OptimalSystem Operation A3

parianceis caiied the 'unii commitment' (UC) probiemand the secondis calleci
the 'load scheduling' (LS) problem.One must first solvethe UC problem before
proceedingwith the LS problem.
Throughout this chapter we shall concern ourselves with an existing
installation,so that the economicconsiderationsare that of operating(running)
cost and not the capitai outiay.

7.2 OPTIMAL OPERATION OF GENERATORS ON A BUS BAR

Before we tackle the unit commitment problem, we shall consider the optimal
operation of generatorson a bus bar.

Generator Operating Cost

The major component of generatoroperating cost is the fuel input/hour, while

maintenance coutributesonly to a small extent.The fuel cost is meaningtulin
case of thermal and nuclear stations,but for hydro stations where the energy
storageis 'apparentlyfree', the operatingcost as such is not meaningful. A
7.1 INTRODUCTION suitablemeaningwill be attachedto the cost of hydro storedenergyin Section
7.7 of this chapter. Presentlywe shall concentrateon fuel fired stations.
The optimal system operation,in general,involved the considerationof l
I
economyof operation,systemsecurity,emissionsat certain fossil-fuel plants, II
optimalreleasesof water at hydro generation,etc. All theseconsiderationsmay t -
L \

makefor conflictingrequirements and usuallya compromisehasfo he madefor -

; P
al)

optimalsystemoperation.ln this chapterwe cor..;iderthe economyof operation t r o

. c 8
only, also called the ec'omonicdi.spcttch problem. 6 o
The main aim in theeconomicdispatchproblernis to rninimize thc total cost 8b=
of generatingreal power (production cost) at various stations while satisfying
e 5
g E
theloadsandthe lossesin thetransmission links.For sirnplicitywe consicler
the o
presenceof thermalplantsonly in the beginning.In the later part of this chapter O 6
f

we will considerthe presenceof hydro plants which operatein conjunctionwith LL

thermalplants.While thereis negligible operatingcost at a hydro plant, there

o (vw)min (MW)max
is a limitationol availabilityol'watcr'over a pcriodof tinrewhich nrustbc used
Poweroutput,MW -'- --
to savemaximum fuel at the thermal plants.
In the load flow problemas detailedin Chapter6, two variablesare specified Fig.7.1 Input-outputcurve of a generatrngunit
at eachbus and the solutionis then obtainedfor the rernainingvariables.The
The input-output curve of a unit* can be expressedin a million kilocalories
specifiedvariablesare real and reactivepowersat PQ buses,real powersand
per hour or directly in terms of rupees per hour versus output in megawatts. The
voltagemagnitudesat PV buses,and voltagemagnitudeand angle at the slack
cost curve can be determined experimentaily. A typical curve is shown in
bus.The additionalvariablesto be specifiedfor load flow solution are the tap Fig. 7.1 where (MW)o'in is the minimum loading limit below which it is
settingsof regulatingtransformers.If the specifiedvariablesare allowed to vary uneconomical (or may be technically infeasible) to operate the unit and
in a regionconstraineci by practicaiconsicierations (upperanciiower iimits on (MW)n,u, is the maximunr output limit. The inpLlt-output curve has
activeand reactivegenerations, bus voltagelimits, and rangeof transformertap discontinuities at steam valve openings which have not been indicated in the
settings),there resultsan infinite numberof load flow solutions,eachpertaining figure. By fitting a suitable degree polynomial, an analytical expression for
to one set of valueso1'specifiedvariables.The 'best'choice in sornesenseof operating cost can be written as
thevaluesof specifiedvariablesleadsto the 'best' load flow solution.Economy
of operationis naturallypredominantin determiningallocationof generationto A unit consists of a boiler, turbine and generator.
eachstationfor varioussystemload levels.The first problem in power system
 | . . - - A - - - ! - - A - ^ r - - ^ : -
.L4+ | MOOern rower DySIeIrl Arlaly
I Considerationsof spinning reserve,to be explained later in this section,require
Ci(Pc) Rs/hourat outPutPc, that
(7.6)
where the suffix i standsfor the unit number.It generatly suffices to fit a second D Po,,^*) Po
degreepolynomial,i.e.
marsin. i.e. Eq. (7.6) must be a strict inequality.
_Q; b,Pc, + d, Rs/hour (7.r) Since the operatingcost is insensitiveto reactive loading of a generator,the
2
rnannerin which the reactive load of the station is sharedamong various on-
The slope of the cost curve, i.".43 is called the incrementalJuel cost(lQ, line generatorsdoes not afl'ectthe operatingeconomy'.
dPo, 'What is the optimal manner
The question that has now to be answeredis:
and is expressedin units of rupeesper megawatthour (Rs/lvIWh).A typical plot in which the load demand Po must be shared by the generatorson the bus?'
of incrementalfuel cost versus power output is sho.wnin Fig. 7.2.If the cost This is answeredby minimizing the operating cost
curve is approximatedas a quadraticas in Eq. (7.1), we have k

(lc)i= aiP"t + bt (7.2) c = D ci(pci) ( 7. 7)

f:l

underthe equalityconstraintof meetingthe load demand,i.e.

l
I (
II f - \' G
D (7.8)
-i. i l- P o = O
(J L
i:t
o
o
o c where k = the number of generatorson the bus.
E B
,d> Further, the loading of each generator is constrained by the inequality
o o constraintof Eq. (7.5).
Etr
o
E Since Ci(Pc) is non-linear and C, is independentof P6t (i+ i), this is a
E
() separablenon-linear programming problem. \
If it is assumedat present,that the inequality constraint of Eq. Q.q is not
effective,the problem can be solved by the method of Lagrangemultipliers.
(MW)min (MW) max Define the Lagrangian as
Poweroutput,MW
fuel costversuspoweroutputfor the unitwhose
Fig.7.2 Incremental
f -
it (Pci)-^[f -""] (7.e)
curve is shownin Fig.7.1
input-output "",
where X is the Lagrange multiplier.
i.e. a linear relationship.For better accuracy incremental fuel cost may be Minimization is achieved by the condition
expressedby a number of short line segments (piecewise lineanzation).
Altcrnativcly,wc can fit a polynomial of suitabledegreeto representIC curve
of, =o
in the inverseform dPo,
Pc;i= a, + {),(lC)i + 1,QC)', + ... (7.3) dC' i
- ) i i = 1 , 2 , . . . ,k (7.1o)
or
dPc,
Optimal Operation

[,et us assulnethat it is known a priltri which generutorsitre t<lrtln to ntccta where lci is the incrementalcost of the ith generator (units: Rs/TvIWh),
d4,,
load clenrand
p:.rrticular on the statton.Ubvtously
(7.4) a Iurctio' ol'.gencra,il;,":':t E q u a t i o n ( 7 . 1 0 ) c a n be written as
DPr,,, ) P, ,'',,*
:": dC^ \
is the rated real power capacity of the ith generatorand Po is
where Pci, ,r.,,"* ( 7 .r l )
d4(il- dr-- dPoo
the total power demandon the station.Further,the load on each generatoris to
irc constrainedwithin lower and upper limits, i.e. *The
effect of reactive loading on generatorlosses is of negligible order.
Pcr, .in 1 Po, 1 Po,, rn.*, I = L, 2, "', k (7.s)
 n-.:-^r A-^^r--- ^ .. L -^-

cost of the plant colrespondsto that of unit 2 alone. When rhe plant load is
40
Mw, each unit operatesat its minimum bound, i.e.2o Mw wiitr plant \ = Rs
35/I4Wh.
When dczldPcz= Rs 44/MWh,
Computer solution fbr optimal loa<ling of generators can be obtained
iteratively as follows: 0.25PG2+30 - 44
1. Choose a trial value of ), i.e. IC = (IC)o.
2. Solve for P", (i = 1, 2, ..., k) from Eq. (7.3). or pnt = JI- = 56 MW
0. 25
3. If ItPc,- Pol < e(a specified value), the optimal solution is reached. The total plant output is then (56 + 20) = 76 MW. From this point onwards,
Otherwise, the values of plant load sharedby the two units are found by assumingvarious
valuesof \. The results are displayedin Table 7.1.
4. Increment(lC) bv A (1"), t I: Po,- ,rl I < 0 or decrement(/c) by A(tr)
fl Table 7-1 Outputof each unit and plant output for variousvaluesof
if [D Pc, - Pr] r 0 and repeat from step 2. This step is possible because ) for Example7.1
P.-, is monotonicallyincreasingfunction of (1g).
Plant ), Unit I Unit 2
consider now the effecr of the inequality constraint (7.5). As (1c) is Plant Output
RszMWh Pcl, MW Pcz, NfW
increasedor decreasedin the iterative process,if a particular generatorloading
P", reachesthe limit PGi,^o or P6;, min, its loading from now on is held fixed 35 40.0
at this value and the balance load is then shared between the remaining 44 76.0
50 130. 0
generators on equal incremental cost basis. The fact that this operation is
55 175.0
optimal can be shown by rhe Kuhn-Tucker theory (seeAppendix n;.
60 220.0\
61.25 231.25
65 250.0
Figure 7.3 shows the plot of the plant .trversusplant output. It is seenfrom
Table 7.7 that at .\ = 61.25, unit 2 is operatingat its upper limit and therefore,
Incremental fuel costs in rupeesper MWh for a plant consisting of two units the additional load must now be taken by unit 1, which then determines
are: the
plant ).
dt
i * -G o
1 .2opct+40.0 60
+
t - -
-dcz-= o.2,pcz+3o.o l c c
dPo, I

Assume that both units are operating at all times, and total load varies from 40 8 5 0
o

MW to 250 MW, and the maximum and minimum loads on each unit are to be @ ^

i= 45
I25 and 20 MW, respectively.How will the load be sharedbetween the two a>
units as the systemload vanes over the full range?What are the corresponding E
t -
P 40
values of the plant incrementalcosts? E
O
c J c
Solution At light loads, unit t has the higher incrementalfuel cost and will,
therefore,operareat its lower limit of zo Mw, for which dcrldpcr is Rs 44 per
MWh. When the ourputof unit 2is20 MW, dczldpcz= Rs 35 p"i UWt. Thus,
with an increasein the plant output, the additionalload should be borne bv unit Plantoutput,MW
Fig.7.3 Incrementalfuel cost versus plant output,as found in Example7.1
t:Hzut"l
.Z.48:i MocjernPower Systemnnaiysis
Net savingcausedby optimumscheduling
is
To find the load sharingbetweenthe units for a plant output of say 150 MW,
we find from the curve of Fig. 7.3, that the correspondingplant X is Rs 52,22 772.5- 721.875= 50.625Rs/lr
per MWh. Optimum schedulesfor each unit for 150 MW plant load can now Totalyearlysavingassumingcontinuous
operation
be found as
0.2Pa*40-52.22; P c t = 6 1 ' 1 1M W This savingjustifies the need for optimal load sharing and the ddvices to be
0.25PG2
+ 30= 52.22; Pcz = 88'89 MW installed for controlling the unit loadings automaticallv.

P c r + P c z = 1 5 0M W
Proceeding on the above lines, unit outputs for various plant outputs are
computedand have beenplottedin Fig. 7.4. Optimum load sharingfor any plant
load can be directlv read from this fieure. Let the two units of the svstem studiedin Example7.1 have the following cost
curves.
Cr = 0.lPto, + 40Pc + 120 Rs/hr
I tzs
I Cz= 0.l25Pzcz+ 30Po, + 100 Rsftrr
= 100
250
220 MW
Eru
o
tI 200
E 5 0
f I
3 150

E 100
o
0 5 0 100 150 200 250 J

Plantoutput,MW -_--> 50
Sunday Monday

Flg.7.4 Output of each unit versus plant output for Example 7.1 l l
12 1 2 6
(noon) PM (night) AM
Time -----
Fig. 7.5 Daily load cycle

For the plant describedin ExampleT.l find the saving in fuel cost in rupeesper Let us assumea daily load cycle as given in Fig. 7.5. Also assumethat a cost
hour for the optimal schedulingof a total load of 130 MW as comparedto equal of Rs 400 is incurredin taking either unit off the line and returning it to service
distribution of the same load between the two units. after 12 hours. Considerthe 24 hour period from 6 a.m. one morning to 6 a.m.
Solution Example 7.I revealsthat unit I should take up a load of 50 MW and the next morning. Now, we want to find out whether it would be more
unit 2 should supply 80 MW. If each unit supplies65 MW, the increasein cost economical to keep both the units in servicefor this 24hour period or to remove
for unit 1 is one of the units from service for the 12 hours of light load.
For the twelve-hourperiod when the load is 220 MW, referring to Table 7.1
165
165rnnn . ,^\rn rr.'tnZ ; lf\r1 rl Fn^ a D^tL- of Example 7.1, we get the optimum scheduleas
| \'U . L I ' r : r t *U)|JIl nr = (U.Ifnt t'tUI'6yll = llL.J l\S/I[
Jso "'
lso Pcr = 100 M W' Pcz = 120 M W
Similarly, for unit 2, Total fuel cost for this period is
[ 0 . 1x 1 0 0 2 + 4 0 x 1 0 0 + 1 2 0 + 0 . 1 2 5 x 1 2 0 2+ 3 0 x 7 2 0 + 1 0 0 ]x 1 2
J*co.rs" + 30)dpor=(0.r25PGz+
:oro;1"
", = Rs. 1, 27, 440
- - 721.875Rs/hr
 ltgtiFl
--t
powersvstemnnAVsis
Modern
If both units operatein the light load period (76 MW from 6 p.m. to 6 a.m.) -
t
also, then from the same table, we get the optimal scheduleas Dynamic Programming Method
Pcr = 20 MW, Pcz = 56 MW In a practical problem, the UC table is to be arrived at for the
complete load
Total fuel cost for this period is then cycle. If the load is assumedto increasein small but finite size stensldvnamin
(0.1 x 20' + 40 x 20 + 120+ 0.125x 5 6 + 3 0 x 5 6 + 1 0 0 ) x 1 2 prograrrurung can be used to advantage for computing the uc table,
= Rs 37,584 wherein it is not necessaryto solve the coordination equations;
while at the
same time the unit combinations to be tried are much reduced
Thus the total fuel cost when the units are operating throughout the 24 hour in number. For
these reasons,only the Dp approachwill be advancedhere.
period is Rs I,65,024.
The total number of units available, their individual cost characteristics
If only one of the units is run during the light load period, it is easily verified and
the load cycle on the station are assumedto be known a priori.Further,
that it is economicalto run unit 2 andto put off unit 1. Then the total fuel cost it shall
be assumedthat the load on each unit of combination of
duringthisperiot:tXf ': units changei'in
suitably small but uniform steps of size /MW (e.g. I MW).
762+ 3o x 76+ 100)x rz = Rs 37,224 Starting arbitrarily with any two units, the most iconomical combination
is
Total fuel cost for this case= L,27,440+ 37,224 determined for all the discrete load levels of the combined output
of the two
= Rs 1,64,664 units. At eachload level the most economic answermay be to
run either unit
or both units with a certain load sharing betweenthe two. The
Total operatingcost for this casewill be the total fuel cost plus the start-up cost most economical
cost curve in discreteform for the two units thus obtained,
of unit l , i .e . can be viewed as
the cost curve of a single equivalent unit. The third unit is now
1,64,6& + 400 = Rs 1,65,064 added and the
procedurerepeatedto find the cost curve of the threecombined
units. It may be
Comparingthis with the earlier case,it is clear that it is economicalto run both noted that in this procedurethe operatingcombinationsof third
and first, also
the units. third and secondare not required to be worked out resulting in
considerable
It is easy to see that if the start-upcost is Rs 200, then it is economical to saving in computationaleffort. The processis repeated,till all
available units
run only 2 in the light load period and to put off unit 1. are exhausted.The advantage of this approach is that having
oitiined the
gPli-u.| way,of loading ft units, it is quite easy ro determine the Jptimal manner
7.3 OPTTMAL UNrT COMMTTMENT (UC) of loading (ft + 1) units.
Let a cost function F" (x) be defined as follows:
As is evident,it is not economicalto run all the units availableall the time. To F,y (x) = the minimum cost in Rs/hr of generating .r MW by
N units,
determine the units of a plant that should operate for a particular load is the
problem of unit commitment (UC). This problem is of importancefbr thermal fN 0) = cost of generating y MW by the Nth unit
plants as for other types of generationsuch as hydro; their operating cost and F*-{x - y) - the minimum cosr of generating (.r - y) Mw by
the remain_
start-up times are negligible so that their on-off statusis not important. ing (1/ - t) units
A simple but sub-optimal approach to the problem is to impose priority Now the application of DP results in the following recursive relation
ordering, wherein the most efficient unit is loaded first to be'followed by the
less efficient units in order as the Ioad increases. FN@)= TnVn9) * Fu-r @ - y)| (7.r2)
A straightforward but highly time-consuming way of finding the most Using the aboverecursiverelation, we can easily determinethe
economicalcombination of units to meet a particular load demand,is to try all combination
of units, yielding minimum operating costsfor loads ranging in
possiblecombinationsof units that can supply this load; to divide the load convenient steps
from the minimum.permissible load of the smallest unit to the
optimally among the units of each combination by use of the coordination sum of the
canaeifies n f q l l $vs^rqurv
qrroiiol-lo rr-i+o i- +L;^ --^^^-^
uurrD. rrr LrrlD PruuttJs .ure
1^- 1-1 r .
total nunlmum oDerating eost
equaiions,so as to finci the most economicaioperatingcost of the combination; and the load sharedby each unit of the optimal combination are
then,to determinethe combinationwhich has the least operatingcost among all ;il.u,i"
determined for each load level.
these.Considerablecomputationalsavingcan be achievedby using branch and The use of DP for solving the UC problem is best illustratedby
bound or a dynamic programming method for comparing the economics of meansof an
example. Considera sample system having four thermal generating
combinationsas certaincombinationsneetnot be tried at all. units with
parameterslisted in Table 7.2.It is required to determinr th.
most-economical
units to be committed for a load of 9 MW. Let the load changes
be in steps of
I MW.
 l1 FE B/5 I

IJ>

--!
F-- - - - - _ _ _ ::"F-------
- - - ------=
- - - - -
- - - - - < - - - - =.c::,--
1

J,t, ~ l)? ?tf ft ~ tYYYUWhi

- - - - --------....---- -
..,_ __ . R....,,j+f-=~>G=----==-:--
)a, -__ ._' l f_
_f1_tu _W_JU L - -- -~_ _ _ _
_@_ ~ .:.____i.__ _

~@
.------ k

• -'- --- 4-- · ~! m l_ y, , C = ;!;_ C? ( IG/) /(

~
t------+--4_·;,~_f/~~'iU/£& we,/,_ fq Sa ~ · i!;f&,· -fo- /i -o
-- -- - - - - ---- --- --- r
r- -- r- -- - - - - - - - - ---
 I
PAGS ~No
~
..---.---1l ttl
lfliFf3 / 5 I ~DATE I 1 .
1

iffc;,('

• t; ::: ~~
J
I ,..

l.t~ J.. Li : J'n. ~ 1?-NrtLJm-

•
, - Ir L; {<Ulo-n tGOY.J d /M1~
,~ ~ d PL I'd ()~i-
a ')-eu'°£AI'{•
~ci
~tl
• Jrn ~ ~ CBe,-f; Q(cZe»e__cl t!)~ ~/1vrwe,,,
f
 ,
i
~ - ~ I>
d
- r-
~i
l1j_ - le
/1€EB !SI

rm -
,S,

f1,, · .
~
---
- - -
~-
~-~
---
rT:j
~ :1 -
---
t --r;_--=- ::iI<. -- cK c:_ fim &-n fr;, 'Yl -
•., - - - ,?Yr-f-l ?r"f - --
-~
t

t
-
8,wn ~!3,,,, ""~ " )._M,!J
l,VYUf.a ~
) M4tA/Wl.
;;!l'e/e,n::t
l>'n4
-
wke, t;& ~
-- -
. -
- ---- - --- -
'"T' -. - - - ----;; ~

ft ~ P~' B P6i Prs, ::: -- PG, . 6~ ,.... Bu B,2. -· -6,~

I
J '
PG,_
- - --
6,__, &1--·· 82.K
,-- ~

I
. ---- . - >--

-
P67(l( ) /31<1 Bt2. ----Bt< ~
• ,.
"
~1.)
(_k XI<
-t> k 3,f~.· ' ·1

u
l
t I
-
PG7,
-6t, 611- 1613
~

• fa-= fGi2- ~
0-= B:ir &1- 82-3 ...
-
('
~

\
Pbis B3, &2.
•,, I
- 831
I ..
' I
I ,. . PL= I Pc;, P02- ~2 J 8t,6,-< .B,3 P~,
~ I
I
&, 0,.~ g~'.3 f'b~
• I ...... 631 83A (333 - fGt.s -
-
'
~
= fr~. PGJ~ ptr/j I 611 P~, + Br~ P<,,~ t- 813 P6 ~
&1 Pr,,, + 8:).iP- (?r,.,:!l + B-< 2 f& 3
~
~ p~I
Ill"

- -t- 8:3;;_ f Gi;;_ + B=i3 P(,,,'J

'
L
-::t> PL_: 811 P6J, 2:·
+__B,~ PGt, f~ + /3,3 Prs,, f G-, 3 . + B~,PGr, _ t
- - - 2....
- - - -

@~ :> {It -t- B~.9 Y'e,,.,_P6i;i + f33i f Gi, pG-,3 t- 83~ f cr, z. 6:,,3- t. 633 ~
il -~ - - - - ;;;z. - - 2.

Ii
r-> /L_== - ~!.Lf~) -,. B~-" f'Gzt. -
+ 833 fG,$
'2-

_+ ~ B,~ f&,/G:J _+ ~B,~_P~, fr:,, 3

-- -

+ ..?(k3 ~f(n.3_
i - I

.,
~ - -
I

.I
I
 l1~EBt5J
[SGENo]
I I I l
.7 --
DATE
. J.,;,

'. ~ ({ 1 ~ 2!- -
_J f'6m
- -
Bm.,,
-
~
} fr,,,- JI},,/ ) "n1~-I - -
71:: I
- ~

- --- - - -
1'( 0
(b-taJ ?1~ .ff fvrm4 =- - - -
- -k -
- ;; r-f- c}- f~M
- -
BmJ;i Pt:,,1)2_ ± 71~ /6);_6/71 ~.2?
J fGz/ (_ ,;:;:~• ;:k - "'V: {.
1
-
... §- t.,,,, 6mJ. ,1,a + l1t/ Bt IJ.,,,-
.ryn-:::I
7 lt,h/. , y y , ~- ::; (k~1Y--
"Tn:/: i
- I
k k

-!"+ ~ 'Yl 2(
B/IJ? f<:,,rn +2 ~~/lm ~ ( + r:<Bri·~, .. )
l 'Yl.:;il
"n?:=z I .,
,rn-1-1 I
' -J
r k 7
I dfL ~
~~ e< Br1· f~t
3:::2( u V
I
I J

J fG?,- 7:= !, 2., -~--k..,

I
I I

'L..

➔ c"( -= I a/ f4, .. + fot 16,;-

!
-f- cl./
~

-=P cJ.u- ~
a/ f k." -f- bt I
d.P6u

k
b/== ~ ( r -
C-t/ f)&z; +
,-, -< B(i f01 i
J~

k
V (,/ /
)
- - -
-h atf&i +b( + ) ~Z__ ~ B,j· f ~ ~;; ---
o=r
-k:- - -
4," f1nt" -+ h/-+ ~ ~ f3/tff!IH
I
_<1 -~ ~ 8(J. ~· :::: A
v_? I
--J4 i

--
......
 /1ff/3l 5J

<O

-A - P. > o
k.

-------1