Electric Power Systems Research, 21 (1991) 147 - 153 147

Simple Algorithm for Economic Power Dispatch

C. P A L A N I C H A M Y and K. S R I K R I S H N A
Department of Electrical Engineering, Thiagarajar College of Engineering, Madurai 625 015, Tamil Nadu (India)
(Received October 11, 1990)

ABSTRACT tion of 2. Nagendra Rao [9] has proposed a

method which is simple and accurate to apply,
A straightforward and computationally but it does not consider the transmission
efficient method is presented for economic power losses. Ramaraj et al. [10] have presented an
dispatch. The familiar B coefficients are used for analytical approach which reduces the time-
the evaluation of transmission losses, incremen- consuming iterative procedure into a mere
tal transmission losses and penalty factors. The positive-root determination of a third-order
method of evaluating the incremental transmis- polynomial in ~. Several approximations, such
sion losses is justified by means of test results. as (i) using the approximate penalty factor
The algorithm is tailored suitably to accommo- method, (ii) modifying the B-coefficients by
date system load as well as network topology assuming Pm = Pn, and (iii) ignoring a number
changes. A sample system is tested to show the of terms in the binomial expansion, have been
computational capability and the proposed ap- made, but they lead to greater deviations in
proach was found to produce a fuel cost which fuel cost and power loss from values found by
deviates less from the accurate value than a the conventional exact method. Also, the co-
previous direct method. efficients of the third-order polynomial contain
a greater number of terms which require exten-
sive off-line computation time. The positive-
1. I N T R O D U C T I O N root determination takes almost as much time
as the iterative method. The application of a
The best way to minimize fuel consumption quadratic form for direct economic power dis-
is of great concern in the present situation. patch has been discussed in ref. 11.
Considering the prevailing energy crisis and In this paper, a new formulation of the
the fast diminishing fuel reserves, any method dispatch problem is derived. The resulting
which leads to minimum fuel utilization for equation is simple and gives the plant genera-
energy production is definitely worth trying. tions directly. The method takes into account
Many methods [1 - 5] have already been used the changes in network topology and system
in practice for the solution of the coordination loads. The B-coefficients are used for the deter-
equation, taking into account the transmission mination of line losses and incremental trans-
losses. A survey by Happ [5] provides a very mission losses. Computational experience
complete review of this subject. Hill and indicates t hat the approach is highly efficient
Stevenson [6] have suggested an effective and is capable of solving large size problems in
method to obtain the incremental transmission a time short enough to be compatible with
losses by the B-coefficients. on-line applications.
Recently, some new analytical methods
have been developed t hr ough which the time-
consuming established iterative way of iden- 2. P R O P O S E D M E T H O D
tifying the generation schedule has been
eliminated. Alvarado [7] calculated the pen- In order to determine the economic viability
alty factor from the transposed Jacobian by a of power dispatching, it is necessary to com-
direct and fast way. R a m a n a t h a n [8] formu- pute the fuel cost. For a dynamic dispatch
lated a closed-form expression for the calcula- scheme, the solution has to be fast and reliable.

0378-7796/91/$3.50 ~t" Elsevier S e q u o i a / P r i n t e d in The N e t h e r l a n d s

The computational strategy should be to ap- obtained. This schedule is obtained directl\
proach the optimal operating point rather without using any complicated solution tech-
than the pure mathematical minima. nique and hence the computer run time is ver!
The proposed algorithm consists of two small. The suboptimal solution is necessary
parts. In the first, a suboptimal schedule is for the initial estimation of network losses
obtained (generations corresponding to the and incremental transmission losses.
loss neglected case) for the evaluation of ini-
tial network losses and incremental transmis- 2.2. Incremental transmission loss evaluation
sion losses. In the second part, actual line The following parameters are defined.
losses and incremental transmission losses are
p, generation of ith plant corre-
determined from their initial estimates, by
sponding to loss included case
simple means. Then, a direct method for the
PP generation of ith plant corre-
optimal generation when transmission losses
sponding to loss neglected case
are considered is discussed.
(suboptimal generation)
p:: transmission loss computed with
2.1. Suboptimal schedule
values of loss neglected case gen-
Assuming quadratic fuel cost curves, the
erations Pp (initial power loss)
methodology for the case neglecting loss [9] is
APL change in transmission loss due
discussed first. Let Pp be the generation of the
to change in generation from PI’
ith plant corresponding to the loss neglected
to P,
case, Pz be the generation of an arbitrary
((:P,,/?P; )” incremental transmission loss of
plant k (k < n), n be the number of active
ith plant evaluated from subopti-
generators in the system, and a,, b,, ah, b, be
ma1 schedule (initial estimate of
the fuel cost coefficients of plants i and k,
incremental transmission loss)
respectively (k = 1,2, . , i, . , n).
((:P,,/(:P; ) = ITL,, incremental transmission
The optimal schedule is obtained when the
loss corresponding to optimal
incremental costs of all the plants are equal.
schedule (loss included case)
Therefore
L, penalty factor of ith plant com-
2a, Pe + bi = 2a, PW + b, (1) puted with value of (?P,,/?P,)
PI, system demand
or
In the usual way for economic dispatch
Pp = (2a,PO, + b, - b,)/2a, (2) equations, a vector of incremental losses asso-
Substituting eqn. (2) in the power balance ciated with the generating units is used to
equation include the effects of system losses. A vector
of these quantities exists for each set of load
i Py = P,, conditions, system parameters and constants.
1=1 Palanichamy and Srikrishna [12] have pre-
sented a new equation for updating the incre-
we get
mental transmission losses as a function of
changing load. From various test results, they
i (2a,PO, + b, - b,)/2ai = P,, (3)
i=l showed that the relationship between the
loads and incremental transmission losses is
Simplifying eqn. (3), a direct expression for
linear for a load change of up to *20X from
the optimal generation for any plant k in
the base load:
terms of the system parameters and the total
demand is obtained:
(5)

where
(4)
Py” = P$” f 20%

Using eqn. (4), the suboptimal schedule (gen- In this paper, eqn. (5) is modified slightly but
eration schedule for the loss neglected case) is its general structure remains the same.
 149

TABLE 1
C o m p a r i s o n of i n c r e m e n t a l t r a n s m i s s i o n losses

Method Incremental Load (MW)

transmission
losses 200 300 400 500 600

C o n v e n t i o n a l a c c u r a t e m e t h o d [1] ITL~ 0.00916 0.01939 0.02932 0.03935 0.04947

ITLu 0.01153 0.01967 0.02793 0.03628 0.04475
ITL a 0.00246 0.00363 0.00483 0.00608 0.00738
ITL 4 0.02340 0.03640 0.04945 0.06255 0.07570
ITL~ 0.01780 0.02338 0.02905 0.03481 0.04067
ITL 6 0.02742 0.03249 0.03762 0.04283 0.04870

P r o p o s e d m e t h o d u s i n g eqn. (7) ITL~ 0.00852 0.01821 0.02798 0.03784 0.04780

ITL2 0.01105 0.01901 0.02704 0.03514 0.04332
ITL:~ 0.00224 0.00324 0.00424 0.00526 0.00628
ITL 4 0.02416 0.03763 0.05122 0.06495 0.07880
ITL 5 0.01792 0.02334 0.02880 0.03432 0.03989
ITL~ 0.02833 0.03346 0.03864 0.04388 0.04916

First, the incremental transmission loss Since incremental transmission losses, and
(OPI,/~Pi) ° of the ith plant is evaluated from hence penalty factors, are quantities which
the suboptimal schedule using the expression depend on the network topology, they have to
~Pi ~° be determined for every change in a system
~ / ] ~-'21~¢=1BikPk (6) network topology.

where Bi~ are the transmission loss co- 2.3. Transmission loss computation
efficients. The desirability of including transmission
The initial network loss p0 is estimated line losses in economic dispatch calculations
from the suboptimal generations and it is has been clearly demonstrated [1 - 3, 5]. Gener-
treated as a load in addition to the base load. ally, line losses are calculated either by the
Then ((~PI,/(~Pi) °, the computed incremental B-coefficient approach [1] or by load flow
loss, is updated using the equation techniques [4, 5]. Here, the B-coefficient ap-
proach is adopted.
,~p~/=\-~j\ ~ ) (7) The total power loss PL is expressed as a
function of the initial line loss p0 and the
The relationship in eqn. (7) between (SPL/~Pi) change in loss AP L as
and (SPI•/(~Pi)° has been justified by means of
test results. For the test system [10], with PL = P ~ + A P L (8)
various load conditions, the incremental
transmission losses were evaluated by an ac- The change in loss is due to the change in
curate conventional method [1]. The incre- generation from the suboptimal value p0 to
mental transmission losses for the same the optimal generation Pi and is expressed as
system were also calculated by eqn. (7), and
are shown in Table 1. AP~• = ~=~
~ ~~ P( PL i -P°) (9)
From the results of Table 1, it is found that
the incremental transmission losses obtained
Substituting eqn. (9) in (8),
by eqn. (7) of the proposed algorithm coincide
well with the incremental transmission losses
PI = po + £ ~PL (10)
of the accurate conventional method. • " i=,~(Pi-P°)
Equation (7) has been applied for the evalu-
ation of incremental transmission losses for Since
many similar systems, and has been found to
be computationally efficient and to give accu- ~PL 1
--1
rate values. t~Pi Li
150

eqn. (10) becomes Read fuel cost coefts.,

system demand and
loss coefts.
PL= P~ + 1- L (P~ - P~/) (11)
i=l

IObtaln sub-optimal schedule

Equation (11) gives the total transmission loss using eqn. (h)
at optimal dispatch, which is a function of the
network topology dependent penalty factor,
the initial power loss and the change in plant Compute PL and ~ PL )o
generations. Whenever there is a change in
system demand and/or change in network to-
].with sub-optlmal generations
pology, the transmission losses must be recom-
puted.
PL • to (--~-~-)
Update (-~.~-) ~ PL
2.4. Optimal generation
When transmission losses are considered,
the optimal generation [1, 2] is obtained if using eqn. (7)

(2aiPi + bi)Li = (2akPk + bk)Lk (12)

or I Compute Pk (k = I, 2, i, ....n)
by eqn. (17) and total fUel
1 bl cost.
Pi = (2akPk + bk)Lk (13)
2aiLi 2ai
The total generation is the sum of the individ-
ual plant generations. Therefore,

P~=(2akPh +bk)Lk ~ 1 ~ bi
i=1 i=12aiLi i=l~a i (14)
From the power balance equation, the total
generation is also equal to Fig. 1. Flow c h a r t for e c o n o m i c p o w e r d i s p a t c h .

~ Pi = PD +PL (15)
i-1 can also be obtained by the proposed al-
Substituting eqn. (11) in (15), gorithm. There is no restriction on the num-
ber of generators (n) and the method is
suitable for a general problem comprising n
~ PI=PD+P°~+ ~ (1-~)(Pi -P°) (16)
i=1 i=l generators. The derived equation for the opti-
mal generation is simple, non-iterative, less
Equations (2) and (13) are substituted in eqn.
time consuming than other methods, accurate,
(16) and equated to (14). Regrouping similar
and gives a guaranteed solution.
terms and simplifying, we get

Pk= P D + P ° ~ + i =~, ~iai -(2a'~P°+bk) 3. P R O P O S E D S O L U T I O N A L G O R I T H M

bk
1--Li/J2~kLki~l 2 a i L i 2 - -2a~
1 . A concise review of the proposed algorithm
~~
X i=I - is presented in the form of a flow chart in
(17) Fig. 1.

Equation (17) gives a direct expression for the

optimal generation of any plant k (k ~<n) in 4. TES T S Y S T E M A N D R E S U L T S
terms of system parameters, initial suboptimal
generations and system demand. Changes in The proposed method is applied to a six-
system demand and network topology necessi- generator system [10] to assess the suitability
tate new optimal generation scheduling which of the algorithm.
 151

TABLE 2
Losscoefficient maxtrix ~rthetestsystem

0.0002 0.00001 0.000015 0.000005 0.0 -0.00003

0.00001 0.0003 -0.00002 0.000001 0.000012 0.00001
0.000015 -0.00002 0.0001 0.00001 0.00001 0.000008
0.000005 0.000001 -0.00001 0.00015 0.000006 0.00005
0.0 0.000012 0.00001 0.000006 0.00025 0.00002
-0.00003 0.00001 0.000008 0.00005 0.00002 0.00021

T h e fuel cost (in U.S. S/h) e q u a t i o n s for the T a b l e 3 gives the test r e s u l t s of the pro-
six g e n e r a t i o n s are posed m e t h o d c o m p a r e d w i t h the a c c u r a t e
c o n v e n t i o n a l i t e r a t i v e m e t h o d [1] a n d the
F1 = 0.005P12 + 2.00P1 + 100 m e t h o d from ref. 10.
F2 = 0.010P22 + 2.00P2 + 200 T h e test r e s u l t s of the p r o p o s e d m e t h o d
c o r r e l a t e well with the a c c u r a t e one. The per-
F:~ = 0.020P:, 2 + 2.00P:~ + 300 c e n t a g e d e v i a t i o n s in the fuel cost a n d p o w e r
F4 = 0.003P42 + 1.95/)4 + 80 loss, with r e s p e c t to the a c c u r a t e results, are
s h o w n in the last two c o l u m n s of T a b l e 3. F o r
F.5 = 0.015P.~ ~ + 1.45P~ + 100 a load of 200 MW, t h e r e is p r a c t i c a l l y no devi-
a t i o n in fuel cost by the p r o p o s e d method,
F(~ = 0.010PG 2 + 0.95P~ + 120
w h e r e a s t h e r e is a d e v i a t i o n u s i n g the m e t h o d
and the loss coefficient m a t r i x is g i v e n in from ref. 10. W h e n the l o a d increases, the
T a b l e 2. p e r c e n t a g e d e v i a t i o n in fuel cost by the

TABLE 3
C o m p a r i s o n of t h e test r e s u l t s

System Method P/ P2 P:~ P4 P~ P6 PI, Fuel cost Cost P o w e r loss

demand (MW) (MW) ( M W ) ( M W ) (MW) (MW) (MW) F t (U.S. S/h) d e v i a t i o n deviation
(MW) (%) (%)

200 Accurate iterative 30.95 15.25 8.15 54.54 28.01 65.29 2.19 1289.47
m e t h o d [1]
Proposed method 31.22 15.32 8.17 54.30 28.01 65.17 2.19 1289.47 0 0
M e t h o d from ref. 10 30.20 15.40 8.20 54.90 28.10 65.40 2.20 1289.46 0.0008 0.4566

300 Accurate iterative 55.26 27.59 14.85 93.06 36.41 77.36 4.53 1536.15
m e t h o d [ 1]
Proposed method 55.64 27.72 14.88 92.64 36.43 77.21 4.52 1536.16 0.0007 0.22
M e t h o d from ref. 10 54.60 27.80 14.80 92.90 36.50 77.70 4.30 1535.56 -0.038 5.077

400 Accurate iterative 79.81 40.11 21.73 131.70 44.97 89.58 7.90 1810.37
m e t h o d [1]
Proposed method 80.32 40.30 21.79 131.02 45.03 89.40 7.87 1810.36 0.0006 0.3797
M e t h o d from ref. 10 79.20 40.50 21.70 130.90 45.20 90.30 7.80 1810.15 -0.0122 1.2658

500 Accurate iterative 104.60 52.79 28.80 170.44 53.69 101.97 12.29 2112.98
m e t h o d [1]
Proposed method 105.27 53.06 28.90 169.45 53.80 101.77 12.25 2112.97 -0.0005 0.3255
M e t h o d from ref. 10 104.00 53.20 28.70 169.00 53.90 103.00 11.80 2111.56 -0.0672 3.9870

600 Accurate iterative 129.64 65.65 36.08 209.29 62.57 114.53 17.76 2444.86
m e t h o d [1]
Proposed method 130.50 66.03 36.23 207.89 62.75 114.30 17.71 2444.80 0.0008 0.2815
M e t h o d from ref. 10 129.00 66.00 35.90 207.10 62.80 116.10 16.90 2442.15 -0.1109 4.8423
 152

TABI,E 4
0.01 L~20
Demand i n MW
0 300 bOO 500 6 O0 Core and time assessments
0.0

Method
-0.02
Accurate iterative M e t h o d from Proposed
m e t h o d I1] ref. 10 method
-0.~
Memory
(bytes) 5380 4636 4063
-o. c6 Run time

-0.o8
Method
{10)~ (s) 255 159 112

300 MW, whereas the p r e s e n t m e t h o d offers a

-o.Io
d e v i a t i o n of only 0.22% for the same load.
T a k i n g the c o n v e n t i o n a l m e t h o d as refer-
-0.12 ence, graphs showing (i) the system demand
and the p e r c e n t a g e deviation in fuel cost and
Fig. 2. D e m a n d vs. p e r c e n t a g e d e v i a t i o n in F t .
(ii) the system d e m a n d and the p e r c e n t a g e
deviation in power loss for the proposed
m e t h o d from ref. 10 becomes large, being more m e t h o d and t h a t of ref. 10 are illustrated in
t h a n 0.1% for a load of 600 MW. Figs. 2 and 3, respectively.
Using the proposed method, t h o u g h t h e r e The c o m p u t e r ( F P 200 CASIO) core storage
are some deviations in the fuel cost c o m p a r e d and r u n time for all t h r e e methods are as-
with the a c c u r a t e method, they are less t h a n sessed in Table 4.
0.001%, which is very low. The deviations in F r o m Tables 3 and 4 it is observed t h a t the
the fuel cost by the proposed m e t h o d and t h a t proposed m e t h o d has good p o t e n t i a l for eco-
of ref. 10 t u r n e d out to be negative because of nomic power dispatch since the results are
the a p p r o x i m a t i o n s made. The l a r g e r the num- very a c c u r a t e , with lower core storage and r u n
ber of a p p r o x i m a t i o n s , the h i g h e r the devia- time r e q u i r e m e n t s t h a n the o t h e r methods.
tion from the most economical condition.
Also, the d e v i a t i o n in t r a n s m i s s i o n loss by the
m e t h o d of ref. 10 is a r o u n d 5% for a load of CONCLUSION

A s t r a i g h t f o r w a r d and c o m p u t a t i o n a l l y
efficient m e t h o d is p r e s e n t e d for economic
power dispatch. The results are very a c c u r a t e
and yet the solution a l g o r i t h m is very simple.
The a l g o r i t h m for the loss neglected case
dispatch is also discussed because the initial
h suboptimal schedule for the final solution is
o b t a i n e d from it. Results are g u a r a n t e e d be-
cause the optimal solution is a function of the
o" 3 suboptimal scheduling and the c h a n g e in
power loss is easily a c c o u n t e d for. Both the
2 off-line and on-line c o m p u t a t i o n times are less
t h a n those of the c o n v e n t i o n a l i t e r a t i v e
m e t h o d and t h a t of ref. 10. C o m p u t a t i o n a l
Proposed
a~ e x p e r i e n c e with the proposed m e t h o d indi-
cates t h a t the m e t h o d is very efficient and is
capable of solving large-size problems with
200 300 400 500 600
the expected a c c u r a c y in a time which is short
Demand i n F,},
e n o u g h to be compatible with on-line applica-
Fig. 3. D e m a n d vs. p e r c e n t a g e d e v i a t i o n in p o w e r loss. tions.
 153

ACKNOWLEDGEMENT sive survey, IEEE Trans., P A S - 9 6 (1977) 841- 854.

6 E. E. Hill and W. D. Stevenson, A new method of
determining loss coefficients, I E E E Trans., P A S - 8 7
The authors are grateful to the authorities
(1968) 1419- 1425.
of Thiagarajar College of Engineering, 7 F. L. Alvarado, Penalty factor from Newton's method,
Madurai, for the facilities provided. I E E E Trans., P A S - 9 7 (1978) 2031 -2037.
8 R. Ramanathan, Fast economic dispatch based on
penalty factors from Newton's method, IEEE Trans.,
PAS-104 (1985) 1624 - 1629.
REFERENCES 9 P. S. Nagendra Rao, Direct method of determining
optimum generation schedules, J. Inst. Eng. (India), 67
1 L. K. Kirchmayer, Economic Operation of Power Sys- (1987) 251 - 253.
tems, Wiley, New York, 1958. 10 N. Ramaraj, R. Rajaram and K. Parthasarathy, A new
2 0 . I. Elgerd, Electric Energy Systems Theory A n In- analytical approach to optimize a generation schedule,
troduction, McGraw-Hill, New York, 1982. Electr. Power Syst. Res., 11 (1986) 147 - 152.
3 V. Venikov, V. Zhuravlov and T. Filippova, Optimal 11 C. Palanichamy and K. Srikrishna, A quick method to
Operation of Power Plants and Electric Systems, Mir, optimize generation schedules, J. Inst. Eng. (India), 69
Moscow, 1984. (1988) 99- 104.
4 H. W. Dommel and W. F. Tinney, Optimal power flow 12 C. Palanichamy and K. Srikrishna, A method for
solutions, IEEE Trans., P A S - 8 7 (1968) 1866-1876. short-term generation redispatch, Electr. Power Syst.
5 H. H. Happ, Optimal power dispatch a comprehen- Res., 17 (1989) 129- 138.