to find an estimate of x (called xesl) that maximizes the probability that
the observed measurement zYeaswould occur. Since we have the probability
density function of zYeas,we can write

prob(zYeas)= I .ya +&;err

Zmcdr
I
PDF(zYeaS)d z y a s as d z y -,0
= PDF(zYeaS)dZYeas (12.6)

The maximum likelihood procedure then requires that we maximize the

value of prob(zTeas),which is a function of x. That is,

max prob(zTeas)= max PDF(zYeas)dzYeas ( 1 2.7)

X X

One convenient transformation that can be used at this point is to maximize

the natural logarithm of PDF(zYeaS)since maximizing the Ln of PDF(zYeaS)
will also maximize PDF(zYaS),Then we wish to find

max Ln[PDF(z;"'"')]
L
or

Since the first term is constant, it can be ignored. We can maximize the function in
brackets by minimizing the second term since it has a negative coefficient, that is,

L
is the same as

(12.8)

The value of x that minimizes the right-hand term is found by simply taking
the first derivative and setting the result to zero:
462 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

FIG. 12.7 DC circuit with t w o current measurements.

To most readers this result was obvious from the beginning. All we have
accomplished is to declare the maximum likelihood estimate of our voltage as
simply the measured current times the known resistance. However, by adding a
second measurement circuit, we have an entirely different situation in which the
best estimate is not so obvious. Let us now add a second ammeter and resistance
as shown in Figure 12.7.
Assume that both r1 and r2 are known. As before, model each meter reading
as the sum of the true value and a random error:

(12.10)

where the errors will be represented as independent zero mean, normally

distributed random variables with probability density functions:

PDF(vl) = ~

a,&
1 exp( s) (12.11)
PDF(v2) =
%fi
~ $)1 exp(
202

and as before we can write the probability density functions of z;leasand z;leas as

(12.12)

1
PDF(zYeaS)= ~
 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 463

The likelihood function must be the probability of obtaining the measure-

ments zTeaSand zTeas.Since we are assuming that the random errors q1 and q2
are independent random variables, the probability of obtaining zTeas and zTeas
is simply the product of the probability of obtaining zYeasand the probability
of obtaining z;leas.

prob(zreasand zTeas) = prob(zyeas)x (prob(z7'"')

= PDF(zyeaS)PDF(zTeaS)dZyeasdZTeas

To maximize the function we will again take its natural logarithm:

max prob(zreasand zTeas)

X

= min
X 20:
+ ( 12.14)

The minimum sought is found by

464 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

giving

(12.15)

If one of the ammeters is of superior quality, its variance will be much smaller
than that of the other meter. For example, if C J <<~ o:, then the equation for xest

becomes
xesl= Geas x r2

Thus, we see that the maximum likelihood method of estimating our unknown
parameter gives us a way to weight the measurements properly according to
their quality.
It should be obvious by now that we need not express our estimation
problem as a maximum of the product of probability density functions. Instead,
we can observe a direct way of writing what is needed by looking at Eqs. 12.8
and 12.14. In these equations, we see that the maximum likelihood estimate of
our unknown parameter is always expressed as that value of the parameter that
gives the minimum of the sum of the squares of the difference between each
measured value and the true value being measured (expressed as a function of
our unknown parameter) with each squared difference divided or “weighted”
by the variance of the meter error. Thus, if we are estimating a single parameter,
x, using N, measurements, we would write the expression

N“! [ Z y a s - fi(X)]2
min ~ ( x =
) 1 (12.16)
X i= 1 o?
where
fi = function that is used to calculate the value being measured by the ith
measurement
02 = variance for the ith measurement
J(x) = measurement residual
N, = number of independent measurements
zYeas= i l h measured quantity

Note that Eq. 12.16 may be expressed in per unit or in physical units such as
MW, MVAR, or kV.
If we were to try to estimate N, unknown parameters using N, measurements,
we would write
 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 465

The estimation calculation shown in Eqs. 12.16 and 12.17 is known as a

weighted least-squares estimator, which, as we have shown earlier, is equivalent
to a maximum likelihood estimator if the measurement errors are modeled as
random numbers having a normal distribution.

12.3.3 Matrix Formulation

If the functions fi(xl, x 2 , . . . , xNs)are linear functions, Eq. 12.17 has a closed-
form solution. Let us write the function fi(xl, x2, . . . , xNs) as

Then, if we place all the fi functions in a vector, we may write

(12.19)

where

[HI = an N, by N, matrix containing the coefficients of the linear functions fi(x)

N, = number of measurements
N, = number of unknown parameters being estimated

Placing the measurements in a vector:

(12.20)

We may then write Eq. 12.17 in a very compact form.

min J ( x ) = [zmeas- f(x)] T I R - l ] [ ~ m e-a sf(x)l (12.21)

X

where
466 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

[ R ] is called the covariance matrix of measurement errors. To obtain the general

expression for the minimum in Eq. 12.21, expand the expression and substitute
[ H ] x for f ( x ) from Eq. 12.19.

min J ( x ) = ( z ~ ~ ~ ~ ~- x[ ~R [- ~~ ~] ]z ~~ [ ~ R ~ - ~~ ] z ~ ~ ~ ~
X

-zmeasT[R-'][H]x + xTIHITIR-'][H]x) (12.22)

Similar to the procedures of Chapter 3, the minimum of J ( x ) is found when

d J ( x ) / d x i = 0, for i = 1 , . . . , N,; this is identical to stating that the gradient of
J ( x ) , V J ( x ) , is exactly zero.
The gradient of J ( x ) is (see the appendix to this chapter)

+
V J ( X )= - ~ [ H ] * [ R - ' ] Z " ' ~ ~2 ~[ H ] T [ R - 1 ] [ H ] ~

Then V J ( x ) = 0 gives

xest = [ [ H I T [ R - ' ] [ H ] ] - ' [ H I T [ R - ' 1~"'~~~ (12.23)

Note that Eq. 12.23 holds when N, < N,; that is, when the number of parameters
being estimated is less than the number of measurements being made.
When N, = N,, our estimation problem reduces to

Xest = [ H I - lZmeas (12.24)

There is also a closed-form solution to the problem when N, > N,, although
in this case we are not estimating x to maximize a likelihood function since
N, > N, usually implies that many different values for xestcan be found that
cause f i ( x e s t )to equal zTeas for all i = 1 , . . . , N, exactly. Rather, the objective
is to find xes' such that the sum of the squares of xps' is minimized. That is,

N.
min C xz = x T x (12.25)
x i=l

subject to the condition that zmeas= [ H l x . The closed-form solution for this
case is
xest = [ H I ' [ [ H ] [ H ] '3 - l z m e a s (12.26)

In power system state estimation, underdetermined problems (i.e., where

N, > N,) are not solved, as shown in Eq. 12.26. Rather, "pseudo-measurements"
are added to the measurement set to give a completely determined or
overdetermined problem. We will discuss pseudo-measurements in Section
12.6.3. Table 12.1 summarizes the results for this section.
 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 467

TABLE 12.1 Estimation Formulas

Case Description Solution Comment
N, -= N, Overdetermined xesl= [[H]TIR-'][H]]-l xestis the maximum
x { H] T [ R - 1-Jzmeas likelihood estimate
of x given the
measurements Par
N, = N, Completely Xest = [H] - lZmeas xestfits the measured
determined quantities to the
measurements Peas
exactly
N, > N, Underdetermined xest= [H]T[[H][H]T]-'zmeas xestis the vector of
minimum norm that
fits the measured
quantities to
the measurements
exactly. (The norm
of a vector is equal
to the sum of the
squares of its
components)

12.3.4 An Example of Weighted Least-Squares State Estimation

We now return to our three-bus example. Recall from Figure 12.2 that we have
three measurements to determine 8, and 8,, the phase angles at buses 1 and 2.
From the development in the preceding section, we know that the states 8, and
8, can be estimated by minimizing a residual J ( 8 , , 8,) where J(8,, 8,) is the
sum of the squares of individual measurement residuals divided by the variance
for each measurement.
To start, we will assume that all three meters have the following characteristics.

Meter full-scale value: 100 MW

Meter accuracy: f 3 MW

This is interpreted to mean that the meters will give a reading within f 3 M W of
the true value being measured for approximately 99% of the time. Mathe-
matically, we say that the errors are distributed according to a normal probability
density function with a standard deviation, g, as shown in Figure 12.8.
Notice that the probability of an error decreases as the error magnitude
increases. By integrating the P D F between -30 and +30 we come up with a
value of approximately 0.99. We will assume that the meter's accuracy (in our
case? 3 MW) is being stated as equal to the 30 points on the probability density
function. Then f 3 MW corresponds to a metering standard deviation of
0 = 1 MW = 0.01 PU.
468 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

I True value of
-
1

U
H I quantity being
I measured

The formula developed in the last section for the weighted least-squares
estimate is given in Eq. 12.23, which is repeated here.

xest= [ [ H I T I R - ' ] [ H ] ] - ' [ H I T [ R - ' 1~"'~~~

where
xest= vector of estimated state variables
[ H I = measurement function coefficient matrix
[ R ] = measurement covariance matrix
zmeas= vector containing the measured values themselves

For the three-bus problem we have

(12.27)

To derive the [ H I matrix, we need to write the measurements as a function

of the state variables 8, and 8,. These functions are written in per unit as

1
M , , = f 1 2 = - (e, - e,) = 58, - 58,
0.2
1
~ 1 =3f i 3 = -(el - e,) = 2.58, ( 12.28)
0.4
1
M,, = f3, =-
0.25
(e, - e,) = -48,
The reference-bus phase angle, O,, is still assumed to be zero. Then
 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 469

The covariance matrix for the measurements, [R], is

OM12 0.000 1

OM32

Note that since the coefficients of [ H I are in per unit we must also write [R]
and zmeasin per unit.
Our least-squares “best” estimate of 8, and 8, is then calculated as

x[ - 5 2.5
0 - 4o][
0.000 1
o.oO01 ]-[:1;
0.0001
312500 -250000]-’[ 325001
=[-250000 41oooO -45800

- 0.094286
where

I:
0.37

From the estimated phase angles, we can calculate the power flowing in
each transmission line and the net generation or load at each bus. The results
are shown in Figure 12.9. If we calculate the value of J(B,, e2>,the residual, we
get

- ~ 0 . 6 2- (58, - se2)l2+ ~ 0 . 0 6- (2.se,)l2 + [ON + (4e2)12

0.0001 o.Ooo1 0.0001
= 2.14 ( 1 2.29)
470 AN INTRODUCTION TO STATE ESTIMATION IN POWER SYSTEMS

2
v
-61.4MW

z 99. 1 MW

+
68.5 MW
13
A
v
8 , =0.028571 37.7 M W ~
e2 = -0.094286

Bus 3
FIG. 12.9 Three-bus example with best estimates of Q1 and Q2.

Suppose the meter on the MI, transmission line was superior in quality to
those on MI, and M,,. How will this affect the estimate of the states? Intuitively,
we can reason that any measurement reading we get from M i 3 will be much
closer to the true power flowing on line 1-3 than can be expected when
comparing M , , and M 3 , to the flows on lines 1-2 and 3-2, respectively.
Therefore, we would expect the results from the state estimator to reflect this
if we set up the measurement data to reflect the fact that M , , is a superior
measurement. To show this, we use the following metering data,

Meters MI2and M32: 100 MW full scale

f3 MW accuracy
(a = 1 MW = 0.01 pu)
Meter M13: 100 MW full scale
k 0.3 MW accuracy
(a = 0.1 MW = 0.001 pu)

The covariance matrix to be used in the least-squares formula now becomes

 MAXIMUM LIKELIHOOD WEIGHTED LEAST-SQUARES ESTIMATION 47 1

We now solve Eq. 12.23 again with the new [ R ] matrix.

1 x 10-4
1x
1 x10-4 I" 5
2.5
o
-5

-4.
0

From these estimated phase angles, we obtain the network conditions shown
in Figure 12.10. Compare the estimated flow on line 1-3, as just calculated, to
the estimated flow calculated on line 1-3 in the previous least-squares estimate.
Setting o M I 3to 0.1 MW has brought the estimated flow on line 1-3 much
closer to the meter reading of 6.0 MW. Also, note that the estimates of flow on
lines 1-2 and 3-2 are now further from the M , , and M3* meter readings,
respectively, which is what we should have expected.

+60.55 MW
M12

-99.35 MW

66.58MW --t
M13

FIG. 12.10 Three-bus example with better meter at M13.