Professional Documents
Culture Documents
E7 - 4
m.wartana@hotmail.com
niputu.agustini@yahoo.co.id
optimization,
NSGA-II,
stability margin, system loadability, UPFC controller
I. INTRODUCTION
Power systems are becoming increasingly more complex
due to the interconnection of regional system and deregulation
of the overall electricity market. It has become imperative to
better utilize the existing power networks to increase
capacities by installing FACTS controllers [7]. The variables
and parameter of the transmission line, which include line
reactance, voltage magnitude, and phase angle are able to be
controlled using FACTS controllers in a fast and effective
way [12]. The benefits derived from FACTS include
improvement of the stability of power system networks, such
as voltage stability, line stability, small signal stability,
transient stability, and thus enhance system reliability.
However, controlling power flows is the main function of
FACTS [13]. Maximal system loadability can also be obtained
with the optimal location and parameter setting of FACTS
controllers [4, 19]. The maximum benefit of the FACTS
controllers depends greatly on how these controllers are
allocated in the power system: namely, on their location and
setting [1].
max
iSH = (i p + iq ) e j i
and f is the maximal limit of f . The load factor f reflects
QLi ( f ) = f QLi
2
Qiju = brVi cos iqVi
(2)
j = 1, ...., M
g ( x, u ) = 0
(4)
Subject to :
k = 1, ...., K
h ( x, u ) 0
where F is known as the objective vector, F1, and F2 are the
bi-objective functions to be optimized, x is the vector of
dependent variables, and u is the vector of control variables.
A. Maximize the system loadability within security margin
This objective is to maximize system loadability by
considering security constraints as given below:
(5)
Nl
Nb
i =1
j =1
the variation of power loads PLi and QLi , which are defined as
[1]:
(8)
PLi ( f ) = f PLi
(6)
(9)
(10)
OLLi =
Pij
max
i and j and the thermal limit for the line between buses i and j
respectively. OLL is the coefficient which is used to adjust
the slope of the exponential function.
The second part BVVj in (11) concerns the voltage levels
for each bus of the power network. The value of BVVj is
defined as [18]:
if 0.9 Vb 1.1
1;
(11)
BVV j =
exp(BVV 1 Vb ) ; otherwise
where BVVj is the bus voltage violation factor at bus j and
BVV represents the coefficient used to adjust the slope of the
exponential function in the above equation. The equation
shows that appropriate voltage magnitudes are close to 1 pu
Similar to OLLi , the value of BVVj is equal to 1 if the voltage
level falls between the minimal and maximal voltage limits.
Outside the range, BVVj increases exponentially with the
voltage variation.
B. Minimization of Active Power Losses of the transmission
lines
This objective is to minimize the active power losses (Ploss)
in the transmission lines and which can be expressed as [1]:
Nl
g [V
F 2 ( x, u ) =
2
i
+ V i 2 2V i V j cos t ( i j ) .
k =1
(12)
(14)
S = Q2 Q1
i =1
Nb
(18)
(23)
2
4Z Q j
2
Vi X
(24)
LQP = 4
Pi + Q j
2
2
V V
i i
(25)
E. Equality Constraints
F. Inequality Constraints
The inequality constraints h(x, u) are limits of control
variables and state variables. Generator active power PG,
reactive power QG, voltage Vi, and phase angle i are restricted
by their limits as follows:
PGmin
i = 1,......,m
PGi PGmax
i
i
min
max
(20)
QGi QGi QGi
i = 1,......,m
min
max
Vi Vi Vi
i = 1,.......,Nb
i = 1,.......,Nb
0.9 i 0.9
The parameter settings of single UPFC controller are
restricted by their limits as follows:
vSmin vS vSmax
(21)
min
max
iSH
iSH iSH
The constraint of transmission loading Pij is represented as
max
(22)
Pij Pij ; ij = 1,......., Nl
H. NSGA-II
In principle, multi-objective optimization has a very
different nature with single-objective optimization. In single
objective, one attempts to obtain the best design or decision,
which is usually the global optimum [1]. But in the case of
multiple conflicting objectives, a typical multi optimization
problem produces a set of solutions which are superior to the
rest of the solutions with respect to all objective criteria but
are inferior to other solutions in one or more objectives [20].
max
Fi
(26)
Fi
Fi min < F i < F i max
i = max
min
Fi
Fi
0,
F i Fi max
min
max
Where Fi
and Fi
are the minimum and the
maximum value of the ith objective function among all nondominated solutions, respectively. The membership function
i is varied between 0 and 1, where i = 0 indicates the
incompatibility of the solution with the set, while = 1 means
full compatibility. For each non-dominated solution k, the
Fig. 2 NSGA-II procedure
1) In order to divide the population into separate fronts, fast
non-dominated sorting approach is applied, in which for each
solution two entities are calculated:
a) Domination count np, the number of solution which
dominates the solution p.
b) Sp, a set of solutions that the solution p dominates.
All the solutions in the first non-dominated front will have
their domination count as zero, and they are eliminated to get
the other fronts. It should be mentioned that, if m is the
number of objective function, dominance of chromosome x
against chromosome p is defined as:
For i = 1: m
fitnessi (x) fitnessi (p).
2) Thereafter for the purpose of maintaining sustainable
diversity in population, chromosomes in each front should be
sorted with crowded-comparison approach. For each
chromosome i, the quantity of crowding distance idistance is
the average distance between two points on either side of this
point along each objective. Therefore in comparison between
two solutions, we prefer solution with lower rank and in the
case of same ranks, the solution that is located in lesser
crowded region (with larger quantity of idistance) is chosen.
After implementation of these sorting procedures, half
population is directly chosen as Pt+1 like the elitist method.
Thereafter the new population Pt+1 is now used for tournament
selection, crossover and mutation to create the other half of
population so-called Qt+1.With calculation the merit of total
population, next iteration is commenced.
I. Best compromise solution
Once the Pareto optimal set is obtained, it is practical to
choose one solution from all solutions that satisfy different
goals to some extends. Due to the imprecise nature of the
decision makers (DM) judgment, it is natural to assume that
the DM may have fuzzy or imprecise nature goals of each
k =
N obj
i =1
is calculated as:
k
i
N obj
k =1
i =1
(27)
k
i
Population
Generations
100
50
Pool
size
25
Tour
Size
2
20
20
UPFC
controller
Max.
SL
Min.
Ploss
Location (line)
Shunt setting (pu)
Serie setting (%.)
SL (%.)
Ploss (pu)
IC (x106$.)
6-7
6-4
0.94
1.01
-8.36
-50.29
264.75
186.10
0.4303
0.1213
17.2890 13.2056
Best
comp.
solution
6-4
1.01
-41.05
190.86
0.1270
15.0133
0.45
0.4
Non-dominated solution
Best compromise solution
Maximum system loadability
Minimum active power loss
Type of
FACTS
SL
(%)
IC
(106$)
UPFC
190.86
139
1
8
15.0133
276.7
Stability
index/factor
constraints
Yes
No
Result
in this study
reported in [13]
0.35
0.3
0.25
0.2
0.15
0.1
180
X: 190.9
Y: 0.127
190
200
210
220
230
240
250
260
270
LQP (p.u.)
FVSI (p.u)
Fig. 3 Pareto front of the best compromise solution of UPFC for IEEE 30-bus
system
LQP
0.20
0.25
0.21
FVSI
0.17
0.10
0.13
0.09
0.05
0.00
0.01
-0.03
-0.07 1 4 7 10 13 16 19 22 25 28 31 34 37 40 -0.10
-0.11
-0.15
-0.20
Lines
Fig. 4 The best compromise solution of FVSI and LQP after UPFC placement
in IEEE 30-bus system
UPFC
controller
Max.
SL
Min.
Ploss
Location (line)
Shunt setting (pu)
Serie setting (%.)
SL (%.)
Ploss (pu)
IC (x106$.)
13-14
22-23
1.00
1.07
-34.06
17.99
189.25
128.02
7.5621
1.3923
27.8050 46.8286
Best
comp.
solution
19-1
0.98
20.00
153.73
2.2780
11.9611
[2].
8
Non-dominated solution
Best compromise solution
Maximum system loadability
Minimum active power loss
[3].
[4].
5
[5].
3
X: 153.7
Y: 2.278
1
120
[6].
130
140
150
160
170
180
190
200
Fig. 6 Pareto front of the best compromise solution of UPFC for Java-Bali 24bus Indonesian system
0.40
0.30
0.20
0.10
0.00
LQP (p.u.)
FVSI (p.u)
0.40
FVSI
0.30
LQP
0.20
0.10
0.00
-0.10 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49
-0.20
-0.30
-0.40
Lines
[7].
-0.10
-0.20
[8].
[9].
[10].
Fig. 7 The best compromise solution of FVSI and LQP after UPFC placement
in Java-Bali 24-bus Indonesian system
VI. CONCLUSIONS
A novel approach based on NSPGA-II has been presented
and implemented to optimal location and parameter settings of
UPFC controller in this study. The problem is formulated as a
real mixed continuous integer multi-objective optimization
problem, where two different objectives are considered
namely: maximize system loadability by maintaining security
and stability constrains (FVSI and LQP) and minimize active
power losses by considering installation cost of UPFC
controller. The optimal location and parameter settings of the
controller are performed for single use of UPFC controller. A
fuzzy-based mechanism is employed to extract the best
compromise solution from the Pareto front. The results show
that NSGA-II provides well distributed non-dominated
solutions and well exploration of the research space. Since
UPFC controller is expensive, this study will be further
extended to address the problem of optimal location of UPFC
controller to minimize installation cost of the controller.
ACKNOWLEDGMENT
The authors would like to thank to the Indonesian
Government Electrical Company for supporting the data and
to Prof. F. Milano for the free PSAT software package.
[11].
[12].
[13].
[14].
[15].
[16].
[17].
[18].
[19].
REFERENCES
[1].
[20].