You are on page 1of 6

A HYBRID MODEL FOR OPTIMAL POWER FLOW

INCORPORATING FACTS DEVICES

Narayana Prasad Padby M. A. R. Abdel-Moamen P. K. Trivedi B. Das

Dept. of Electrical Engineering


University of Roorkee
Roorkee -247667
India

Email: ~

Abstract: Optimal power flow (OPF) is one among the most decreases the overall stability of the system[2]. To achieve
important algorithms available to utility for generating least cost better power flow control over the transmission lines without
generation patterns in a power system satisfying transmission and
violating the stability margin of the system, application of
operational constraints. In day to day life, the forecasted loads used
flexible ac transmission system (FACTS) technology is
in classical OPF algorithms are increasing with time and are also not
currently being pursued very intensively [3]. FACTS
completely free from errors. Increase in load demands leads to
technology is essentially the art and science of achieving
overloading of the transmission lines and forecasted errors cause
loss of optimality. So classical optimal power flow algorithms may better controllability over various electrical quantities in a
not be able to provide optimal solutions and they are limited to power transmission system. Different FACTS device, such as
fixed type of cost characteristic curves and can not handle frequent static var compensator (SVC), solid state synchronous
increase and variation in load demands. in this poper a ncw hybrid compensator (STATCOM), thyristor controlled series
model for OPF incorporating FACTS devices has been proposed to capacitor (TCSC), unified power flow controller (UPFC) etc,
overcome the above said diftlculties. In the proposed model load
are among the most potential controllers for application to
demands, generation outputs and cost of generation are treated as
power system to achieve better controllability. SVC and
fuzzy variables. la the first stage fuzzy dynamic programming has
STATCOM essentially control the voltage of a bus in a
been used to determine a set of feasible generation patterns. In the
second stage dc power flow algorithm has been used to determine a
system. TCSC essentially controls power flow over a line and
set of valid sub-optimal generation patterns satisfying transmission UPFC controls both the bus voltage and power flow over a
nnd opcmtionnl constraints. Finally in the third stage over loading line.
of transmission lines can be eliminated by installing FACTS
devices in the system. To verify the validity of the proposed model If the transmission capacity constraints are
a 5 generator and 10-bus test system with and without modification included[4], there is a possibility that the optimum generation
has been used. It was found that the solutions obtained are quite
point may shifi to satisfy the transmission line capacity
encouraging and suitable for modem deregulated environment.
constraints. At any operating point of a transmission system
there must be some transmission lines through which the
Key Words: Optimal Power Flow, Fuzzy Logic atui Facts Devices
actual power flow is quite small compared to their
operational limits. If the power flow through the under
I. INTRODUCTION
utilized lines can be increased by putting a TCSC in the lines,
then power flow in the over-loaded lines may be diverted to
Optimal power flow is one of the most important
those under utilized lines and consequently, the power flow
operational functions of the modern clay energy management
in the overloaded lines may reduce and ultimately come
system. The purpose of the optimal power flow is to find the
down to a level wkch is below their operating limits.
optimum generation among the existing units, such that the
total generation cost is minimized while simultaneously
II. PROBLEM FORMULATION
satisfying the power baluncc equations and various otlwr
constraints in the system. The most common constraints in
Mathematically, the optimal power flow problem[5] can be
the operation of a power transmission system are constraints
formulated as follows :
on the voltage magnitudes of the buses and constraints on the
The objective is to minimize the total cost of generation. i.e.
reactive power genemted by the generators. For reliable
operation of a power system, the voltage magnitudes of the
minimize
buses and the reactive power[ 1] generated by the generators
are constrained to stay within specified limits. However, due FT = ~Fi(PGi ) (1.1)
to the increasing load demand, more and more power is iaf
required to be pushed over the existing transmission lines but where,
increase in the power flow above a certain operating level

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 510
0-7803-6674-3/00/$10.00 (C) 2000 IEEE
III. PROPOSED MODEL
FT = Total cost of generation($/hr)
Fi(f’Gi) = The cost of PGi generation by i’h generator($/hr) However, in practical scenario, the load demand is
NG = Number of generators not always known precisely [(i], rather an approximate value

Fi(pd
= aip~i
+bi PGi+Ci($~) (1.2) of the load demand is more likely to be known. In other
words, there is an uncertainty or vagueness associated with
Where, ai, bi, Ci are the cost coefficients for i’” generator. the system load demand. This uncertainty can not be
adequately expressed by a crisp variable. On the other hand,
The above minimization problem is subjected to this vagueness can be adequately handled in fiuzzy set theory.
certain system constraints. The most common constraints are This theory provides a strict mathematical frame work in
which vague conceptual phenomenon can be studied
● Active Power Balance Ecmation for the Svstem rigorously. In this theory, the variables, functions, etc.
The total generation in the system must be equal to connected with the imprecise phenomenon to be studied are
total load plus the total loss in the system, i.e. expressed as fuzzy variables and fuzzy functions.
Consequently, to solve optimal power flow under imprecisely
defined variables such as load demand etc., the objective
function and all or some of the constraints need to be
expressed as fuzzy objective function and fuzzy constraints.
This uncertainty in load demand and generation output and
where, PIO,~= Total active load in the system production cost are treated as a fuzzy variables and different
P,.,, = Total active power loss in the system suitable membership functions are chosen.

● Limits on the Outtmts of the Generation Units In this paper, the system load demand has been
The output of each generating unit must be within classified into three categories ; low, medium and high.
some specified minimum and maximum limits, i.e. Consequently, the total generation capacity of an individual
generator is also classified into three categories. The
membership functions of generation capacity levels versus
P&i” < pGi < P:’” ,fori= 1,2, . ..NG (1.4)
different load levels have been shown in Fig. 1.1.

The membership function chosen for the cost is written as.


where,
P Cli = The unit MW generated by i’” generntor

P~’” = The specified minimum MW generation by pC(i,j) = exp(-Ac(i, j)) (1.6)


ith generator

PGI
‘ax = The specified maximum MW generation by
where,
i* generator

AC (i,
j,= C(i, j) - Cm,.(i, j)
. O~eratin~ Line Constraints
The power flow over a transmission line should not Cmin(i,j)
exceed the specified maximum limit because of
where
stability considerations, i.e.

WC (i, j) = The cost membership value for i’h bus


PJj~ Pyax i=l,2.., ........n
(1.5) generator~witb j’h discrete generation (P~ij)
j=l,2 .......... . n
C(i, j) = cost per MW for i* bus generator with j’h
where,
discrete generation given by the expression
pij = active power flowing in line joining i’h & j’h bus
C(i, j) = ai P~ij + bi + ci/p~ij
p ~’ = maximum allowable active power flow in
C,,,i,,(i, j) = minimum cost per MW for i’h generator
line joining i’” & j’” bus
n = number of kmscs in the system bus with j’” ciiscrete generation,

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 511

0-7803-6674-3/00/$10.00 (C) 2000 IEEE


Load
Low Medium High
Generation
\(MW)

‘ow (XA ,G(xl~ ,G,x)l\

P mm P P P

,G(x~ ;pm= ,G(;A””

Medium

P;1 I,,llm ]>,,,u


-h
,;;,> ~~
,G::>p,,,w

High

P,,,i,, PIlmx Pmm ma% min

Figure 1.1: Membership functions for generation level (p~(.)) versus load demand levels(P)

Load demands and MW generation of the generators L~i” + minimum load demand of the system
have been classified into three categories based on the
L.m -+ maximum road demand of the system
following formula.
L~i~ = (L~flX - L~in)
. Load in the range : (L,,,,,,) - (L,,,im+ 0,25LJ,)
The generators have been classified into three
+ low range
different categories, based on the following criteria.
,. Loads in the range :(L~in + 0.25L,,J - (L~u - 0.25 L,,,)
i) Low range : P~.X <50 MW
+ medium range
ii) Medium range :50 MW < P~”,oXs 150 MW
● Loads in the range : (L.,n, - 0,25 L,,,) - (Ln,m)
iii) High range : PG.U >150 MW
-+ high range
Where, Pc,mw -+ maximum generating capacity of a
where,
generator.

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 512

0-7803-6674-3/00/$10.00 (C) 2000 IEEE


IV. PROPOSED HYBRID MODEL ALGORITHM Step 10: Check the line and other constraints.

The step by step algorithm of the proposed hybrid model is Step 11 : If the line and other constraints are within their
presented as follows: individual limits, then generation and cost of the set (Pol,
PG2- . . . . P~o) is belongs to the most economic solution.
Step 1: Read all inputs data.
Step 12: If line constraints violated, select the next lower of
(a) Generation units, P&’n , P ~, for i = 1,2 . . . . . .NG menlbersllip ~ ,,~vand go to step 8.

(b) Transmission line limits p~ for i = 1,2 . . ... . .n Step 15: Install the FACTS devices in the lines with over
loading and repeat step 1 to 12.
j= 1,2 . . .....n

(c) Cost coefficients ai, b,, c, for i = 1,2,.. ..N(3 V. RESULTS AND DISCUSSION

(d) Line parameters such as resistance, reactance, line To illustrate the application of the proposed model a
charging susceptance and FACTS device parameters for 5 generator, 10-bus system has been considered(refer
line between i{h and j{~ bus. appendix – A). OPF problem has been solved using different
membership functions for generators and their corresponding
Step 2: Set PIW,= O.
cost timctions taking into account of uncertainties in the load
Step 3: Pflay = PIO.~+ Plw, demands.

It has been found that observed that for the most


Step 4: Set the generation of all the generators (PGl,P~2,
optimum generation pattern, the total generation cost is
P~3,. . . . . .P~~) such that
139.078583 units but unfortunately two lines, namely lines 1-
9 and 3-7 are violating the maximum power flow constraints,
However, it is observed that for the 36’h sub-optimal
Step 5 : Determine PC ( i, j ) and ~G(i>j) generation pattern, all the system constraints are satisfied.
‘or all ‘e
The results under this condition are shown in Table 1.3 and
generation patterns obtained from step 4 and calculate
1.4.
~R (i, j) for each p(j,, i ENG USing the following formula.
To explore the possibility of using TCSC to relieve

PRk (d, k(i,j))


(i,’j)=min
(Pc the overloaded lines, OPF problem has been solved using
‘1’CSC at various lines with vurying compensation levels. It
fork= 1,2,3 . . . . . ..NG was found that for two TCSC installations, all the system
constraints are satisfied. These two cases are :
Step 6: Calculate the set membership value (SMV) for all
the feasible combinations and arrange them in the decending a) TCSC installed on the line 1-6 with 50%
order. compensation and
..
Psmvw =maximum ()
( ~1 1, J , p2(i,j), . . . . . . . . . p~(i,j)) b) TCSC
compensation
installed
levels.
on the line 2-6 with 30%

forw=l,2,3 , . . . . . number of generation sets


From these above case studies, the total generation
Step 7 : Select the generation pattern corresponc@g to the costs are less than that obtained at the operating condition
highest membership value(obtained from step 6) and run the given in Table 1.1. Hence both these cases are economically
viable as in both these cases, the cost of TCSC installation
DC power flow and calculate the corresponding p~~
would be offset by the savings accrued in the generation cost.
However, in this @ase(a) the generation cost is less than the
Step 8: Compute error= lPl~, - P~~ 1.
cost in case (b). So solution set (a) is preferable as the savings
in this case is higher and consequently, the cost of the TCSC
Step 9: If error >6 (to]crancc), set P,,),, = P~J~ and go to installation would be offset more quickly. I Ience, from the
step 3. Or otherwise go to step 10. above discussion, TCSC with 50% compensation on the line
1-6 is recommended.

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 513

0-7803-6674-3/00/$10.00 (C) 2000 IEEE


Table 1.1 : Optimum generations and their corresponding Table 1.4: Line flows corresponding to

costs. the generation pattern of Table 1.3

Line joining Line flow (p.u)


Unit No. Generation Generating cost (units) Bus Bus
(MW) 1 2 0.1285940
1, 85.00 29.816999 1 6 0.1768700
2. 70.00 33.839001 1 9 0.2713670
3. 55.00 26.161751 2 3 0.0709960
4. 55.00 28.252501 2 6 0.2180210
5. 59.110260 24.008343 3 7 0.2878110
4 7 0.2124870
4 8 0.1061760
5 6 0.145000
5 10 0.1738680
Table 1.2: Line flows for the optimum generation pattern 6 9 0.0099870
8 10 0.046722
Line joining Line flow (pu) 9 10 0.0658310
Bus Bus
1 2 0.1346941
2 6 0.146318
VI. CONCLUSIONS
1 9 0.308809
2 3 0.093831
In this paper, OPF has been solved considering the

2
uncertainty in the load demand. The uncertainty in the load
6 0,189410 ~
demand has been incofiorated in the study by the use of
3 7 0,312321
fuzzy logic. To gain insights regarding the advantage or
4 7 0.187601
demerits of using fuzzy dynamic programming an attempt
4 8 0.131157
has been made to solve the OPF problem. Also, possibility of
5 6 0.063340
using TCSCS to satisfy the transmission capacity constraints
5 10 0.255540
in the OPF has also been expIored.
6 9 0.160411
8 10 0.020083 Consideration of uncertainty in the load demand
9 10 0.127892 results into higher generation cost at the most feasible
operating point compared to the cost when the uncertainty in
the load demand is not considered. Whereas the generation
pattern when the load uncertainty is considered is more
reliable than the generation pattern obtained without the “
Table-1.3 : Generation patterns m~d their corresponding consideration of the uncertainty in load demand. With the use
costs without violating line constraints. of TCSC, the total generating cost comes out to be less than
that obtained without TCSC. Hence the savings obtained

Unit Generation because of the difference in generation cost would offset the
Generating cost (units)
No. (MW) cost of TCSC implementation.
t
I 115.00 29.637001
2 70.00 33.839001
VII, REFERENCES
3 55.00 26.161751
1. Precdavichit, P., and Srivastava, S. C., “Optinxd Reactive
4 40.00 26.76000 Power Dispa[ch Considering FACTS Devices”, Electrical
Power Systcms Research, Vol. 46, pp. 251-257, 1998.
5 44.110260 22.684479

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 514
0-7803-6674-3/00/$10.00 (C) 2000 IEEE
2. Gotham, fl. J., and 1Iaydt, G. T., ‘{Pcrww Flow Control in Systems Table-A3: Load Distribution
with FACTS devices”, Electrical Machines and Power Systems,
vol. 26, pp. 951-962, 1998, Bus # Load Bus # Load
3. Hingortsni, N., “Flexible AC Transmission”, IEEE Spcctlum, Vol.
30, No. 4, pp. 40-45, April, 1993. 1 0.3 +jO.1 6 0.6 +jO.15
4. Ji. Yrrmr Fan, Lan Zhang, “Real – Time Economic Dispatch with
Line Flow and Emission Constraints Using Quadratic 2 0.4 +jO.15 7 0.2 +jO.1
Programming”, IEEE Transirctions on Power Systems, Vol. 13,
3 0.2 +jO. I 8 0.4 +jO.1
I+o. 2, pp. 320-324, May 1998, ,
5. Wood, A.J. Wollenberg, B F., “Power Generation Operation and 4 0.2 +jO. I
0.3 -tjO.15 9’
Control”, John Wiley and Sons, 1984.
6. Zimermann, H. J., “FUZZV Sets, Decision Makirre and Exoert 5 0.3 +jO.1 10 0.6 +jO.1
Systems”, Khrwer Academ;c Publishers, 1987. - ‘

APPENDIX – A

TEST DAFA BIOGRAPHY

Table-Al: Line Characteristics Naravana Prasad Padhv obtained his degree of engineering and

T
Master of Engineering in
Line Bus # Impedance 1990

z
and 1993,
# respectively. in 1997, he
obkrined his Pb. t). dcgrcc
1 1-2 0.02 + jO.08 0.01 0.6 from Anna University,
Chennai, India. He joined
2 1-6 0.06 +jO.25 0.02 0.3 Birla Institute of
Technology & Science as
3 1-9 0.04 +jO.16 0.02 0.3 on Assistant Pmfcssor in

4 2-3 0,06 +jO.25 0.02 0.3


,. Jif
..
1997. I IC is presently with
the faculty of Electrical

6
2-6

3-7
0.06 +jO.25

0.06 +jO.25
0.02

0.02
0.3

0.3’
;
I

He taught course in Basic Electrical


P I
~ ~:
Engineerin~,
Roorkee,

Engineering,
Roorkee.
University

Power Systems and


of

Artificial Intelligence. His field of interest is Artificial Intelligence


7 4-7 0.04 +jO.16 0.02 0.3
Applications to Power System Optimization Problem.
8 ‘4-8 0.06 +jO.25 0.02 0.3

9 5-6 0.04 +jO.16 0.02 0.3 M. A. R. Abdc+-hloamcn, P. K. Trivcdi and Il. I)as arc working as a
research scholar, post graduate student and faculty in the power
5-1o 0.06 +jO.25 0.02 0.3 systems group, department of electrical engineering respectively.

11 6-9 0.02 +jO.08 0.01 0.6

8-10 0.04 +jO.16 0.02 0.3

13 9-1o 0.08 +jO.32 0.025 0.2

Table-A2: Unit Characteristics

I Unit I Bus# 1 Cost function Pmax I Pmin 1


I I I
1 6 20i-7p+ 1.2P~ 1.2 0.1

2 7 20+9p+l.l P2 1.2 0.1

I 318 21+9p+0.7P2 i~
I 4]9
,
23+9p+l.0P2
,
1.2
I
0.1

I 5 10 19+8p+0.8P2 1.2 0.1

0-7803-6674-3/00/$10.00
0-7803-6672-7/01/$10.00
(C)(C)
2000
2001
IEEE
IEEE 515

0-7803-6674-3/00/$10.00 (C) 2000 IEEE

You might also like